text
stringlengths 454
608k
| url
stringlengths 17
896
| dump
stringlengths 9
15
⌀ | source
stringclasses 1
value | word_count
int64 101
114k
| flesch_reading_ease
float64 50
104
|
|---|---|---|---|---|---|
A Python application runs on a single thread, unless you explicitly enable multithreading.
Why is multithreading useful? Code in Python is ran in sequence, one instruction after another.
If you define a function that sleeps 3 seconds and then prints something, like this:
import time def greet(): time.sleep(3) print('hello') greet() print('world')
The
world string is printed after 3 seconds, because we call
time.sleep(3) inside the
greet() function.
This is just a silly example, but imagine processing an image, getting a resource from the network, or writing a big file to disk. Anything that can take a lot of time.
With multithreading we can run the function that takes a lot of time into a separate thread, and go on with our program in the meantime.
The
threading standard library module helps with implementing multithreading. You import
Thread from it:
from threading import Thread
Then we pass the function we must execute as the
target argument to the
Thread() function, getting a thread object:
t = Thread(target=greet)
then we call its start() method to start the thread:
t.start()
Try running this code:
from threading import Thread import time def greet(): time.sleep(3) print('hello') t = Thread(target=greet) t.start() print('world')
You will now see
world printed 3 seconds before
hello shows up in the console.
The program does not end until the thread (or all the threads it started) end, unless you start a thread as a deamon.
Those are the basics of multithreading. This is complex, and can lead to the introduction of bugs if not done well.
Download my free Python Handbook
More python tutorials:
- Beginning GUI Programming in Python with `tkinter`
- Python Introspection
- How to use Python map()
- Python Numbers
- Python, read the content of a file
- How to check if a variable is a string in Python
- Python List comprehensions
- Django in VS Code, fix the error `Unable to import django.db`
- Python Polymorphism | https://flaviocopes.com/python-multithreading/ | CC-MAIN-2021-17 | refinedweb | 328 | 71.75 |
Agenda
See also: IRC log
<scribe> Scribe: Ed Rice
Propose next teleconference: 28 November - conflict with AC meeting?
Regrets Tim, Norm, Vincent and likely Henry
Norm: Propose cancel next week meeting
Ed +1
<timbl> +1
Resolved, next weeks meeting will be cancelled.
Propose next teleconference: 5 Dec
Propose Noah as scribe.
Approve minutes of last teleconference?
<Vincent> Minutes 14 nov.
Resolved: approve minutes of last teleconference.
Agenda review
<Zakim> DanC, you wanted to request discussion of tagSoupIntegration
agenda accepted with Dan's addition for tagSoupIntegration
<Zakim> DanC, you wanted to suggest clarifying the scope to "never send passwords in the cear across the Internet"
Ed talks about the proposed exceptions to where it is ok to send passwords in the clear.
1) I just want to keep a page off the search engine, its not really 'secure'.
2) my network is secured, so I dont need to secure the password.
Dan: I think 'should not' would be ok, instead of a must not.
Timbl: if your running on a
secured network, that doesn't mean there are no virus's on
it.
... the one machine could be connected to other networks which are then not as secured.
... or it could be sniffed by someone else on the VPN.
<DanC> (this discussion of VPNs and firewalls starts to sound like IETF IPv6 discussions)
Dave: One person in my group went to a conference who said firewalls really dont work and its pushing more security on the local machine because only one client inside your firewall can compromise your entire network.
<DanC> Web Security Context Working Group
Dan: W3C has a new security context WG. We may want to have them review this.
Ed, should we change it to 'should' then or send it to the working group?
Dan: I'm in favor of either of these.
Norm: If we say 'should' how do
we point to anyone and say your in violation of the finding or
not?
... So, I have a marginal preference for 'must' if we can get the community to buy in to it.
Dan: All these things are about managing risk.
dorchard: I think of must as a 'if you don't follow a "must" your outside the architecture', and we're saying sometimes its ok to violate a 'must' so I think a 'should' is probably ok on this one.
DO: I think that should is the right word to use.
<DanC> hmm... tim's idea has some appeal; put the MUST onus on the user agent to make the user aware of the risk. hmm....
<DanC> that's already in there: "A client or browser MUST NOT transmit passwords in clear text."
<Zakim> DanC, you wanted to express reservation about advocating a change that we don't have experience with.
Timbl: I suggest adding 'user
agents MUST warn when a password is to be sent in the
clear'
... and the user SHOULD NOT send passwords in the clear
... and the server SHOULD NOT request passwords be sent in clear text.
<DanC> (hmm... perhaps s/MUST warn/MUST not send without informed consent/ ? I think there's some precedent in WAI. but yes, that's just wordsmithing)
Dan: I'd like to see the use cases identified in a paragraph or two as well.
<DanC> replay:
<DanC> Timbl: I suggest adding 'user agents MUST warn when a password is to be sent in the clear'
<DanC> Timbl: and the browser SHOULD NOT send passwords in the clear
<DanC> Timbl: and the user SHOULD NOT send passwords in the clear
<DanC> timbl: and the server SHOULD NOT request passwords be sent in clear text.
<DanC> (disregard 2nd line of replay)
<scribe> ACTION: Ed to produce a new version with these changes. [recorded in]
<DanC> indeed, raman , it's "user SHOULD NOT send passwords in the clear"
VQ: lets talk about what we'd like to achieve?
Schedule is Monday - Wednesday afternoon.
Dave: when to use Get - issue 7 (wstransfer)
Norm: I'm hoping to have namespace document 8 and semantic web dialog, web architecture. I'm hopeful I'll get these done by the face to face.
Timbl: Yes, particularly the semantic web.. we need to get the TAG to focus more on this.
Norm: I will also need to leave early on wednesday.
<Zakim> DanC, you wanted to note my TAG priorities are tagSoupIntegration and extensibility/versioning, though I haven't thought about group priorities much
Dan: My personal interests are tagSoupIntegration and versioning/extensibility.
DO: We never really closed on what we need to do since the last face to face. This paper has become more of a thesis and we discussed possibly breaking it up.
VQ: well, maybe we should discuss to try and resume progress on this.
TV: If I was coming I would like
to see more on the TAGSoup issue.
... I will definatly not be able to attend the f2f due to prior commitments.
Ed: I'd like to try and close on passwords in the clear at the f2f if we could.
<timbl> Raman, you could not call in even, I gather?
TV: yes, I can call in during the evening hours. I'd like to participate in the TagSoup.
VQ: I'll send out a draft agenda
so we can try and settle at least one week in advance.
... other topics for F2F?
Tag discusses..
<DanC>
VQ: anything else to cover
today?
... Meeting adjourned, next meeting in two weeks. | http://www.w3.org/2006/11/21-tagmem-minutes.html | CC-MAIN-2014-52 | refinedweb | 909 | 72.05 |
When I first started using ReSharper a couple years ago and with each subsequent release, I’ve had a lot of fun just exploring all of those “little” features that sometimes go unnoticed, but yet, can greatly increase your productivity in Visual Studio. With the latest 3.0 release, I’m in that mode again. I’m finding many more cool “little” features that just make my life easier. For instance, here’s just one example…
Basic namespace/folder maintenance
Being a bit picky about project/code organization, I’m a stickler for good folder structures and keeping the namespaces in sync. In the past when I’ve needed to move a class to a new folder/namespace I’d go through the following steps:
- F6 to show ReSharper’s very nice Rename dialog
- Type in the new namespace for the class
- Create the folder in the solution explorer
- Cut and paste the file from the solution explorer into the new folder
Very straightforward and fast. This is one of those organizational steps that you may think cannot be improved upon. But just leave it to JetBrains…
Now, thanks to ReSharper 3.0, all I have to do is this:
- Create the folder in the solution explorer
- Cut and paste the file from the solution explorer into the new folder
- F12 to the nice new warning that ReSharper gives me…
- Alt-Enter to have ReSharper automatically fix it for me
The key thing about this is that I don’t have to actually type the namespace anymore. May seem small, but many of these “little” features combined with each other can make you much more productive in your development efforts.
Off to discover some more candy…
Post Footer automatically generated by Add Post Footer Plugin for wordpress. | http://lostechies.com/joeybeninghove/2007/06/28/resharper-3-0-like-a-kid-in-a-candy-store-again/ | CC-MAIN-2014-52 | refinedweb | 297 | 53.95 |
PTP/testing/6.x
Contents
Test Plan for PTP 6.x Release
This plan describes the tests that will be undertaken to verify the 6.x series of PTP releases - Eclipse 6.0.0 will be available June 27, 2012 with the Eclipse Juno (4.2) Simultaneous release.
Test Setup
The following steps should be carried out prior to testing PTP. Refer to the release notes for the appropriate PTP version if necessary.
- Make sure you have Java 1.5 or later
- Start with a fresh Eclipse installation
- Use a new workspace for testing (or remove old testing workspaces)
- Download and install the "Eclipse IDE for Parallel Application Developers" package
- from Select "Developer Builds" at the top to get pre-release downloads.
- Depending on your test plan requirements, install PTP server components using instructions from the 6.0 release notes.
- Launch Eclipse on the test machine.
- If you require an MPI program for testing, use the following code:
#include <stdio.h> #include <mpi.h> int main(int argc, char *argv[]) { int i, rank; MPI_Init(&argc, &argv); MPI_Comm_rank(MPI_COMM_WORLD, &rank); for (i = 0; i < 10; i++) { printf("hello from %d loop %d\n", rank, i); sleep(1); } MPI_Finalize(); return 0; }
Test Iterations
The following table shows the test iterations for the RC builds. PTP is +2 which means that its build is completed on the +2 day. Testing can begin using the Juno aggregation site () on +3 and later. The EPP build is generally available at the end of the +3 day, but official EPP testing can begin on the EPP day once the packages have been announced.
RC4 testing is still useful as any bugs can still be fixed in future bug fix releases or service releases (SR1, SR2). However, these fixes will not be included in the Juno release.
Testers
Please add your name to this list, or remove the comment if you are still participating in 6.0
- Greg Watson
- Beth Tibbitts
- Dave Wootton
- Alan Humphrey
- Wyatt Spear
- Jeff Overbey
- Jay Alameda (remove this if participating in 6.0)
- Galen Arnold
- Rui Liu (remove this if participating in 6.0)
- Jie Jiang (remove this if participating in 6.0)
- Max Billingsley III (remove this if participating in 6.0)
- Roland Schulz (remove this if participating in 6.0)
- John Eblen (remove this if participating in 6.0)
- Hari Krishnan
Test Matrix
The following test matrix shows who will be testing each component on the different supported architectures. Note that because PTP is client/server, you should list both the client and server architectures for each component being tested. For example, if you are testing PE on a Mac OS X x86 client and an AIX backend, you should list both these architectures.
When you find bugs, please open a bug report against the relevant component.
* Only the GTK 2 version of Eclipse will be tested
** Only the Cocoa version of Eclipse will be tested
*** AIX testing is for the server side only.
α Limited testing only
(client) indicates that eclipse runs here
(server) indicates remote target runs here
(R) will test remote usage
(L) will test local usage
## old (not updated to PTP 6.0 testing) - remove ## if you update this and/or confirm you are still testing this
Test Outlines
Initial overview of some features to be tested:
- Install & Update: install parallel package, try update scenarios from & to different RCs/releases, make sure feature names, versions, providers, ids, etc are correctly set in installation information for each of the features (Beth)
- Help: bring up help, make sure the pieces are all there, make sure topics are up to date, any version numbers referenced are correct, make sure links work between help docs (Beth) | http://wiki.eclipse.org/index.php?title=PTP/testing/6.x&oldid=305150 | CC-MAIN-2014-52 | refinedweb | 616 | 62.48 |
SYNOPSIS
#include <libaio.h>
int io_destroy(aio_context_t ctx);
Link with -laio.
DESCRIPTION
io_destroy() removes the asynchronous I/O context from the list of I/O
contexts and then destroys it. io_destroy() can also cancel any out-
standing asynchronous I/O actions on ctx and block on completion.
RETURN VALUE
On success, io_destroy() returns 0. For the failure return, see NOTES.
ERRORS
EFAULT The context pointed to is invalid.
EINVAL The AIO context specified by ctx.
NOTES
Glibc does not provide a wrapper function for this system call.
The wrapper provided in libaio for io_destroy()_getevents(2), io_setup(2), io_submit(2)
COLOPHON
This page is part of release 3.23 of the Linux man-pages project. A
description of the project, and information about reporting bugs, can
be found at. | http://www.linux-directory.com/man2/io_destroy.shtml | crawl-003 | refinedweb | 129 | 69.99 |
Any function prototypes, and type definitions that can be exported from a source code file are put in a header file. From the point of view of the main application, these functions are external.
The compiler reads the included header file and knows where a particular function comes from. Without this it would just report an undefined function error.
When the application is linked, all calls to external functions are resolved by the linker and built into an exe or dll (.so on Linux).
Dangers
Because file_b might include file_a and both include file_d, it's important to protect against this by using the #ifndef directive to check that file_d is only included once. So the 'include for file_d should be wrapped inside a #ifndef ... #endif and these four lines should be in both file_a and file_b.
#ifndef FILE_D
#define FILE_D
#include <file_d.h>
#endif
If the pre-processor includes file_a first then the FILE_D will not exist and so it will be defined and file_d.h will be included. Next when file_b is processed, FILE_D will exist and so file_d.h will not be included for a second time. | http://cplus.about.com/od/glossar1/g/headerdefn.htm | CC-MAIN-2014-10 | refinedweb | 190 | 75 |
You should be used to installing new modules using pip. You have probably used a requirements.txt file to install multiple modules together with the command.
pip install -r requirements.txt
But what about if you need more flexibility. Why would you ever need more flexibility? If you look at my introduction to YAML post, the code supports either the yaml or ruamel.yaml module. There is no way to add conditional logic to a requirements.txt file so a different strategy is needed.
pip is just a module so it can be imported like any other module. This not only gives you access to the main method, which takes an argument list just as if you were calling pip from the command line, but also to its various methods and classes. One of these is the WorkingSet class which creates a collection of the installed modules (or active distributions as the documentation calls them). Using this we can create the conditional logic needed to ensure one of the yaml modules is installed as below.
import pip package_names = [ ws.project_name for ws in pip._vendor.pkg_resources.WorkingSet() ] if ('yaml' not in package_names) and ('ruamel.yaml' not in package_names): pip.main(['install','ruamel.yaml'])
WorkingSet returns a few other useful properties and methods apart from the package_name. The location property returns the path to where the module is installed and the version property naturally returns the version installed. The requires method returns a list of dependencies.
As with most modules, if you’re interested in finding out more dig around in the source code. | https://quackajack.wordpress.com/tag/import/ | CC-MAIN-2018-47 | refinedweb | 263 | 59.4 |
hi there
I want to give a for loop in my programm that if any student fails in more than four subject he will not be able to go to the next level. if average is less than 40 then the student will be failed . i tried in many ways. but compiler is not taking at all. below i am just showing the logs for the loop but not the whole program. the program has got other parts. Basically i could not put the exact logic.
#include <stdio.h> main() { int s,t; Printf ("This student is not allowed to go to the next level as he fails in more than two subject\n"); scanf("%d",&s); /*here what will be the address of s ?*/ for (t=1, t>=4, t++) } | https://www.daniweb.com/programming/software-development/threads/140459/looping-problem | CC-MAIN-2018-43 | refinedweb | 131 | 91.31 |
I am really new to coding and I have been teaching myself how to code by using EDX.org. This week I have been studying about Cryptography and I have to create a Vigenère’s cipher. I wrote the code and for the most part, it's correct. However when I compiled the program, it's showing a segmentation error. I have been trying to figure our why this is happening and I am totally stuck. Could you take a look at my code and tell me whats wrong?
#include <stdio.h>
#include <stdlib.h>
#include <cs50.h>
#include <string.h>
#include <ctype.h>
int index(int k, int c);
int main (int argc, string argv[1])
{
//check for correct criteria
if (argc = 2, isalpha(argv[1]))
{
string text = GetString();
string key = argv[1]; //store key word
int Totalshift = strlen(key); //number of shift for keyword
int shift = 0;
//loops over the whole text
for (int i = 0, n = strlen(text); i <n; i++ )
{
char p= text[i];
char k = toupper(key[shift]); //Upper case for each character
if (isupper(p))
{
//convert to 0 index
p = p - 65;
k = k - 65;
int crypt= index (k , p);
printf("%c", crypt+65);
shift= (shift+1) %Totalshift;
}
else if (islower(p))
{
p = p - 97;
k = k - 65;
int crypt= index (k , p);
printf("%c", crypt+97);
shift= (shift+1) %Totalshift;
}
else
{
printf("%c", p);
}
}
printf("\n");
}
//error message
else
{
printf("ERROR!\n");
return 1;
}
}
//index function
int index(int k, int p)
{
return (k+p)% 26;
}
string
No. Never, ever hide pointers.
int main(int argc, char ** argv)
Then:
//check for correct criteria if (argc = 2, isalpha(argv[1]))
Here, you assign to the variable (parameters behave like local variables in that respect) the value
2, thus destroying the previous value (which holds the number of arguments given to the program). The result is the value which was assigned, thus
2. Then, there's the comma operator: You discard that
2, and then call
isalpha(argv[1]), which clearly shows why you should always turn on warnings and never, ever hide pointers:
argv[1] is of type
char *, thus a pointer to a character array (or, as we know in this case, a character array terminated with
'\0', which is called a C string). Since
isalpha takes an
int as parameter the value of the pointer ("the memory address") is implicitly converted to (a probably very large)
int value. Quoting from above link, emphasis mine:
The c argument is an int, the value of which the application shall ensure is representable as an unsigned char or equal to the value of the macro EOF. If the argument has any other value, the behavior is undefined.
Which is possibly the source of the segmentation fault.
Finally, that
GetString really looks fishy to me. Assuming that it allocates some memory (for the string that it presumably reads from the user) ... where do you free that memory? Is it really allocating memory, or possibly returning a pointer to an array with automatic storage duration (a local variable, so to say)? | https://codedump.io/share/RTjcuaAKbgXD/1/segmentation-fault-in-c-while-coding-vigen232res-cipher | CC-MAIN-2017-09 | refinedweb | 514 | 59.84 |
Here's an interesting bit of Python code I hacked together – it's a script that takes an image and warps it so that it is tileable (making it suitable for a repeating backgound or a texture in a game).
If you use it on a photograph, it will come out looking like a fair-ground mirror. But it works well when applied to a pattern, or something more abstract, such as the fractal image on the left.
The code is public domain – use it for whatever the heck you want!
Example Output
Update: Here's another, more interesting example, The original is here.
The Code
import Image from math import * def maketilable(src_path, dst_path): src = Image.open(src_path) src = src.convert('RGB') src_w, src_h = src.size dst = Image.new('RGB', (src_w, src_h)) w, h = dst.size def warp(p, l, dl): i = float(p) / l i = sin(i*pi*2 + pi) i = i / 2.0 + .5 return abs(i * dl) warpx = [warp(x, w-1, src_w-1) for x in range(w)] warpy = [warp(y, h-1, src_h-1) for y in range(h)] get = src.load() put = dst.load() def getpixel(x, y): frac_x = x - floor(x) frac_y = y - floor(y) x1 = (x+1)%src_w y1 = (y+1)%src_h a = get[x, y] b = get[x1, y] c = get[x, y1] d = get[x1, y1] area_d = frac_x * frac_y area_c = (1.-frac_x) * frac_y area_b = frac_x * (1. - frac_y) area_a = (1.-frac_x) * (1. - frac_y) a = [n*area_a for n in a] b = [n*area_b for n in b] c = [n*area_c for n in c] d = [n*area_d for n in d] return tuple(int(sum(s)) for s in zip(a,b,c,d)) old_status_msg = None status_msg = '' for y in xrange(h): status_msg = '%2d%% complete' % ((float(y) / h)*100.0) if status_msg != old_status_msg: print status_msg old_status_msg = status_msg for x in xrange(w): put[x, y] = getpixel(warpx[x], warpy[y]) dst.save(dst_path) if __name__ == "__main__": import sys try: src_path = sys.argv[1] dst_path = sys.argv[2] except IndexError: print "<source image path>, <destination image path>" else: maketilable(src_path, dst_path)
Smells like a job for Numpy ( [scipy.org]):
Pauli, Nice! I think it is a job for Numpy.
Once more, this time demonstrating the use of scipy.ndimage for the interpolation: | http://www.willmcgugan.com/2009/7/18/make-tilable-backgrounds-with-python/ | CC-MAIN-2014-10 | refinedweb | 381 | 76.32 |
Fix the use of units in Gecko
RESOLVED FIXED in mozilla1.9alpha3
Status
()
People
(Reporter: roc, Assigned: sharparrow1)
Tracking
(Depends on 2 bugs, Blocks 4 bugs)
500997, 513837, 97861, 322938, 353860, 369618, 369684, 369690, 369693, 369698, 369882, 370006, 370444, 370466, 370553, 370629, 370631, 373381, 375172, 376690, 378927, 380438, 381074, 381250, 381792, 382421, 382458, 382961, 386406, 391868, 395677, 396811, 398336, 401904, 412646, 421700, 428975, 434841
Bug Flags:
Firefox Tracking Flags
(Not tracked)
Attachments
(2 attachments, 16 obsolete attachments)
Currently we have a proliferation of units and conversions in Gecko which are very difficult to understand, which induce a lot of conversion code (some special-cased for specific situations like Print Preview), and which appear to be leading to rounding errors (hard to tell because things are too complex). Here's my proposal for fixing the situation: 1) Define 'layout units' to be exactly 1/100th of a CSS pixel. 2) Use these layout units everywhere. 3) Scale these units to device units inside GFX. Do not perform any device unit scaling outside GFX. 4) In Widget, provide a scale factor that maps layout units to window system pixels. 5) In GFX, provide a scale factor that maps physical lengths to layout units. This scale factor is device specific. This scale factor would be used only in the style system to convert lengths in physical units into layout units. dbaron wrote: > For example, in an ideal system, I might want to have > three constants for: > 1. converting reference pixels to app units (which would be points for > printing or fractions of pixels for display) > 2. converting physical lengths to app units, and > 3. an overall scale. Rougly, I'm suggesting fixing 1. to some constant (say, 1/100), 2. corresponds to my 5), and 3. is achieved by varying 3) and 5). I have to go now, I'll attach some rationale later.
I suppose for the sake of consistency with existing code we should call layout units 'app units'. After some more thought I think the magic number should be 1/720 of a CSS pixel. I'm assuming that we bend the CSS rules a bit and choose the size of a CSS pixel to make sure that the ratio of screen pixels to CSS pixels is always an integer (i.e., 1 for most of today's screens, 2 for screens like IBM's 200dpi LCD panel, maybe more later). If we don't do this it would be a disaster for the clarity of many Web pages, no matter what else we do. Then for ratios 1, 2, 3, 4, 5 and 6 we get an integral number of app units per screen pixel. Beyond that it probably doesn't matter. By fixing app units to 1/720 of a CSS pixel, we guarantee that if the author specifies lengths in the form NNN.Npx, all constraints will be computed exactly. For example, a horizontal stack of boxes of specified lengths will always fit perfectly into a box whose width is specified to be the sum of the box lengths. 1/720 of a CSS pixel is nominally about .4 microns, so we get good fidelity even for very high resolution output devices. But we still get 1.4 million pixels in 30 bits, which is enough for a > 100,000 line document with reasonable font height.
Inorder to fix the roundoff problems.. keep in mind: Anytime something is draw onto the screen it has to either a.) placement on the screen has to use a layout unit that is coicident with the screens pixel.. roundoff of this unit would not help. or b.) The error from this placement has to be kept around for any calculations relative to this object so it can be calculated correctly. I have created examples in another bug (63336), to show this is the problem behind our roundoff problems. A unit that keeps an error might be the best.. like floating or a fixed unit (16 bits decimal, 16 bits fractional) I totally agree with the cleanup of our conversion and unit cleanup, we have duplicate units, duplicate calls, calls that have poliferated outside of GFX, calls with confusing names. I am in for helping out on this one.
dcone wrote: > like floating or a fixed unit (16 bits decimal, 16 bits fractional) You are aware that output devices like printers can easily hit coordinates beyond 65536, right ?
It was just an example.. we could use a 24 decimal and 8 fractional or floating or double. I really did not give thought to the exact unit to use. But it would only have to take care of our layout/application needs like twips does, the conversion to the device units would be kept in the 32 bit longs.
In this framework, the print device tells us the physical size of a CSS pixel, which we use to size physical lengths such as points and inches. Print preview, shrink to fit and print scaling use the same number for physical size of a CSS pixel, but also specify a scale factor to GFX to shrink the printed output to fit on the screen or the page. This may require modifying GFX to make sure the GFX transform is properly applied to fonts (and everything else that gets rendered).
> placement on the screen has to use a layout unit that is coicident with > the screens pixel.. roundoff of this unit would not help. Authors can specify lengths that aren't integer multiples of screen pixels, so we can't entirely avoid rounding for screen display. The best we can do is make sure that the conversion from app units to screen pixels is nice and clean. That's what my proposal aims for: CSS pixels (the most important author unit) are an integer multiple of app units, and screen pixels are also an integer multiple of app units. 1/720 of a CSS pixel is essentially a fixed point unit. There's no need for the denominator to be a power of two, and in fact there are some advantages to not making it a power of two. I don't think we should use floating point for layout. That makes errors inevitable, such as in the case I suggested about making boxes fit exactly inside their parent. We may or may not want to use floating point for GFX scaling to device coordinates (we currently do). We might eliminate some problems by using integer arithmetic there too.
Maybe 1/60 would be a better number. It can still represent NNN.Npx exactly, and it still handles screens with 1, 2, 3, 4, 5 and 6 screen pixels per CSS pixel. It also gives us over a million lines of 10px text.
Should this be done now, or as part of Gecko 2 (and thus blocking bug 168884)?
Gecko 2 is just Gecko with bugs fixed.
QA Contact: gerardok → moied
The reporter in bug 178330 is asking for the same thing as dcone --- round every measurement to device pixels before we do layout. This has a number of unpleasant properties but if it's what the people want, I guess we could do it.
I suspect it's not what the CSS working group wants. In particular, I've heard mention that any size larger than zero should result in non-adjacency (but not necessarily visibility).
More WG guidance would be most welcome.
Kevin and I were discussing this yesterday and Kevin brought up the impact this would have on printing. Don do you have any thoughts on this? We must all keep in mind that printing is not a GFX/transformation matrix issue. It is a layout issue. We must always reflow for printing. It's not that I like twips, but when layout is in twips it is completely disconnected from any device or the definition of what a pixel may or may not be on any particular device. Also, fixing this, may fix 153080. I do see an interesting issue with font sizes and how/when they are defined with either Points or Pixels. Currently the conversion from Points to Twips is a device independent conversion (thus contributing to the problems in 153080) and Font sizes defined in Pixels, of coarse, is a device dependent conversion. So currently the PresShell defines a bunch of default faults in Pixels but their "real" size in twips cannot be calculated until there is a Device. So there is a bunch of really screwed up code throughout making assumptions about where the current font size in twips came from (again Bug 153080).
> We must all keep in mind that printing is not a GFX/transformation matrix > issue. Understood. However, I believe that print preview is, or at least should be, a process where we layout exactly as if we were printing, and then apply a GFX transformation to make the result fit on the screen. Correct me if I'm wrong. By the way, I hope you realize that CSS pixels are intended to be *somewhat* device independent. Read if you haven't already. Especially the example "In the second image, an area of 1px by 1px is covered by a single dot in a low-resolution device (a computer screen), while the same area is covered by 16 dots in a higher resolution device (such as a 400 dpi laser printer)." So CSS pixels are device dependent in that a CSS pixel should be an integral number of device pixels, but they are device independent in that we set that integer to get a CSS pixel as close to 1/96in as we can. [Hmm, the spec makes it sound like we should have a "how far away from the computer are you sitting" preference :-).] > but when layout is in twips it is completely > disconnected from any device or the definition of what a pixel may or may not > be on any particular device. Yeah, and that's actually a problem because it creates rounding errors visible on the screen unless the user selects certain "good" DPI values which make pixels an integral number of twips. We must guarantee that a screen pixel is an integral number of app units. That connection has to be there or we lose. That is true no matter how we fix bug 63336. [This constraint only applies when device pixels are bigger than app units. When device pixels are smaller than app units, as would sometimes be the case for printing under my proposal, I don't think we have to worry about the app unit/device pixel ratio, because the only visible rounding effects should be one-device-pixel changes in the sizes of objects depending on where they're positioned, and those changes would be very hard to spot.] Note that a major advantage of moving away from twips is that we let the user use arbitrary DPI values. The DPI value will only be used to convert physical lengths (pt, in, etc) to app units. I agree with you that it would be nice to have app units be as device independent "as possible" given the above constraints. That's why I'm proposing using fractions of CSS pixels as the basis. > So currently the PresShell defines a bunch of default faults in Pixels but > their "real" size in twips cannot be calculated until there is a Device. I think my proposal fixes this. A font size specified in CSS pixels maps to the same number of app units no matter what the device.
OK, so this all sounds good, now what do we all think about scaling? For discussions sake, let's let "scaling" be a reflow issue and "zooming" be a magnification issue (gfx transformation matrix). Note that "text zooming" (TZ)is a reflow issue. The fonts sizes are scaled (and nothing else is) and then reflow occurrs. Currently we turn off zooming for printing/PP for a couple of reasons: 1) Scaling was not designed into the current system so we co-opt the Canonical Pixel Scale for doing all the scaling and currently Point sized fonts are not scaled (Bug 153080) 2) If a user has TZ set AND scaling set what exactly does that mean? Do you just apply both? This could confuse end users as to the size of their final printout or what they see in PP.
Scaling is where the user enters "N%" into the Page Setup box? What exactly is that intended to do? If the user specifies 50% scaling, should a "1 in" CSS box be printed as 1 in or as 0.5 in? I assume the latter. If so, then here's how we do scaling by N%: -- GFX returns a page width/height multiplied by 100/N -- GFX returns the same DPI value that it would without scaling -- Layout as normal for printing. Layout knows nothing about scaling. -- GFX applies a transformation to scale everything by N/100 In other words, we just pretend the paper is bigger, and then we scale down the resulting output to fit onto the real paper. This is totally orthogonal to text zoom. If you scale by 50% then all text and images shrink by 50%. If you scale by 50% and text zoom by 2x then text ends up the same size but all images (and other measured lengths) shrink by 50%. If you do print preview with scaling in effect: -- GFX returns a page width/height multiplied by 100/N -- GFX returns the same DPI value that it would without scaling -- Layout as normal for printing. Layout knows nothing about scaling OR print preview (as far as units are concerned) -- GFX applies a transformation to scale everything by N/100 So far, just like above... -- GFX applies a transformation to scale everything by CSS pixels for window width/CSS pixels for page width Note that in both of these stacks there is a final transformation from CSS pixels to device pixels, and these transformations are different for the printer and the screen. Print Preview is totally WYSIWYG. Everything is laid out and rendered just as it will be printed, with a pure geometric transformation at the end to make the results fit in the window.
I don't quite understand how printing and screen units interact right now... but would the proposal on the table also address the issue we have where changing the screen dpi changes the size of the printed output?
Yes. Screen DPI would have no effect on printed output. Well, almost. Screen DPI would be used in two ways: -- GFX uses the screen DPI to figure out how many screen pixels to use for each CSS pixel, when rendering to the screen -- GFX uses the screen DPI (and the above decision) to compute how many CSS pixels there are per inch on the screen (this only affects documents being laid out for the screen, and does not affect layout for printing or print preview) The first decision affects printing of widgets whose dimensions are taken from the OS. E.g., if the OS tells us that a scrollbar is N pixels wide, then we need to convert that to CSS pixels for layout, and that conversion depends on the screen DPI. E.g. for a 200dpi screen we might choose to have 2 screen pixels per CSS pixel. So the OS says a scrollbar is 30 pixels wide, but that's really only 15 CSS pixels, and that's what we use in layout, for both screen and printing. Other than that, screen DPI should have no effect on printing.
Ok... so the size of images in printed output will still depend on the screen DPI then, for intrinsically sized images....
No, they won't. An intrinsically sized image of XxY pixels will be laid out with a size of exactly X x Y CSS pixels --- NOT screen pixels. Yes, this means for a 200dpi screen most images will be scaled by 2x in each direction, but this is actually what you want.
This is not an theoretical question, by the way. Some people where I work have IBM T221 displays, which are 200dpi. They look fabulous but you go blind surfing the Web with today's browsers. This proposal will largely fix that.
Ah, cool! I was wondering whether we'd treat image pixels as CSS pixels... That may make perf a bit worse on high-dpi displays (since we will be scaling continuously) but I think it's well worth it.
It will perf worse if CSS pixels != device pixels. It also will look worse, since the scaling algorithms are ... simplistic. See bug 98971, bug 73322, and bug 4821
Even the dumbest scaling algorithm will work fine, since we'll guarantee that theere is always an integral number of device pixels per CSS pixel. E.g., on a 200dpi screen, a normal image will display just as it would on a 100dpi screen, using a 2x2 block of device pixels for each image pixel. The only other simple alternative is simply to display everything at half the size, which I believe is worse. [Of course the best alternative would be to implement a better scaling algorithm so the image on a 200dpi screen would be the same size as the image on a 100dpi screen, but would actually look better. But that's beyond the scope of the issues here.]
Well, roc's proposal gets my vote. Let's roll! :-)
Component: Layout → Layout: Misc Code
QA Contact: moied → ian
Component: Layout: Misc Code → Layout
sorry for the component change.. didn't mean to. How is this going to impact viewing image on a site for the normal user? What I mean is if you have a 72x72 pixel image on a normal user's screen, we don't want to be resizing it at all.. it would be a huge perf hit if we were.
Component: Layout → Layout: Misc Code
We won't be resizing anything for 'normal' users. I suggest using the following equation, which seems to more or less follow from the W3C definition of a CSS pixel: devicePixelsPerCSSPixel = min(1, round(screenDPI/96)) So devicePixelsPerCSSPixel is 1 until screenDPI >= 144. There will come a day when normal users have screens with that kind of DPI --- but by then hopefully things will have sped up enough we won't be worrying about the scaling.
Doing this will not fix any round off problem.. we all agree to that? What are the other reasons to fix the units in Gecko, I don't understand what we gain using a CSS pixel or some scale of that as an internal unit. The internal unit we decide to use has to be a unit that maps well to all devices and any unit that is used on that device.. thats the only requirment for an interal unit.. correct? Also.. permormance issues.. and continually scaling.. BIG TIME HIT IN SPEED.. not something you can discount. Other problems we have.. and need to be addressed.. setting anything in pixel units (like fonts), what does this mean when we specify a pixel unit for a scren then print out. This causes alot of conversion problems..because the printer has a completely different pixel size. We have to be careful about what we are trying to solve and the bounds we want to stay in. We have been messing around with page layout issues and will get into this even deeper with things coming down the road... so I think its really important to really understand all the issues.
I feel like I've already thoroughly explained the requirements for units and how my proposal meets them. I'm not sure how to make it clearer. But I'll try to summarize. Layout has to work in units that are some integer fraction of CSS pixels. If we don't, then we incur roundoff errors in layout that mean, for example, 7 boxes of 7 pixels wide each might not fit exactly inside a 49-pixel wide box. [Right now layout DOES work in units that are some integer fraction of CSS pixels, but only because we force users to use a screen DPI that makes the twips to pixels ratio an integer. That sucks.] There are various other reason why doing layout in something related to CSS pixels makes sense: e.g. a CSS pixel is somewhat resolution independent, yet it will always have a nice relationship to screen pixels. Also, rods can tell you how we have font prefs specified in CSS pixels, and it's convenient to let them be device-independent. There are some roundoff issues that this does not solve --- namely, roundoff when we try to display a document that has features with finer resolution than the output device can display. Those problems are covered in bug 63336. Even if we decide to address those issues by (for example) rounding up the sizes of laid-out things to whole pixels, that can be done independently of the proposal here. > Also.. permormance issues.. and continually scaling.. I think I just explained in comment #27 that this would only happen for devices with high DPI (say, >= 144DPI), which is very few screens today. And when you do have a screen like that, scaling is better than going blind, trust me. > what does this mean when we specify a pixel unit for a scren then print out. I already addressed this at length. This proposal actually improves the printing situation a lot. Please remember that we're talking about CSS pixels and a CSS pixel is roughly the same size on the screen as on the printer.
Why don't we plan a meeting to discuss this. I have many questions, so does Rod. I can study and read in the meantime to understand what this all means to the GFX engine and what it would take to accomplish this for the advantages we want to get out of it.
I have written up something a little more logically organized:
If there's a meeting about this, please dial me in.
A CSS pixel 1/96 of an inch.. right (lets say normal arms and normal eye angle). I am getting stuck on this point.. I am assuming a CSS pixel is a virtual device at 96 dpi.. of certain angle and viewing distance. I will set up a meeting next week.. chim in if you want to be included. So far its me, roc,rod,ian and I will get kevin to attend.
A CSS pixel is not always 1/96 of an inch. From CSS 2.1: > Pixel units are relative to the resolution of the viewing device, i.e., most > often a computer display. If the pixel density of the output device is very > different from that of a typical computer display, the user agent should > rescale pixel values. What this means is that for typical computer displays a CSS pixel should be exactly a screen pixel, even if the screen is not 96dpi. So if you have a 72dpi screen a CSS pixel will be 1/72 of an inch; if you have a 120dpi screen, a CSS pixel will be 1/120 of an inch. If you have a 200dpi screen my proposal suggests we should make a CSS pixel be 1/100 of an inch (as close to 96dpi as we can get while still making a CSS pixel be an integral number of screen pixels). A phone meeting sounds fine.
I'd also want to be included in this meeting. Also note that this bug is somewhat independent of moving to correct CSS pixels, in that fixing this doesn't really require that we do so. So please don't get hung up on the scaling issues here. Nevertheless, moving to correct CSS pixels is something that we should do.
If the meeting is Tues or Thurs, I would like to dial in.
In comment 35, I probably should have said "moving to correct CSS pixels for screen displays", since we already scale CSS pixels for printing (although we may not do it quite right).
> Also note that this bug is somewhat independent of moving to correct CSS pixels That's true, but moving to correct CSS pixels for high-DPI screens imposes an additional constraint on the choice of N, so I think it's worth at least taking that into account now.
How about on Tuesday afternoon, early for us East-coast-ers? 1pm West coast time?
Could we give times in UTC or explicitly UTC-08 or -05 or whatever the relevant time zone is? I'm in UTC+00 at the moment.
Component: Layout: Misc Code → Layout: Block & Inline
Version: Trunk → 1.0 Branch
Here's my summary of yesterday's meeting: This general approach was viewed favourably by everyone. In particular the idea that fonts should be scaled by the current GFX transform seemed to be accepted. There is one major unanswered question, and it needs some deep thought. That is the question of whether layout should align boxes to device pixels, or whether all such rounding should be deferred to GFX. Looking at it another way, the question is how to render a fragment like this on a normal screen: <div style="width:6px; height:9px; background:black;"> <img style="width:1.5px; height:9px; background:red;"> <img style="width:1.5px; height:9px; background:green;"> <img style="width:1.5px; height:9px; background:blue;"> <img style="width:1.5px; height:9px; background:yellow;"> </div> If we round the IMG widths to device pixels during layout, then either an IMG will wrap to the next line or some black will be visible at the end of the line. Arguably, either is unacceptable. On the other hand, if we do the rounding during GFX rendering, then some of the IMGs will be rendered 1px wide and some 2px wide. One could also argue that that is unacceptable. So the question is, which unacceptable behaviour are we going to choose? :-)
I think rounding/conversion from app units to device units (pixels) should be done at the very last possible moment. Who knows when some code might want to add 3 "app units" to x-pos. Converting at layout restricts gfx from ever modifying position/size accurately.
My opinion is that on the whole, we should try to preserve layout even if it means strange rendering. If we have a layout which involves fractional pixels then we simply can't render it 100% accurately; I'd rather be off by a pixel here and there than to risk a gross layout change, such as causing something to wrap that shouldn't, or causing something to be uncovered that was supposed to be completely covered. However, I'm open to using specific hacks to get things aligned on screen pixels when we have leeway in certain situations. For example, no matter what layout units we use, we probably should lie about the size of 1mm or 0.1in to make it an integral number of screen pixels. Similarly karnaze has table hacks that try to align %-sized table columns on pixel boundaries. I think that's fine and good, but it works only because of this particular situation: authors shouldn't rely on getting an exact length from a percentage, and even if we give them a slightly incorrect length we know (because it's in a table) that we won't be screwing up the overall layout too much. When an author specifies an exact length I don't think we have quite the same freedom.
I agree that we should try to honour fractional pixels as closely as possible. In the example you give, I would expect, at page zoom = 200%, to see: +--+--+--+--+--+--+--+--+--+--+--+--+ | red | green | blue | yellow | +--+--+--+--+--+--+--+--+--+--+--+--+ ...and at 100%, either: +-----+-----+-----+-----+-----+-----+ | red |green| blue |yello| +-----+-----+-----+-----+-----+-----+ ...or: +-----+-----+-----+-----+-----+-----+ | red | green |blue | yellow | +-----+-----+-----+-----+-----+-----+ ...depending on our exact implementation. I think it is vital that zooming (or changing device, which should be exactly equivalent) should not radically change the layout, which is the only possibility if we don't do it this way. > For example, no matter what layout units we use, we probably should lie about > the size of 1mm or 0.1in to make it an integral number of screen pixels. Well we certainly can't do both at the same time. If we say 1mm is an integral number of pixels, that forces our resolution to an integral multiple of 25.4dpi, a kind of restriction from which we are trying to get away. Furthermore if we say that 1mm is an integral number of pixels then 0.1in must be an integral multiple of 2.54 pixels, which is most definitely not an integral number of pixels itself. Similarly, 1pt can only be an integral number of pixels if the resolution is an integral multiple of 72dpi. On another note, it was mentioned at one point that rounding at rendering time rather than at layout can cause odd-looking effects while scrolling, as the pixels fall on different boundaries. This can be easily solved by doing all the maths for rendering relative to the canvas origin rather than the rendering surface origin, and simply rendering the right part to the screen. Note that an OpenGL GFX would probably automatically have OpenGL do anti- aliasing for the pixels that contain two colours...
Alias: pixels
> Well we certainly can't do both at the same time. We can cheat by lying about the length of a millimetre and an inch. Suppose I have a 92dpi screen. Then 1mm is 3.54 dots. Let's lie and say it's 4 dots. We've made 1mm 10% bigger than the true millimetre. (Of course we then make 1cm be 40 dots.) Similarly, 0.1in is 9.2 dots, but we can lie and make it 9 dots. Then 1in is 90 dots so it's 2% smaller than a true inch. Furthermore 1in would be 22.5mm instead of 25.4. I wouldn't recommend messing around with points; just leave them at 1/72 of 1in. :-). > scrolling Actually this is most easily solved by making sure all scroll offsets are always screen-pixel-aligned. That's what we currently do anyway. > OpenGL GFX It's not always possible to anti-alias everything in OpenGL, and even on an anti-aliased device we want things to be pixel-aligned as much as possible, or it will look bad. But yes it would help us render stuff that can't be aligned, like my example.
Using the origin boundary will not solve this. We use lots of relative measurments, widths, etc, etc. So easily solved.. I disagree. As soon as you start drawing a path.. for example..and have an ending point.. and the next path depends on this ending point for its starting point.. all bets are off. Anti Aliasing in OpenGL.. is very expensive.. and not a gimme. Anti-aliasing is not built into OpenGL.. although there is mechanism with there buffers to Anti alias.. but its not part of OpenGL.. path drawing or line drawing. OpenGL is a 3D drawing enviorment.. so any 2D drawing is an after thought. You can not even draw concave polygons in OpenGL unless you supply your own routines to rasterize them.
The problems specificallly related to scrolling are easily solved. Other problems are not. OpenGL is WAY offtopic for this bug, but anti-aliased line and polygon rasterization is built into the spec (GL_SMOOTH mode), although implementations may or may not do it well. Also, you can draw concave polygons without rasterizing them yourself, using some clever tricks with the stencil buffer, XOR mode, and the "fan" triangle-drawing mode. But let's just forget Hixie ever mentioned OpenGL :-).
Re: comment 41 and ff. Interestingly, this problem is reminiscent to what happens with scalable fonts, when trying to fit a glyph sequence in a given, bounded, space. What happens with fonts is that those glyphs get scaled differently (a.k.a. non-linear scaling / hinting). With fonts however, the font designer _knows_ the list of characters encoded in their fonts, and hardcode a built-in mechanism that knows how to particularly weight 'm' vs 'a' vs etc, so that the overall result is almost always pleasant/uniform on any given string taken from the font under consideration. With abritray boxes however, say B1, B2, sometimes one would want the compensation to go in B1, sometimes one want it to go in B2, in a consistent way, (e.g., in a vertical stacking within table cells where the user may want a neat alignment of the right edge of B1 elements for example, or in the case of horizontal staking, a neat line up of the stacked baseline). It might be in this area that the trickiest problems could arise. You might this article useful: [BTW, TeX uses a hundredth (1/100) of the wavelength of the light as its internal unit. Hence while rendering might look jagged at places on the screen (eg with xdvi), when printed, round-off errors fall below the visible threshold and that's why it still looks so nice even with today's high resolution printers.]
> TeX uses a hundredth (1/100) of the wavelength of the light TeX has a luxury we do not -- it's rendering on something about the size of a printed page (give or take an order of magnitude). It only needs to keep track of what's going on on that one page. We're often forced to render on a canvas that's hundreds of thousands of lines tall... And the worst part is that with, eg, relative positioning things at the bottom of that canvas can affect what's going on at the top, so we can't just render one part of it. :(
Re: comment 45, I thought inaccurate computation of real-world units for arbitrary DPI settings was one of the problems this was intended to solve.
> > TeX uses a hundredth (1/100) of the wavelength of the light > > TeX has a luxury we do not Yep, I know that. The point is that even with its paranoic precision Knuth-like unit an all this compute time that makes some jealous (batch vs web-interactive), integer round-off errors arise, and are all the more crude on the screen (as noted in comment 46). So if folks do indeed agree that roc's proposal solves other issues in a foward-looking manner, then well, let's have a try, as a solve-all system is impossible. But using a fractional unit (as TeX or comment 2) might probably lay a much solid foundation for the proposal.
> :-) Psychotic test case writers are the people who will be proclaiming loudly that Mozilla is non-standards compliant if 2.54cm != 25.4mm != 1in != 72pt. Also, if we have a dpi setting then we really should make an effort to make it useful -- accurate lengths are a part of that.
Come to think of it, there's no need to worry that much about absolute units. If we do render them in apparently uneven ways, it'll only help convince people to avoid them. We're already trying to convince authors to only use 'em' and '%' anyway (most of them use 'px', grrr).
You don't want authors to specify line widths in ems, do you? Ugh. Most likely 0.1em will not be an integral number of screen pixels, in which case we are probably going to see rounding error artifacts in em-based layouts unless we do something *really* evil.
Line widths should typically be specified by using 'em' or '%' units, yes. How else can you get a scalable multi-media stylesheet? I don't understand why you would get any artifacts. Could you attach a non-pathological testcase which demonstrates what you mean?
Why can't we just round accurately but consistently, always scroll by integer numbers of pixels, and use units that are fractions of a pixel so that any integer number of pixels can be expressed in our units?
Here's a testcase which uses ems and will have problems when 0.1em is not an integral number of screen pixels; some of the boxes will have different sizes. Right now we do even worse, because we sometimes show one-pixel wide lines separating some of the boxes, but that's just an inconsistent rounding bug somewhere, not a design problem. If we rounded consistently we would never see this particular problem.
> How else can you get a scalable multi-media stylesheet? You can use px in a scalable multimedia stylesheet. It is roughly 1/72 of an inch. On a low resolution, antialiased device you SHOULD use px for line widths because it's the only way to guarantee you don't get crap-looking antialiased 1.5 pixel wide vertical and horizontal lines. And as far as I know, px is also the only way to guarantee you get boxes displayed with consistent sizes as measured in screen pixels. > Why can't we just round accurately but consistently, always scroll by integer > numbers of pixels, and use units that are fractions of a pixel so that any > integer number of pixels can be expressed in our units? I think we all agree that we should do these things at least. The question is whether we should do additional work to try to lay out objects at screen pixel boundaries. dcone says "yes". I say "sometimes, but we must be sneaky to avoid breaking layouts". Are you saying "no"? This question could be deferred, but it's worth considering now because one particularly convenient way to force things to lay out at screen pixel boundaries would be to make layout units be screen pixels. Personally I think this would break layouts too much (e.g. by causing inappropriate line wrapping), and so I think we should just choose to make app units be 1/60 of a CSS pixel and later see where we can judiciously add more round-to-screen-pixels logic to layout.
> some of the boxes will have different sizes Why is that a problem?
>> How else can you get a scalable multi-media stylesheet? > You can use px in a scalable multimedia stylesheet. Sorry, I should have been clearer. By "scalable" I meant one which honours the user's font size settings. The only valid use of pixels in stylesheets, IMHO, is for 1px borders.
>> some of the boxes will have different sizes > Why is that a problem? Because it looks ugly (and people file bugs about it). > By "scalable" I meant one which honours the > user's font size settings. Why should line widths depend on the user's font size settings? It sounds like you want font size settings to act as a general "zoom everything" control. I would rather implement a seperate "zoom everything" control in the UA for that. That would handle intrinsically sized images and everything, without burdening authors.
I'm not saying everything should obey the user's font size, and indeed I explicitly mentioned 1px borders as a good use for pixel units. But solid pages should be built with 'em's and '%'s so that they adapt to the user's environment and interact well with user stylesheets. >>> some of the boxes will have different sizes >> Why is that a problem? > Because it looks ugly (and people file bugs about it). I would much, _much_ rather have 0.7em sometimes being 10 pixels and sometimes 11 pixels rather than have strange, unpredictable hacks that make some or all of 0.1em, 0.1cm, and 0.1in be integral numbers of pixels, just to ensure that a subset of sites happen to render with square squares. How about the site that has 0.15em? Or the site that uses 0.5em of 0.5em? Or the site that uses percentages of a prime number of pixels? We can always come up with cases that will end up as non-integral. If authors want to ensure that they get exactly equal results, then they should indeed use pixels (and hope the user hasn't zoomed in or out). But in the majority of cases, we should just lay out documents at the subpixel level, and then render onto the pixel grid using well defined rules (e.g. a pixel takes the color of just inside its top right corner) and consistent rounding.
Here's a testcase for what I mentioned in comment 44: I think we can easily fix Gecko to render that correctly at all zoom levels, even 100%. (We actually don't do that badly at 800% in print preview at the moment.)
Just a little comment from a user POV (I've just read the whole bug history) Whatever unit you choose, please make it based on em not millimeters, inches or some hardwired "96dpi" rule. The days of the 96dpi screen are numbered. They'll die with crts. Processors are still getting faster and that's not just for games. (if you want some sort of rounding to make old devices speedier just make it an option - the user is clearly the one who can decide on a quality vs speed trade-off) I happen to use true 100dpi displays. Native apps know it and act accordingly. I can assure you wherever a browser (or another app) decides I shall use a 96dpi display I see it at once. So don't give me **** about 96=100 (or even 144=96). A user can feel it. I don't want to scale things in my mind or with my eyes just to save some CPU time. I *paid* (or my employer did) for a powerfull system to do things for me. Last I've checked it was a bargain compared to an eye operation some time in the future. Books are still the best reading media because they do *not* satisfy themselves with good enough (if you doubt this just try to xerox a few pages slightly changing the scaling between each of them, and then try to read it). The internet is still mostly a written media. That means accurate font scaling must/should be paramount in a browser, and hence internal units be em-based.
One comment on : I think I prefer N=20 over N=60, since I don't think thirds and sixths matter very much since it's not possible to specify them accurately in decimal. (The advantage is that it gives us three times the space to avoid nscoord overflow.)
Also, regarding making fonts obey the current transform -- currently nsTransform2D seems capable of representing any affine transformation (i.e., the composition of a linear transformation and a transposition) in 2 dimensional space. Most (all?) of its callers assume, however, that it represents only the composition of a scaling or a transformation (i.e., they assume it's two independent transformations on the two axes). We should perhaps make the code less general since we don't need it. However, I could imagine that we might want the ability to scale the two dimensions by different amounts (e.g., different dpi in each dimension, although this shouldn't matter for printing, I'd think), and it might be hard to make fonts obey that (although they might just come out "right" anyway). Should we throw out the ability of the transform to scale the dimensions by different amounts (we don't support that now anyway, although X does)? If not, how do we make fonts obey the current transform?
Never mind comment #66 -- I was forgetting about the issue of screens with 3 device pixels per CSS pixel.
Consider SVG ... everywhere we currently read the transform matrix is probably a bug. We need general affine transforms. IMHO font renderers that can't yet handle general affine transforms should just assert that the matrix is one they can handle.
If we want general affine transforms we need to fix all the rendering context methods that do anything with rects or ellipses -- i.e., anything that uses nsTransform2D::TransformCoord(nscoord *aX, nscoord *aY, nscoord *aWidth, nscoord *aHeight), since any caller of that method is inherently broken.
ahhh, I thought you were talking about callers from layout. Sure, callers within Gfx make all sorts of assumptions. But SVG-friendly Gfx implementations require tons of work anyway. As far as this bug is concerned we just need to change the font API to specify font sizes in app units.
Version: 1.0 Branch → Trunk
A start. Still some issues with this. The biggest ones are Print Preview is messed up, page margins are messed up, header/footer printing is broken, and there are some rounding problems, like text selectom moves text. This is also Windows-only at the moment because of the GFX changes.
Any chance you could summarize what it is that you're doing?
The basic idea is that I changed nsIDeviceContext so that instead of providing methods for converting between device pixels and twips, it provides methods for converting between device pixels and CSS pixels, CSS pixels and application units, and CSS Pixels and Twips. Then, I changed most of the consumers converting between device pixels and twips to converting between app units and CSS pixels. I think this is the right idea. This patch isn't really complete at the moment (not everything is converted, some things were incorrectly converted, and a lot of things are incorrectly labeled). Note that one I've fixed the patch, a 200dpi screen will display anything sized in CSS pixels using two device pixel per CSS pixel. I currently have it set up so the App Unit per CSS pixel ratio is a variable, but it only gets initialized to 60 and never actually changes (and probably will never need to change during runtime). Should I define it to be a fixed number? If so, should I get rid of the device context calls to retrieve this number?
Hey, this looks VERY cool. A few comments... Don't use #if 0. Just remove the code. > I currently have it set up so the App Unit per CSS pixel ratio is a variable, > but it only gets initialized to 60 and never actually changes. That's good. Let's leave it like that. - nsMargin(nscoord aLeft, nscoord aTop, - nscoord aRight, nscoord aBottom) {left = aLeft; top = aTop; + nsMargin(PRInt32 aLeft, PRInt32 aTop, + PRInt32 aRight, PRInt32 aBottom) {left = aLeft; top = aTop; Please don't do this and don't remove the nsIntMargin stuff! nscoord will change to float eventually. See bug 265084. I've never liked this two-step process where we obtain a scaling factor and then do some transformation. Couldn't we just have methods on nsPresContext, perhaps like this? // rounds down PRInt32 AppUnitsToIntCSSPixels(nscoord aV); nsIntPoint AppUnitsToIntCSSPixels(nsPoint aPt); etc? What do you think?
I think the scaling factor thing was done for performance reasons, although I can't see it being much of a win, considering how infrequently it's used. I'll go ahead and do that.
It should be relatively easy to make sure that the conversion functions are fully inlined, so performance is not an issue.
Still has issues: most noticeable are that images in printing are tiny and some text moves around when it is selected or hovered over. Large DPI screens are still mostly broken. I've been trying to fix device/css pixel issues, but haven't been making much progress; I must still have some of the device pixel and CSS pixel callers mixed up. I probably won't be able to do much on this during the week, but if anyone wants to help, that would be nice.
Attachment #177341 - Attachment is obsolete: true
I moved my units page into the wiki, it's now here: Feel free to edit. One thing I rembembered during the conversion is that it's really important that the CSS pixels to app units value be a constant across the entire app. So I think we should make CSSPixelsToAppUnits static in nsIDeviceContext. In fact I think for clarity's sake we should just have these in nsIDeviceContext: static PRInt32 AppUnitsPerCSSPixel() { return ...; } PRInt32 AppUnitsPerDevPixel() { return ...; } PRInt32 CSSPixelsPerInch() { return ...; } Everything else can be derived from these.
Okay, say we have only those 3 methods on nsIDeviceContext. Then say some code needs to convert from App Units to Device Pixels. Should every use be written out explicitly, or should there be some utility functions somewhere, say in nsIDeviceContext.h, like the following? inline PRInt32 NSAppUnitsToDevPixelsFloor(nsIDeviceContext* aDeviceContext, nscoord aAppUnits) { return NSToIntFloor((float)aAppUnits / aDeviceContext->AppUnitsPerDevPixel()); }
I think the main helpers should be on the prescontext. There could be static helpers in nsIDeviceContext for code that doesn't have a prescontext, but that's hardly anyone.
Should I convert callers to use the form nsPresContext::CSSPixelsToAppUnits(lengthInPixels) rather than aPresContext->CSSPixelsToAppUnits(lengthInPixels)?
yeah ...
Working patch with suggested interfaces. This still needs more work, primarly fixing printing and print preview and tracking down a few bugs. However, it is ready for comments at the interface level. Interface-wise, I think I need to change AppUnitsPerDevPixel on the PresContext to be a float. I don't see any other alternative that allows for a full zoom implementation in which device pixel rounding works. I know that a non-integral number of app units per dev pixel has been a source of bugs, but I don't see an alternative. Note that I'm not suggesting changing the device context method. Also, the difference would not be noticible with this patch alone.
The attachment appears to not be plain text... AppUnitsToDevicePixels shouldn't be affected by a general zoom transformation, since zooming should not affect layout. IMHO we should do a general zoom just by changing the active transformation in the device context when we render. This doesn't currently work because our current Gfx implementations don't apply the current transformation's scaling to fonts, but we'll fix this in the new Gfx implementation based on Cairo.
I posted the other one gzipped; I guess I shouldn't do that. The case I'm wondering about concerns, for example, the table code that rounds to pixels. Suppose zoom is set to 140%. If AppPixelsPerDevicePixel doesn't change for that document, we'll round to multiples of 60 app units when the number of app units in a device pixel is really 300/7~=42.86. Without accounting for that somehow, any code that does pixel rounding will be worthless.
Zooming shouldn't change the layout, so the table code is just going to have to do the right thing assuming no zoom. The results won't be quite right for non-integral scaling factors, but I think we can live with that.
Well, zooming should change the layout, since zooming in effectively shrinks the window width.
That really depends on what you want the zooming feature to be. If you want to generally surf around with everthing scaled up or down, sure, we'd adjust the window width so content fitted into the window. If you just want to temporarily zoom in to see some part of a page in more detail, I claim you don't want to change the layout.
There's significant user demand for the former (since it would fix many of the problems with text zoom)..
I'm trying to figure out where to store the scaling factor for printing. I'm thinking that I should store it on the view and have the view manager transform the rendering context in nsViewManager::Display and nsViewManager::Refresh. Does that sound like the best way to do it?
In theory it's now a generic scaling factor, not a printing-specific thing, right? We could store it on the PresShell and make it accessible via nsIViewObserver. How about that?
(In reply to comment #91) >. '1px' zoomed by 150% should render exactly the same as '1.5px' zoomed by 100%. I don't see the problem here.
Pages whose elements have dimensions like 1.5px tend to look kinda bad. Basically, some of the boxes will be 2px and others will be 1px depending on where they happened to be laid out. Sure, we can enable all possible zoom factors if you don't care about that.
So long as we're doing layout with sub-device-pixel resolution, rounding at rendering time is fine, IMHO. So yes, I think we should not worry about preventing users from being exposed to this problem.
I've kinda gotten scaling working, but I'm having some issues with clipping. At this point I'm looking for some comments on whether the way I'm storing zoom on the view make sense and some idea what I'm doing wrong with clipping.
It's looking really good. (In reply to comment #97) > Created an attachment (id=180349) [edit] > I've kinda gotten scaling working, but I'm having some issues with clipping. > At this point I'm looking for some comments on whether the way I'm storing > zoom on the view make sense I think doing per-view or even per-view-manager scaling is not a good idea, since it becomes unclear how to handle subdocuments with different scale factors. How about making the scale just be a property on the root view manager? It doesn't need to affect view coordinate setting at all. It just gets applied to the rendering context before painting and to incoming event handling coordinates. I think that would simplify your code quite a bit. > and some idea what I'm doing wrong with clipping. No idea, sorry. What sort of problems are you seeing?? If I were to put the zoom on the view manager, the PageContentFrame would need to be the root view in a view manager. Are you proposing I add that?
(In reply to comment #99) >? There is. > If I were to put the zoom on the view manager, the PageContentFrame would need > to be the root view in a view manager. Are you proposing I add that? No. I hadn't thought of the stuff that should not be scaled. That's really annoying... If we have to have a per-view scale factor, then at least call it nsIView::SetScale.. Am I missing something or is the scale factor not currently being applied to painting and event handling? That's all a big hairy ball to get right. I need to think about whether there's a better way.
I'm tempted to suggest that we impose a uniform scale over the entire view manager tree and deal with print preview by explicitly scaling up the scrollbar and page borders.
> If we have to have a per-view scale factor, then at least call it > nsIView::SetScale. Fine. I actually changed it to nsIViewManager::SetViewScale to keep with the precedent. >.
(In reply to comment #102) > >. I think unscaled is the correct way, anyway. > . Yeah. I think we'll just have to go with the per-view scale and make it work. We can simplify the task by enforcing a requirement that only views associated with certain kinds of frames are allowed to have a scale. That will limit the places where we have to consider coordinate transformations taking scale into account. For now, that would only be nsPageContentFrame.
(In reply to comment #100) > (In reply to comment #99) > > the things that need to be scaled. I'm actually a bit confused why what I'm > > doing isn't working. Is there not a single view tree for documents and their > > subdocuments? > > There is. Actually, that's only true for screen rendering at the moment (see). It's not really relevant to this bug, but I don't want anyone who reads this bug to get confused like I did.
This patch still has a couple visible issues. One is certain elements shifting around by a pixel when they are repainted. I'm not sure what's causing this. The other is that the platform-specific font measurement code doesn't account for the scaling factor. Because of the work that would be required to port the complete fix to all the GFX backends, I guess there's not much point to trying to get this in before Cairo is enabled by default. Therefore, I guess I'll just keep this patch up-to-date in my tree and let checking it in wait until Cairo is the default on all platforms. In case anyone's interested, with this patch, 200dpi screen rendering is usable, but there are some noticable issues. I think I'll end up disabling having more than one device pixel per CSS pixel on screens when this gets checked in and deal with the issues in another bug.
Attachment #180349 - Attachment is obsolete: true
That sounds fabulous.
Updated to trunk with some minor fixes for 200dpi displays. Besides the rounding and font scaling bugs, the only noticable issues with this patch are that on 200dpi displays, SVG doesn't scale and the autocomplete popup is in the wrong spot. The changes I've made to fix screen/device pixel bugs are kludgy, but they show where issues exist (look for conversions from device pixels to app units to CSS pixels). The biggest offender is improper use of DOM events, which have coordinates in CSS pixels, not device pixels. XUL really shouldn't be using the DOM coordinates to handle mouse events.
Assignee: roc → sharparrow1
I am interested in seeing bug 4821 implemented. Could someone explain whether fixing this bug (177805) automatically fixes bug 4821 or whether additional work is necessary? If the latter, what work would remain? Finally, I don't see a target milestone. Is there any desire to see this released soon (by the people working on this)? The last comment included a patch, but that was six months ago.
I figured I might as well post this in case anyone wants to look at it. Overall, it's working well for regular screen rendering, but it needs work in some areas and a some cleanup. Windows only at the moment, but it should be very easy to get working on other platforms that use thebes. I've been seeing some rounding problems testing at 200dpi: the most noticable is that images are rendering oddly, for example the toolbars and Google image searches. Other problems at 200dpi: There are problems rendering the borders of textboxes. I still need to fix tooltips. I haven't reimplemented scaling for printing yet. I'm getting a little confused about how to do it using display lists. Should I create my own kind of display item? And how do I hook it up to scale the content?
Attachment #185921 - Attachment is obsolete: true
We should think about how this work plays with the resolution independence in Mac OS X. See <>.
+ static const PRInt32 mAppUnitsPerCSSPixel = 60; Should be gAppUnitsPerCSSPixel + PRInt32 mAppUnitsPerDevPixel; + PRInt32 mAppUnitsPerInch; I think all 3 of these, and the return types of associated methods, should be nscoords. nscoord is the type of appunits.
(In reply to comment #110) > We should think about how this work plays with the resolution independence in > Mac OS X. See <>. It's pretty much exactly compatible with what hyatt was talking about there.
Actually I think we should get rid of mAppUnitsPerCSSPixel and just make it a build-time constant --- still sourced ultimately from nsIDeviceContext::AppUnitsPerCSSPixel(), as you have it now (so we can make it dynamic if we later want to).
I'll make those changes to nsIDeviceContext. roc, could you give me some pointers about doing scaling with display lists? I think I need a new type of item, but I'm not sure how to go about that.
> I haven't reimplemented scaling for printing yet. I'm getting a little > confused about how to do it using display lists. Should I create my own kind > of display item? I think the best approach would be to create an nsDisplayItem subtype for scaled page contents which is actually a leaf display item. When asked to paint, it would add a scale transformation to the incoming rendering context, build a display list for the page contents, paint it, and then pop the transformation. This avoids confusion because the page contents are really in a different coordinate system to the outside, so we shouldn't put them in the same display list structure. nsSVGForeignObjectFrame does something a bit like this. This would require more work if we want events to be able to target content in print preview, but we don't (currently). If we ever do need events, we follow/extend the way foreignobject works.
Okay, this takes care of most of the issues. One issue remaining is themed rendering and printing. Textboxes get printed with one-device-pixel thick lines, which are many times thinner than the borders on the screen.. I'm not sure if the MathML changes are right, particularly with the sub and sup frames; the behavior won't change, but I'm not sure it was right originally. Someone familiar with that code should check I'm doing the right thing. I fixed up the table code in a very minimal way. I'm not sure if I should worry about cleaning it up now, since this patch is already so big. I guess this patch will have some time to sit until Thebes progresses more (a couple months?).
Attachment #226459 - Attachment is obsolete: true
. OK, interesting. Not your bugs then. > I guess this patch will have some time to sit until Thebes progresses more (a > couple months?). Yes. We'll need to enable Thebes on all platforms and then get to the point where we don't want to support non-Thebes builds, not even for comparision purposes.
(In reply to comment #117) > ? > No, it doesn't, although foreignobject scaling gives weird results for scrollbars (weird in a different way from the result of my patch). The theme code isn't aware of CSS pixels, and I think that that is the issue. I'll have to look at the theme code a bit more carefully.
Flags: blocking1.9?
+ mAppUnitsPerDevPixel = AppUnitsPerCSSPixel() / PR_MAX(1, dpi / 96); I think this should round dpi/96. So e.g. 150dpi should lead to pixel-doubling. But I could be wrong, I'm not 100% sure.
(In reply to comment #116) > The other big issue is the painting of images. Toolbar images at 200dpi are > drawing with pixels from adjacent icons. At 200dpi, images are drawing with > blurry edges. This sounds like bug 324698 -- basically Cairo's image sampling is braindead and needs to be fixed. Making pixman do the right thing at image edges is on my todo list, I've just been very afraid to delve into that code.
(In reply to comment #119) > + mAppUnitsPerDevPixel = AppUnitsPerCSSPixel() / PR_MAX(1, dpi / 96); > > I think this should round dpi/96. So e.g. 150dpi should lead to pixel-doubling. > But I could be wrong, I'm not 100% sure. So, more like PR_MAX(1, (dpi + 48) / 96))? Of course, we can always tweak it even after the checkin. If we're going to support arbitrary scaling, though, we really shouldn't be using that number directly anywhere, becuase any assumptions about it will break when a page is scaled (the print stuff deals with it, but the screen is a lot more complicated.)
Status: NEW → ASSIGNED
(In reply to comment #121) > So, more like PR_MAX(1, (dpi + 48) / 96))? Of course, we can always tweak it > even after the checkin. Yeah. >?
(In reply to comment #122) > >? Nothing important. Stuff that's being taken care of by other work going on. Eventually it really shouldn't matter what it's set to because there will be other ways to scale everything.
Flags: blocking1.9? → blocking1.9+
While this may not come as much of a shock given that the last patch was made in June, WIP v8 fails miserably when attempting to apply it to the current trunk.
After reflow branch lands (next week or two), someone will need to remerge this, which will be tons of work, but then we can review and land it!
(In reply to comment #125) > After reflow branch lands (next week or two), someone will need to remerge > this, which will be tons of work, but then we can review and land it! My take: 1. What about BeOS and OS/2? A bit of research seems to show that BeOS is out of the picture because trunk doesn't build on their version of GCC. For OS/2, I've asked in the newsgroup, and hopefully I'll know soon. 2. Are cairo-gtk2 and cairo-cocoa really stable enough to completely break non-Cairo builds? (Or rather, will they be a month from now?) On Windows, everything seems to be working well enough, but I'm not sure whether the builds are good enough to dogfood for other platforms. 3. It might be a good idea to fix print preview before checking this in, so that it is possible to test print preview; on the other hand, it might be easier to fix once the canonical pixel scale stuff is gone. (The issue is that the AltDevice stuff is currently stubbed out in thebes gfx.) 4. (In reply to a private email.) Testing isn't really necessary at the moment, because the patch will have to be redone to apply to the trunk. Once the patch is updated, though, I'd really appreciate testing, since this patch touches a lot of code. 5. I think I'll update the patch sometime in December, unless someone else wants to do it. Unfortunately, most of it has to be done by hand; having the older patch as a reference helps, of course, but the codebase constantly changes, and each conversion really needs to be evaluated individually because people haven't been very good about using the right conversions (the distinction between scaled pixels and device pixels is only visible when printing at the moment). 6. Is the reflow branch really landing in a week or two? Cool :) (Nobody's mentioned it online.)
1. I believe OS/2 currently builds. However it does lag behind at times and I don't think it should block landing a patch of this magnitude and importance, so feel free to break it if that helps. 2. cairo-gtk2 and cairo-cocoa are on by default now. We'll ship them or die trying. We are ready to break non-cairo builds, or we will be soon enough. 3. It would be good if we could have print preview working in this patch. How hard is that? 4. I'm sure our QA supermen (and women) are standing by! 5. Makes sense. You definitely only want to update once more before landing. 6. The trunk is closed and being prepped for 1.9a1, which should happen next week. I believe we're landing reflow branch ASAP after that.
Well 6. isn't quite accurate, but basically yes, we're landing the reflow branch soon.
(In reply to comment #127) > 3. It would be good if we could have print preview working in this patch. How > hard is that? Print preview isn't actually broken; it's just not using all the right data. What's currently broken is that we're not getting the right page size and the font measurement is wrong. Hmm, actually it should be really easy to fix those, since they only occur in a few places. Print preview font rendering would still be broken, but that never worked properly (bug 109970).
(In reply to comment #127) > 1. I believe OS/2 currently builds. However it does lag behind at times and I > don't think it should block landing a patch of this magnitude and importance, > so feel free to break it if that helps. Robert, I agree. But I still wanted to point out that it was really nice of Eli to post in the OS/2 newsgroup and point us to this issue. The manpower on the OS/2 port is very limited so any hints like these are always very welcome!
Okay, new version updated to trunk. I think this is ready for testing, if anyone is willing. I'd especially appreciate testing with high-resolution displays and printing, since issues are much more likely to come up there (print preview acts sort of strange, but that's true on the trunk as well). Known issues: there are some rounding errors rendering images and some minor theming issues when device pixels per CSS pixel is more than 1. Also, SVG is not rendering at the right size. If anyone wants to test changing the dpi, you can use the hidden preference "layout.css.dpi". You don't have to restart the browser, but changing it messes up currently open windows. I'm going to do a bit more cleanup before asking for review.
Attachment #227012 - Attachment is obsolete: true
One issue I'm seeing with the patch is that there is a border missing for the inactive tabs in Firefox, see:
I'm seeing white stripes on images in print preview, when they are out of view and then scroll them into view: (image url: )
.
(In reply to comment #134) > . Not quite right. Both tabs are shifted 1px down, which means that the inactive tab's label and both close buttons are shifted 1px up.
With the patch, iframes without width or height set, seem to get no width:
Text is moving with the patch when selecting some text on this page: I found this on
(In reply to comment #136) > With the patch, iframes without width or height set, seem to get no width: > Fixed. (In reply to comment #137) > Text is moving with the patch when selecting some text on this page: > > I found this on Fixed. Thanks for the testing! I'm still working on fixing the other issues, and I'll post a new patch when I'm done.
We're hot to try this on the OLPC machine which has a 200dpi screen. When you update the patch again, we'll try it out.
Assuming SVG isn't totally broken, can somebody with the patch applied please test if it fixes bug 364638? There should be 6 black squares in the testcase in attachment 250406 [details], not just 4.
Mac OS X, the patch applied well, but I get a build error almost immediately: /Users/phiw/cocoafox/mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp: In member function 'nsresult nsThebesDeviceContext::SetDPI(PRInt32)': /Users/phiw/cocoafox/mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp:139: warning: unused variable 'OSVal' /Users/phiw/cocoafox/mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp: In member function 'virtual nsresult nsThebesDeviceContext::Init(void*)': /Users/phiw/cocoafox/mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp:225: error: 'NSFloatPointsToTwips' was not declared in this scope /Users/phiw/cocoafox/mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp: In member function 'virtual nsresult nsThebesDeviceContext::GetDeviceContextFor(nsIDeviceContextSpec*, nsIDeviceContext*&)': /Users/phiw/cocoafox/mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp:521: warning: unused variable 'rv' make[6]: *** [nsThebesDeviceContext.o] Error 1 make[5]: *** [libs] Error 2 make[4]: *** [libs] Error 2 make[3]: *** [libs_tier_gecko] Error 2 make[2]: *** [tier_gecko] Error 2 make[1]: *** [default] Error 2 make: *** [build] Error 2
philippe: try replacing NSFloatPointsToTwips(x) with NSToCoordRound(x * mAppUnitsPerInch / 72.0f);
Eli, I think we should introduce helpers for converting nsRects, nsMargins, and nsPoints in various directions. However, if you don't want to do it in this patch, then that's fine with me.
(In reply to comment #142) > philippe: try replacing NSFloatPointsToTwips(x) with > NSToCoordRound(x * mAppUnitsPerInch / 72.0f); > Thanks, this brought me further. Unfortunately, more breakage: /Users/phiw/cocoafox/mozilla/widget/src/cocoa/nsNativeThemeCocoa.cpp:286: error: 'class nsDerivedSafe<nsIDeviceContext>' has no member named 'TwipsToDevUnits' /Users/phiw/cocoafox/mozilla/widget/src/cocoa/nsNativeThemeCocoa.cpp:351: error: 'NSTwipsToIntPixels' was not declared in this scope make[6]: *** [nsNativeThemeCocoa.o] Error 1 ... (but now I have to bail out, got some work to do.) Looking forward to an updated patch.
I think you want float t2p = 1.0f/dctx->AppUnitsPerDevPixel(); and replace NSTwipsToIntPixels with NSAppUnitsToIntPixels.
(In reply to comment #140) > Assuming SVG isn't totally broken, can somebody with the patch applied please > test if it fixes bug 364638? There should be 6 black squares in the testcase in > attachment 250406 [details], not just 4. This isn't going to fix that bug. The style system stores all lengths in app units, and this patch doesn't change that; it just changes what exactly app units refer to.
Updated patch. Problem with lines in print preview partially fixed, although image rendering really needs some work. Problem with tabs not fixed; I'm not sure what's causing the problem. For 200dpi displays, there are some significant issues with "cracks" in web layouts (for example,). SVG scaling isn't working; I'm not completely sure how it should work, though. I also need to look into why the scrollbars look bad (at least on WinXP).
Attachment #250329 - Attachment is obsolete: true
(In reply to comment #145) > I think you want > > float t2p = 1.0f/dctx->AppUnitsPerDevPixel(); > > and replace NSTwipsToIntPixels with NSAppUnitsToIntPixels. Not right; NSAppUnitsToIntPixels takes integer app units per pixel. (This is different from the old NSTwipsToIntPixels behavior, which took floating point pixels per app unit.) I put in the Mac changes from the comments in the latest patch; I'm not sure if it compiles yet, though.
(In reply to comment #148) > > I put in the Mac changes from the comments in the latest patch; I'm not sure if > it compiles yet, though. > Unfortunately, is still doesn't compile: The error I get now: /Users/phiw/cocoafox/mozilla/widget/src/cocoa/nsDragService.mm: In function 'NSRect GetDragRect(nsIDOMNode*, nsIScriptableRegion*)': /Users/phiw/cocoafox/mozilla/widget/src/cocoa/nsDragService.mm:150: error: 'class nsPresContext' has no member named 'TwipsToPixels' /Users/phiw/cocoafox/mozilla/widget/src/cocoa/nsDragService.mm:153: error: 'NSTwipsToIntPixels' was not declared in this scope make[6]: *** [nsDragService.o] Error 1
I think float t2p = frame->GetPresContext()->TwipsToPixels(); nsRect screenOffset; screenOffset.MoveBy(NSTwipsToIntPixels(widgetOffset.x + viewOffset.x, t2p), NSTwipsToIntPixels(widgetOffset.y + viewOffset.y, t2p)); outRect.origin.x = (float)screenOffset.x; outRect.origin.y = (float)screenOffset.y + (float)NSTwipsToIntPixels(rect.height, t2p); outRect.size.width = (float)NSTwipsToIntPixels(rect.width, t2p); outRect.size.height = (float)NSTwipsToIntPixels(rect.height, t2p); should be nsPresContext* presContext = frame->GetPresContext(); nsRect screenOffset; screenOffset.MoveBy(presContext->AppUnitsToDevPixels(widgetOffset.x + viewOffset.x), presContext->AppUnitsToDevPixels(widgetOffset.y + viewOffset.y)); outRect.origin.x = (float)screenOffset.x; outRect.origin.y = (float)screenOffset.y + (float)presContext->AppUnitsToDevPixels(rect.height); outRect.size.width = (float)presContext->AppUnitsToDevPixels(rect.width); outRect.size.height = (float)presContext->AppUnitsToDevPixels(rect.height);
Doesn't compile on my Linux box: /home/user/soft/src/mozilla-test/mozilla/gfx/src/ps/nsDeviceContextPS.cpp: In member function `virtual nsresult nsDeviceContextPS::InitDeviceContextPS(nsIDeviceContext*, nsIDeviceContext*)': /home/user/soft/src/mozilla-test/mozilla/gfx/src/ps/nsDeviceContextPS.cpp:206: error: `mTwipsToPixels' was not declared in this scope /home/user/soft/src/mozilla-test/mozilla/gfx/src/ps/nsDeviceContextPS.cpp:206: error: `NSIntPointsToTwips' was not declared in this scope ...and so on.
After doing the edits suggested in comment 150 (thanks Roc), I got one step further. Now I get errors in widget/src/cocoa/nsChildView.mm: : 'class nsIDeviceContext' has no member named 'DevUnitsToAppUnits' /Users/phiw/cocoafox/mozilla/widget/src/cocoa/nsChildView.mm:1915: error: 'NSIntPixelsToTwips' was not declared in this scope (and there will possibly come more errors after that one ...)
(In reply to comment #151) > Doesn't compile on my Linux box: > /home/user/soft/src/mozilla-test/mozilla/gfx/src/ps/nsDeviceContextPS.cpp: In you shouldn't be building gfx/src/ps in a cairo build. this patch won't work in non-cairo builds
(In reply to comment #153) > you shouldn't be building gfx/src/ps in a cairo build. this patch won't work > in non-cairo builds Oops, sorry. However, it still doesn't compile with ac_add_options --enable-default-toolkit=cairo-gtk2 in .mozconfig: src/thebes/nsThebesDeviceContext.cpp:247: error: `NSFloatPointsToTwips' was not declared in this scope Compiles cleanly without the patch. But I'm not sure I'm doing everything right.
Added in compile fixes for Mac and Linux; untested, though.
Attachment #250482 - Attachment is obsolete: true
WIP v11 on Mac: 1/ one file (widget/src/cocoa/nsNativeThemeCocoa.cpp) does not exist anymore, it is replaced by widget/src/cocoa/nsNativeThemeCocoa.mm (it is the same file, by contents, actually).. 3/ I ran tested that build for a while. It works. Noticed that text in .GIF files was quite blurred. Didn't find any major layout breakage. I did see some problems with irregular spacing between elements in a page in the vertical space. But those are possibly coming from Cairo bugs.
(In reply to comment #156) > WIP v11 on Mac: > 1/ one file (widget/src/cocoa/nsNativeThemeCocoa.cpp) does not exist anymore, > it is replaced by > widget/src/cocoa/nsNativeThemeCocoa.mm > (it is the same file, by contents, actually). Okay; I'll update to tip. >. That function basically does the opposite, so it might cause problems. (I'm not sure how critical that code is.) > 3/ I ran tested that build for a while. It works. > Noticed that text in .GIF files was quite blurred. Blurred? What do you mean? Is the UI getting scaled up? >.
(In reply to comment #157) > (In reply to comment #156) > >.) Oops... missed that. Corrected that here, and the build completed successfully. (and while I was at it, I made a Camino build as well. No problems either.) > > 3/ I ran tested that build for a while. It works. > > Noticed that text in .GIF files was quite blurred. > > Blurred? What do you mean? Is the UI getting scaled up? .gif files inserted as elements in an html page. Here is an example: <> from <> The text is slightly blurred in the centre. The impression is one of an image that has been slightly resized in the html (but they are not). On a site under development, I have a gif file that is positioned exactly on top of a .gif background image. With this patch applied, there is a 1px offset. Resizing the window horizontally, and I can see the image slightly moving; sometimes the alignment on the left is correct, sometimes on the right. The look of it is that the foreground image has a 1px difference in size. Both foreground and background image have exactly the same (horizontal) size. Here is an other example, from the Fx toolbar; there is a 1px difference in size between the images and the inputfields <> > > >. > Here is an example. Screenshot <> page: <>. Zooming in or out gives different result. All tests on a standard PowerBook 1.5Ghz, running OS X 10.4.8.
> - view->ScrollTo(NSToCoordRound((aX/ratio)*context->PixelsToTwips() - portRect.width/2), > - NSToCoordRound((aY/ratio)*context->PixelsToTwips() - portRect.height/2), > + view->ScrollTo(nsPresContext::CSSPixelsToAppUnits(aX/ratio) - portRect.width/2, > + nsPresContext::CSSPixelsToAppUnits(aX/ratio) - portRect.height/2, Typo: aX vs aY
> @@ -954,36 +954,31 @@ nsThebesRenderingContext::DrawImage(imgI ... > - nsIntRect pxDr; > - pxDr.x = NSToIntRound(FROM_TWIPS(twDestRect.x)); > - pxDr.y = NSToIntRound(FROM_TWIPS(twDestRect.y)); > - pxDr.width = NSToIntRound(FROM_TWIPS(twDestRect.width)); > - pxDr.height = NSToIntRound(FROM_TWIPS(twDestRect.height)); > + > + nsIntRect pxDr(twDestRect); > + pxDr.ScaleRoundOut(1.0f / mP2A); This adds 1 pixel to width and height for small (normal) integers because 60 * (1/60) > 1 when using floats and ScaleRoundOut() uses ceil(): The "gaps" testcase in comment #158 shows normal behavior if 1em doesn't happen to translate to an even number of pixels (or the margin between h5's isn't 1em).
I'm seeing a few problems on linux: * Tabs are shifted down (similar to comment #123) * There's a line of screwed up drawing up in the title bar * One of the rounded images in the minefield start page is drawing way too big vs..
Maybe the following code in nsCSSRendering::PaintBackgroundWithSC shouldn't be disabled for Linux? #if (!defined(XP_UNIX) && !defined(XP_BEOS)) || defined(XP_MACOSX) // Setup clipping so that rendering doesn't leak out of the computed // dirty rect aRenderingContext.PushState(); aRenderingContext.SetClipRect(dirtyRect, nsClipCombine_kIntersect); #endif But I don't see how this could interfere with the units patch.
Just ran across a bug in WIP v11. At line 3025 we have if (offset.x > width || offset.y > height) { - p0.x = - floor(fmod(offset.x, gfxFloat(width)) + 0.5); - p0.y = - floor(fmod(offset.y, gfxFloat(height)) + 0.5); + p0.x = - floor(fmod(offset.x, gfxFloat(width) * scale) + 0.5); + p0.y = - floor(fmod(offset.y, gfxFloat(height) * scale) + 0.5); } else { The if statement should be changed too, and compare offset to the scaled width and height.
I have made an updated patch, diffed against yesterday's tree: What I noticed with the patch is that images are sometimes 1 pixel higher/wider: I suspect this is the reason why the 'Go' and 'Search' button are 1 pixel too high and have a weird dent. I also suspect (or rather hope) that it is the issue for the missing border line for inactive tabs (comment 132).
(In reply to comment #162) >. > Okay, now I feel kind of silly; I just checked, and it turns out the tabs are srewed up in exactly the same way without my patch (at least on my computer). I might still be introducing a regression, but at least now I'm convinced there are factors outside this patch messing up the tabs. In my latest version, I've gotten rid of the GFX changes that were causing additional rounding problems in various other places. This unfixes the gaps in zoomed content in print preview, but I think that's more of a fundamental problem with the image rendering APIs. I'll post my latest version shortly.
Current state. I think I might have to mess with the sizing/painting for native widgets a bit more to get them to cooperate with printing correctly, although that's not a very high priority. About the issue with the tabs: I'm pretty sure there is some issue, but there's also an issue relating to tabs and certain OS font sizes, which masks the rounding issue.
Attachment #250639 - Attachment is obsolete: true
I'll update this for the textrun landing...
> What I noticed with the patch is that images are sometimes 1 pixel > higher/wider: > > I don't see these bugs on Mac, nor the related chrome bugs..
Attachment #252251 - Attachment is obsolete: true
Obsoletes for v13? I tried to made some fixes for gfx/src/gtk2.... It buildable, but does not work ;( Eli could you look on my changes for gfx/src/gtk... and say what wrong there?
(In reply to comment #170) > Created an attachment (id=252688) [details] > Patch v13 > >. Okay, that's fine with me. I haven't seen anything more than minor cosmetic regressions, so it shouldn't be an issue to land. Is there anything else I should do at this point?
I think we should probably just leave it alone until we land unless a significant new regression is found. There will probably be one more merge required just before we land. romaxa: My v13 works on gtk2, maybe you can use it?
Comment on attachment 252704 [details] [diff] [review] Updated to trunk (After gfx landing) We're not going to put a non-cairo version of this patch on the trunk, so this discussion is not appropriate for this bug. I'll send you an email shortly.
Attachment #252704 - Attachment is obsolete: true
v12 (v13 is just a subset of v12?) still contains code like ScaleRoundOut(1.0f / mContext->AppUnitsPerDevPixel()) These will all cause 1px rounding errors (don't know how critical they are) for integral multiples of AppUnitsPerDevPixel. I'd suggest to create an InvScaleRoundOut() which divides by the supplied parameter instead of multiplying, and use InvScaleRoundOut(mContext->AppUnitsPerDevPixel()) instead. Then the result will be exact (unless compiled with -ffast-math or similar). BTW, NSToIntRound() causes problems too, but that is unrelated to this patch (filed bug 368280). + mAppUnitsPerDevPixel = AppUnitsPerCSSPixel() / PR_MAX(1, dpi / 96); Should be PR_MAX(1, (dpi + 48) / 96), see comment 121.
Sorry, v13 was an incomplete diff. I have a bad habit of doing that. I'll update to trunk and post a new patch. Most of the time where we use ScaleRoundOut, we use it because it's explicitly OK to make the rectangle bigger than necessary. Of course, we probably do sometimes get it wrong, and making it bigger than necessary can cost performance too. One of the good things about this patch is that it gives us a more solid framework for fixing various kinds of rounding bugs. Thanks for the reminder about the mAppUnitsPerDevPixel calculation. We probably should change it. It would be nice to make contact with people who have a range of exotic screens and get some feedback about exactly when pixel-doubling should kick in...
(In reply to comment #176) > It would be nice to make contact with people who have a range > of exotic screens and get some feedback about exactly when pixel-doubling > should kick in... When 150 dpi screens become usual, people probably want a pref for it (for the 48 in the formula, which should be the default). Would also be useful for testing.
Will this bug fix resolution independent UI in OS X? If not I will submit a new bug. Currently it is broken.
I would think this would allow res independence, or at least be a blocker bug for it. As there are more and more devices appearing that will support different resolution independence this would be nice. I think the Nokia 770/n800 run at around 225 dpi, and the iPhone will be 160dpi I think plus MacOSX supporting it (probably for the iPhone) this would be nice.
Also there are some problems with changing DPI for SVG content, it is not changed at all.) { **************************************************** For setup any DPI value: mozilla/gfx/src/thebes/nsThebesDeviceContext.cpp: - mAppUnitsPerDevPixel = AppUnitsPerCSSPixel() / PR_MAX(1, dpi / 96); + mAppUnitsPerDevPixel = ((float)AppUnitsPerCSSPixel() * 96.) / (float)dpi;
(In reply to comment #180) > Also there are some problems with changing DPI for SVG content, it is not > changed at all. Yeah; I spent some time on that, but I never came up with a fix that seemed correct. I'm not really sure where the SVG code decides what scaling factor to use for rendering when there isn't a width/height set. >) { Makes sense; it works? (I'm assuming this is in addition to a forced reflow.) On a side note, it might be a good idea to make the DPI pref change listener force a reflow; I might add it eventually, although in general listening to the pref doesn't really make sense without some way to detect DPI changes.
> + mContext->AppUnitsPerDevPixel() == fm->mAppUnitsPerDevUnit) { With this changes we are not required to do force reflow... we just need to redraw layout (widget_redraw).... Tested ;) at least for text elements it works fine..
(In reply to comment #183) >. Okay; I actually have a version of the patch synced to tip, just so you know. Do you want me to post it?
Yes please. I'll compare it to mine. Also, I assume you've done some testing to make sure it's still working as expected on Windows? (I've just tested my patch on Mac and things seem OK, Linux next.)
Ok, there are some problems with changing DPI on fly ;) data:text/html,<input type=submit works ok after changing layout.css.dpi preference + window_redraw.... But if I will do reload for the page... button looks like "bold".. and not scaled properly... this not happens for next url: data:text/html,<input type=submit Something done for Images in mozilla/content.... but nothing for other form controls... I'm trying to do changing of DPI on fly without any reflow... ;), in this case we can create same zoom (full page, and such fast), like in Opera and IE7.
Changing DPI on the fly is not the best way to do zooming, and I don't think it's a high priority for Firefox, although I guess we'd accept patches for it if it's not too much code.
But anyway it works wrong, if we will change DPI for google.fi and do reload of the page... . images and other elements change sizes, but there are some problems for input field and buttons... same not happens for ya.ru page, even after reloading the page.
(In reply to comment #188) > But anyway it works wrong, if we will change DPI for google.fi and do reload > of the page... . images and other elements change sizes, but there are some > problems for input field and buttons... > same not happens for ya.ru page, even after reloading the page. I have no idea how any layout data could survive a reload; is that really what's happening on ya.ru? Native theme widgets are intentionally sized in device pixels. That shouldn't apply to inputs or buttons, though. (In reply to comment #187) > Changing DPI on the fly is not the best way to do zooming, and I don't think > it's a high priority for Firefox, although I guess we'd accept patches for it > if it's not too much code. Yeah, it's not really very useful for Firefox. We could make a really cool slider thing that makes the UI change size, but it would be pretty worthless. That said, I don't think it'll be very much code, so we could probably take it on the trunk. (I think it should just be a restyle reflow plus clearing the font cache, and maybe a couple of other things I don't know about.) Actually, Oleg, could you file a separate bug and mark it depending on this one? It would make things much clearer. (In reply to comment #185) > Yes please. I'll compare it to mine. Also, I assume you've done some testing to > make sure it's still working as expected on Windows? (I've just tested my patch > on Mac and things seem OK, Linux next.) Yeah, it's still working; I'll post it in a moment.
If you guys are ready to land this, we can keep the tree closed through tomorrow to give you a chance to do so -- just let me know. Assuming the builds from today are ok we'll check in the version bump to 1.9a3 and then you guys should be good to land tomorrow morning (pacific), possibly even tonight. I'd like to reopen the tree by tomorrow later afternoon pacific time though.
A quick review of the changes in mozilla/layout/style/ in attachment 252251 [review]: The nsCSSGroupRule.cpp change seems unrelated to this patch. Though the only reasons not to include it are that the code in question should be cleaned up further, and that it might make the history confusing. Fixing the indentation in nsComputedDOMStyle::GetTextIndent, GetMaxHeight (twice), GetMaxWidth (twice), GetMinHeight, and GetMinWidth, would be nice, though. I noticed the change of rounding in nsROCSSPrimitiveValue::SetAppUnits(float), although I suspect it's fine. Seems like it would be good to file a followup bug on making nsCSSValue not use twips anymore. I also suspect you know what you're doing with changing nsStyleBorder and nsStyleOutline from using the pres context's conversion to the device context's, although I don't follow it.
(In reply to comment #192) > A quick review of the changes in mozilla/layout/style/ in attachment > 252251 [details]: > > The nsCSSGroupRule.cpp change seems unrelated to this patch. Though the > only reasons not to include it are that the code in question should be > cleaned up further, and that it might make the history confusing. I'll try to remember not to check that in. > Fixing the indentation in nsComputedDOMStyle::GetTextIndent, > GetMaxHeight (twice), GetMaxWidth (twice), GetMinHeight, and > GetMinWidth, would be nice, though. Okay. > I noticed the change of rounding in > nsROCSSPrimitiveValue::SetAppUnits(float), although I suspect it's fine. I'm pretty sure that rounding down percentage values is wrong. Being off by a twip doesn't usually make a difference, though. > Seems like it would be good to file a followup bug on making nsCSSValue > not use twips anymore. Okay; do you prefer inches? Points? Passing in a prescontext? Or something else? (They're real twips because they come from physical units.) > I also suspect you know what you're doing with changing nsStyleBorder > and nsStyleOutline from using the pres context's conversion to the > device context's, although I don't follow it. Erm, I think you read that part of the patch backwards? It doesn't look the least bit unusual to me.
I ran reftests and got two unexpected failures, one related to my test setup and one which was already a known problem on Mac, so I think that's OK. Review comments: nsEventListenerManager::PrepareToUseCaret and GetCoordinatesFor need to be documented that they return device pixel coordinates. Somewhere we need to document that gfxContext is normally in (transformed) device coordinates, so the caller of gfxContext methods is responsible for applying an appunit to device-pixel conversion. I think this is ugly and counterintuitive --- if you transform to device coordinates you don't expect scaling, translation or rotation happen after that --- but that's the way it is for now. I'm a bit concerned about nsCanvasRenderingContext2D::Render. It uses mWidth/mHeight as if they were device pixels, but they're actually CSS pixels. We need a followup bug to fix <canvas> for high-DPI situations. (Basically we need to make the internal surface width/height match the target device's device pixels, so that scaling happens when script draws into the canvas, not at render time.). + * Get/set the print scaling level Could you say something in the comment about when it's safe to call SetPageScale and what it does? Index: layout/generic/nsHTMLReflowState.cpp + mComputedPadding.top = nsPresContext::CSSPixelsToAppUnits(mComputedPadding.top); + mComputedPadding.right = nsPresContext::CSSPixelsToAppUnits(mComputedPadding.right); + mComputedPadding.bottom = nsPresContext::CSSPixelsToAppUnits(mComputedPadding.bottom); + mComputedPadding.left = nsPresContext::CSSPixelsToAppUnits(mComputedPadding.left); + nsPresContext::CSSPixelsToAppUnits(mComputedBorderPadding.top); mComputedBorderPadding.right = - NSIntPixelsToTwips(mComputedBorderPadding.right, p2t); + nsPresContext::CSSPixelsToAppUnits(mComputedBorderPadding.right); mComputedBorderPadding.bottom = - NSIntPixelsToTwips(mComputedBorderPadding.bottom, p2t); + nsPresContext::CSSPixelsToAppUnits(mComputedBorderPadding.bottom); mComputedBorderPadding.left = - NSIntPixelsToTwips(mComputedBorderPadding.left, p2t); + nsPresContext::CSSPixelsToAppUnits(mComputedBorderPadding.left); These should all be devpixels, right? Ditto in layout/xul/base/src/nsBox.cpp, nsBox::GetPadding/nsBox::GetBorder. if (aEvent->message==NS_TEXT_TEXT) { - ((nsTextEvent*)aEvent)->theReply.mCursorPosition.x=NSTwipsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.x, t2p); - ((nsTextEvent*)aEvent)->theReply.mCursorPosition.y=NSTwipsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.y, t2p); - ((nsTextEvent*)aEvent)->theReply.mCursorPosition.width=NSTwipsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.width, t2p); - ((nsTextEvent*)aEvent)->theReply.mCursorPosition.height=NSTwipsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.height, t2p); + ((nsTextEvent*)aEvent)->theReply.mCursorPosition.x=NSAppUnitsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.x, p2a); + ((nsTextEvent*)aEvent)->theReply.mCursorPosition.y=NSAppUnitsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.y, p2a); + ((nsTextEvent*)aEvent)->theReply.mCursorPosition.width=NSAppUnitsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.width, p2a); + ((nsTextEvent*)aEvent)->theReply.mCursorPosition.height=NSAppUnitsToIntPixels(((nsTextEvent*)aEvent)->theReply.mCursorPosition.height, p2a); } if((aEvent->message==NS_COMPOSITION_START) || (aEvent->message==NS_COMPOSITION_QUERY)) { - ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.x=NSTwipsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.x,t2p); - ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.y=NSTwipsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.y,t2p); - ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.width=NSTwipsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.width,t2p); - ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.height=NSTwipsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.height,t2p); + ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.x=NSAppUnitsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.x, p2a); + ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.y=NSAppUnitsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.y, p2a); + ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.width=NSAppUnitsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.width, p2a); + ((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.height=NSAppUnitsToIntPixels(((nsCompositionEvent*)aEvent)->theReply.mCursorPosition.height, p2a); } if(aEvent->message==NS_QUERYCARETRECT) { - ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.x=NSTwipsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.x,t2p); - ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.y=NSTwipsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.y,t2p); - ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.width=NSTwipsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.width,t2p); - ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.height=NSTwipsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.height,t2p); + ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.x=NSAppUnitsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.x, p2a); + ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.y=NSAppUnitsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.y, p2a); + ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.width=NSAppUnitsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.width, p2a); + ((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.height=NSAppUnitsToIntPixels(((nsQueryCaretRectEvent*)aEvent)->theReply.mCaretRect.height, p2a); } I think these should all be converting to devpixels. You need to rev the UUID in nsIPrintOptions. The nsStyleBorder and nsStyleOutline changes are just to ensure that the rounding of border and outline widths that occur in the style system are rounding to device pixels, not CSS pixels, which makes sense to me. Fix those and I think we're good to land. I just need to compare my patch against yours to see if I slipped in any fixes that you need.
+ nscoord newX = mHandScrollStartScrollX + NSIntPixelsToAppUnts(deltaX, p2a); + nscoord newY = mHandScrollStartScrollY + NSIntPixelsToAppUnts(deltaY, p2a); typo "Unts" In nsThebesContext::SetDPI, OSVal is unused on Mac. You might want to just declare it where it's assigned for GTK2 and Windows. That's all I have. Good to go!
Eli, what times would you be available to land this in the window vlad gave in comment 190? (And, FWIW, the structure that survives reload in comment 189 is probably the device context. I've been seeing problems with dynamic DPI changes (which result from screen size changes when I dock/undock) ever since the switch to cairo. But that belongs in other bugs.)
BTW, a followup idea for this: We should expose the current CSS-to-device-pixel ratio in the useragent string we sent to HTTP servers. This would allow them to send an image of the appropriate resolution.
(In reply to comment #194) >. Yes, I messed up by not using sizeAppUnits everywhere. However, it's definitely supposed to be DevPixelsToAppUnits. Otherwise, the canvas doesn't get scaled up properly.
This seems to have broken balsa (gcc 3.4) and SunOS' nba & putt builds (SeaMonkey). And also TB win tbox & linux crazyhorse builds.
(In reply to comment #199) > This seems to have broken balsa (gcc 3.4) and SunOS' nba & putt builds > (SeaMonkey). And also TB win tbox & linux crazyhorse builds. See bug 369588.
I think that this bug is a cause of bug 369618.
(In reply to comment #132) > One issue I'm seeing with the patch is that there is a border missing for the > inactive tabs in Firefox, see: > All tabs are shifted down. This works as a workaround: .tabbrowser-tab { margin-bottom: 1px !important; }
After this patch is landed, the gtk2 toolkit doesn't build any more on Solaris 10. Is there any plan to fix this problem?
(In reply to comment #203) > After this patch is landed, the gtk2 toolkit doesn't build any more on Solaris > 10. Is there any plan to fix this problem? With this patch, we only support cairo-gtk2. That is completely intentional. (We've been planning to break non-Thebes GFX for a while now; this patch is the first to break it.) If there's something that breaking specifically on Solaris, file a bug; however, that seems unlikely.
So the point is that the configuration of those tinderboxes needs to be fixed to a supported one. see also bug 369588 for a similar issue.
Attachment #252688 - Attachment is obsolete: true
Attachment #254214 - Attachment is obsolete: true
And marking fixed; yay :)
Status: ASSIGNED → RESOLVED
Last Resolved: 12 years ago
Resolution: --- → FIXED
Could this have caused bug 370631?
Target Milestone: --- → mozilla1.9alpha3
I conclude from the testing I did with the nightlies that this bug caused the regression in bug 382458
The way you are rounding floats to integers is not correct and still produces strange cumulative effects (even if there's no more any 1-device-pixel roundoff error). Apparently you have fixed it using everywhere the function round(x) = ceil(x + 0.5) But in my opinion, the standard IEEE rounding is far better (and it is available in hardware): it rounds to the nearest EVEN integer when the diretion of rounding is not clearly decidable (equal absolute difference between the rounded result and the unrounded parameter). This becomes far more important now that SVG starts being integrated in HTML5, and zooming to arbitrary scales though a transform funtion is part of its design: the effect of rounding will be visible as well when zooming scaled bitmap images (see the discussions about this within graphics 3D engines and how mipmaps are used): ONLY the standard IEEE round-to-nearest-even mode (which is also normally the default rounding mode for floatting-point computings in strict mode) produces the correct isotropic rounding. In other words, don't use the ceil() function (which is also costly in terms of hardware acceleration in CPUs or GPUs), just use the IEEE rounding mode. Or at least to round on the nearest even half-pixel still using this IEEE rounding mode to then round this number of half-pixels to an integer number pixels, if you still want some compatibility with the current solution. In other words use: round(x) = IEEEroundtoeven(IEEEroundtoeven(x * 2.0) / 2.0) (Note that the IEEE rounding mode produces very desirable isotropic effects when handling negative coordinates, which will occur because of zooming and various transforms, as it gives a perfect symetry around 0, producing perfect symetries around any other arbitrary center, when objects are later translated. Note also that works are already being done to integrate 3D objects and projections to 2D within SVG, using 4D transform matrixes, exactly like what is done within GPUs.) Examples: x | x * 2.0 | IEEE round to | final IEEE | | nearest even | round(/2.0) ----------+-----------+---------------+------------ 0.00...0 | 0.00...0 | 0.0 | 0.0 -0.00...0 | -0.00...0 | -0.0 | -0.0 | | | 0.00...1 | 0.00...1 | 0.0 | 0.0 -0.00...1 | -0.00...1 | -0.0 | -0.0 ----------+-----------+ | 0.24...9 | 0.49...9 | 0.0 | 0.0 -0.24...9 | 0.49...9 | -0.0 | -0.0 | | | 0.25...0 | 0.50...0 | 0.0 | 0.0 -0.25...0 | -0.50...0 | -0.0 | -0.0 | +---------------+ 0.25...1 | 0.50...1 | 1.0 | 0.0 -0.25...1 | -0.50...1 | -1.0 | -0.0 ----------+-----------+ | 0.49...9 | 0.99...9 | 1.0 | 0.0 -0.49...9 | -0.99...9 | -1.0 | -0.0 | | | 0.50...0 | 1.00...0 | 1.0 | 0.0 -0.50...0 | -1.00...0 | -1.0 | -0.0 | | | 0.50...1 | 1.00...1 | 1.0 | 0.0 -0.50...1 | -1.00...1 | -1.0 | -0.0 ----------+-----------+ | 0.74...9 | 1.49...9 | 1.0 | 0.0 -0.74...9 | -1.49...9 | -1.0 | -0.0 | +---------------+----------- 0.75...0 | 1.50...0 | 2.0 | 1.0 -0.75...0 | -1.50...0 | -2.0 | -1.0 | | | 0.75...1 | 1.50...1 | 2.0 | 1.0 -0.75...1 | -1.50...1 | -2.0 | -1.0 ----------+-----------+ | 0.99...9 | 1.99...9 | 2.0 | 1.0 -0.99...9 | -1.99...9 | -2.0 | -1.0 | | | 1.00...0 | 2.00...0 | 2.0 | 1.0 -1.00...0 | -2.00...0 | -2.0 | -1.0 | | | 1.00...1 | 2.00...1 | 2.0 | 1.0 -1.00...1 | -2.00...1 | -2.0 | -1.0 ----------+-----------+ | 1.24...9 | 2.49...9 | 2.0 | 1.0 -1.24...9 | -2.49...9 | -2.0 | -1.0 | | | 1.25...0 | 2.50...0 | 2.0 | 1.0 -1.25...0 | -2.50...0 | -2.0 | -1.0 | +---------------+ 1.25...1 | 2.50...1 | 3.0 | 1.0 -1.25...1 | -2.50...1 | -3.0 | -1.0 ----------+-----------+ | 1.49...9 | 2.99...9 | 3.0 | 1.0 -1.49...9 | -2.99...9 | -3.0 | -1.0 | | | 1.50...0 | 3.00...0 | 3.0 | 1.0 -1.50...0 | -3.00...0 | -3.0 | -1.0 | | | 1.50...1 | 3.00...1 | 3.0 | 1.0 -1.50...1 | -3.00...1 | -3.0 | -1.0 ----------+-----------+ | 1.74...9 | 3.49...9 | 3.0 | 1.0 -1.74...9 | -3.49...9 | -3.0 | -1.0 | +---------------+----------- 1.75...0 | 3.50...0 | 4.0 | 2.0 -1.75...0 | -3.50...0 | -4.0 | -2.0 | | | 1.75...1 | 3.50...1 | 4.0 | 2.0 -1.75...1 | -3.50...1 | -4.0 | -2.0 ----------+-----------+ | 1.99...9 | 3.99...9 | 4.0 | 2.0 -1.99...9 | -3.99...9 | -4.0 | -2.0 | | | 2.00...0 | 4.00...0 | 4.0 | 2.0 -2.00...0 | -4.00...0 | -4.0 | -2.0 | | | 2.00...1 | 4.00...1 | 4.0 | 2.0 -2.00...1 | -4.00...1 | -4.0 | -2.0 ----------+-----------+ | 2.24...9 | 4.49...9 | 4.0 | 2.0 -2.24...9 | -4.49...9 | -4.0 | -2.0 | | | 2.25...0 | 4.50...0 | 4.0 | 2.0 -2.25...0 | -4.50...0 | -4.0 | -2.0 | +---------------+ 2.25...1 | 4.50...1 | 5.0 | 2.0 -2.25...1 | -4.50...1 | -5.0 | -2.0 ----------+-----------+ | 2.49...9 | 4.99...9 | 5.0 | 2.0 -2.49...9 | -4.99...9 | -5.0 | -2.0 | | | 2.50...0 | 5.00...0 | 5.0 | 2.0 -2.50...0 | -5.00...0 | -5.0 | -2.0 | | | 2.50...1 | 5.00...1 | 5.0 | 2.0 -2.50...1 | -5.00...1 | -5.0 | -2.0 ----------+-----------+ | 2.74...9 | 5.49...9 | 5.0 | 2.0 -2.74...9 | -5.49...9 | -5.0 | -2.0 | +---------------+----------- 2.75...0 | 5.50...0 | 6.0 | 3.0 -2.75...0 | -5.50...0 | -6.0 | -3.0 | | | 2.75...1 | 5.50...1 | 6.0 | 3.0 -2.75...1 | -5.50...1 | -6.0 | -3.0 ----------+-----------+ | 2.99...9 | 5.99...9 | 6.0 | 3.0 -2.99...9 | -5.99...9 | -6.0 | -3.0 | | | 3.00...0 | 6.00...0 | 6.0 | 3.0 -3.00...0 | -6.00...0 | -6.0 | -3.0 Note that even in this case, the second IEEE rounding to nearest even is stable: it returns exactly the same integers as if a single IEEE rounding to nearest even was used, without first rounding to integer numbers of half-pixels. In other words, a single rounding function can be used to produce fully isotropic images, independantly of their visual translation: round(x) = IEEEroundtoeven(x) And thanks to CPU- and GPU-builders, it is already hardware-accelerated; so builtin intrinsic C/C++ functions that map to these hardware instructions, so this should perform really fast ! (as long as the CPU or GPU is instructed to use this rounding mode).
Also note that the ceil() and floor() functions do not benefit as much of the hardware acceleration (in addition to being anisotropic, a undesirable effect for producing images). This is because (conceptually) the ceil or floor functions are defined as: floor(x) = (x < 0) ? -positiveCeil(-x) : positiveFloor(x); ceil(x) = (x < 0) ? -positiveFloor(-x) : positiveCell(x); And hardware does not like performing test signs, so it is often not implemented (on the opposite, the IEEE round-to-nearest-even mode only requires testing a single bit in the mantissa, depending on the integer value of the base-2 exponent, independantly of the sign of the number to round). There will then be no intrinsic builtin for these functions, and the C/C++ compiled code will generate conditional-branches, which defeats the branch prediction and slows down the performance. Note that this IEEE rounding mode is the STANDARD DEFAULT mode for ANSI/ISO C/C++ when computing with all floatting-point types (float, double, long double, as well as the newer fixed decimal types) and there are excellent reasons for this rounding mode being the default (notably for numbers with fixed precision, which may be implemented using integer arithmetics with the ALU only and not the FPU, or using parallel vector extensions heavily used in 3D kernel renderers).
Also, don't use the division by 2.0 (as described above) but a multiplication (this will help C/C++ optimizers if they don't detect it). inline long double _Gfx_Fastest_IsotropicRoundLD(long double x) { return IEEEroundtoeven(x); // look for actual builtin intrinsic if available } inline long double _Gfx_FastCompatible_IsotropicRoundLD(long double x) { return _Gfx_Fastest_IsotropicRoundLD( _Gfx_Fastest_IsotropicRoundLD(x * 2.0) * 0.5); } inline double _Gfx_Fastest_IsotropicRoundD(double x) { return (double)_Gfx_Fastest_IsotropicRoundLD((long double)x); } inline double Gfx_FastCompatible_IsotropicRoundD(double x) { return (double)_Gfx_FastCompatible_IsotropicRoundLD((long double)x); } inline float _Gfx_Fastest_IsotropicRoundF(float x) { return (float)_Gfx_Fastest_IsotropicRoundIEEEroundtoevenLD((long double)x); } inline float _Gfx_FastCompatible_IsotropicRoundF(float x) { return (float)_Gfx_FastCompatible_IsotropicRoundLD((long double)x); } Check the internal macros or builtin functions that define the IEEEroundtoeven() function as it is just conceptual here.
> as it gives a perfect symetry around 0 This is a drawback, not a benefit. The key to using floor(x + 0.5) is that this gives consistent rounding direction and avoids "missing pixel" seams. If we were to actually use round-to-nearest-ties-to-even, then markup like this: <div> <div style="height: 0.5px"></div> <div style="height: 0.5px"></div> <div style="height: 0.5px"></div> <div style="height: 0.5px"></div> </div> would result in the four inner divs spanning the vertical ranges [0, 0), [0,1), [1,2), [2, 2). That is, things specified to be the same size in the source would NOT be the same size in the rendering. You may want to also read
If you have only 2 device pixels, you can only create 2 vertical spannings (you can't have 4). The IEEE rounding mode will still result in exactly 2 non overlapping pixels that cover the full range correctly. It's simply IMPOSSIBLE to have 4 equal vertical ranges on the device. And perfect symetry around 0 is highly desirable in almost all cases (including the fact that symetry will behave correctly for object borders that should be as precisely as possible position on the pixel boundaries, without loosing these borders). If you have subpixel precision at the device level, treat them with the precision of these subpixels, but you should still still use the IEEE round-to-nearest-even-tie mode (but subpixel precision does not even exist for now on the vertical axis on display, as subpixel precision is typically used only for flat LCD/LED panels; but it exists on printing devices like bubble jet printers) With floor (x + 0.5), not only this is costly in terms of CPU/FPU or GPU instructions, but you get undesirable border effects when computing coordinates in complex layouts with floatting point values. And what you finally get ONLY when source coordinates where EXACTLY computed as (0.0, 0.5, 1.0, 1.5) will be the integer ranges: [0, 1), [1, 1), [1, 2), [2, 2) So you'll still have only 2 non-zero-width ranges (the 1st and the 3rd). So here also the same sizes in the source will result in distinct sizes in the rendering. Your argument is not valid. But any very tiny difference in the source floatting-point coordinates will cause the 2 visible ranges to be selected randomly within these 4 ranges (depending on the precision of the coordinates), producing anisotropic images. That's something that the IEEE rounding mode to nearest even tie completely avoids in almost all cases. This IEEE rounding mode is then much more consistant, as it will correctly round all the lowest differences that exist when the computed floatting-point coordinates have only a few ups of differences that are mich smaller than the device precision.
> If you have only 2 device pixels What made you think that's the case? There are various things that need to be rounded to _CSS_ pixels, which can be multiple device pixels in size. > And perfect symetry around 0 is highly desirable in almost all cases Except the case when there is a fundamental anisotropy to the content. Like say a web page (which has a clearly defined up and down for the most part; SVG can be an exception). Please do read roc's blog post. He explains what the issues are much more clearly than my simple (and somewhat flawed) example does.
Note that your link to the mozilla article just discusses the case of rounding towards or away from zero. It DOES NOT disacuss about the rounding modes towards even ties, which is definitely NOT harmful And you should read articles about 3D rendering, about why it is much better to use it rather than floor(x+0.5) towards minus infinity of equal ties, which produces also other effects like incorrect colors when zooming out thin black and white patterns at 50%: you will get a full-white or full-black fill, and this will constantly flicker between those two when slowly translating such patterns with subpixel floatting-point positions With the IEEE rounding mode set to nearest even tie, you'll get a consistant pattern of black and white at ALL positions, when not using smoothed rendering, or you'll be able to compute consistant gray levels with a smoothed rendering, except within a few ups at a precision MUCH smaller than the device pixel precision, i.e. a level of precision where the difference of smoothing greys will be almost invisible.] The typical case occurs when computing floatting points coordinaes like: double w3 = 3/7; double w1 = 1/7; double w2 = 2/7; return w3 - w1 - w2; which may return a small negative number (such as -1e-12), due to limited precisions : the difference will come from the few ups of precision used to represent the three independantly stored values w1, w2, w3. If you round this result with floor(x+0.5), you'll see that it rounds to -1 (instead of the expected exact 0 that you'll get with IEEE rounding mode towards even tie) and things that should be completely invisible when rendered will suddenly become visible.
And in the blog, you'll see the important comment made by Dijskra. He is right when saying that the IEEE round to nearest-even-tie is FAR better than floor(x+0.5) or ceil(x+0.5) suggested in the blog article to replace the rounding modes toward zero used in the C/C++ basic typecasts of float/double to integer types. Reconsider your position, and create another article stating that floor(x+0.5) and ceil(x+0.5) is ALSO considered harmful for graphics rendering.
Note also that the IEEE 754 round-to-nearest even tie is also mandatory in font renderers (notably those performing correct hinting), this is not just for SVG, and the SVG integration within HTML will continue to grow in HTML5. There's also no such concept as "CSS pixels" which are also arbitrary floatting point measurement units that have nothing in common with the device precision, where rescaling will occur anyway using all sorts of scaling factors that are definitely not part of the CSS specifications (not even in CSS 3).
> And in the blog, you'll see the important comment made by Dijskra. Are you sure you're not misattributing Peter Moulder's comment? > for graphics rendering Where does graphics rendering come in? The discussion in this bug, for the most part, and in the blog post, is about _text_ and border rendering and rendering of websites, not graphics rendering. Graphics rendering is a different ballgame, quite possibly. Certainly different constraints apply.
> There's also no such concept as "CSS pixels" There is, in fact. CSS defines a unit called "pixel" which is only very loosely related to device pixels. However some things in practice (border widths come to mind) have to be rounded to integer numbers of CSS pixels.
Rounding towards even is the same as rounding towards zero for half the midpoint values and rounding away from zero for the other half. So it has exactly the issues roc's blog post describes. > The typical case occurs when computing floatting points coordinaes like: Note that we currently do not in fact use floating point coordinates internally > which may return a small negative number (such as -1e-12), ... > If you round this result with floor(x+0.5), you'll see that it rounds to -1 That statement is false. It will round to 0.
And sorry for my example, I forgot to substract 1/4 to the computed expression, or to divide the result of the computed differences by 2 (this is effectively where the undesirable and unstable cutoffs occur).. You should read the specification of JPEG 2000, PNG, MPEG, H.264, and Ogg Video formats that also have the same constraints for the prefered rounding mode (remember that HTML 5 will integrate these media types very tightly). The concept of "CSS pixels" mapping to a fixed integer number of device pixels is simply flawed. "CSS pixels" are floatting point measures that JUST need to have AT LEAST a IEEE-754 64-bit precision. A higher precision should be accepted, but ECMAScript still does not support Number values with higher precisions, even if ECMAScript interoperates through OMG IDL interfaces that define them precisely, and even if there's hardware support for them (for example "long double" 80-bit precision in C/C++ and on x86 FPUs, or high-end GPUs).
> Note that we currently do not in fact use floating point coordinates internally But pages are designed to use them, as they are exposed through DOM and ECMAScript that already use flotting points. That's where the precision ups will occur when computing coordinates, and this is the only world where "CSS pixels" are measured, independantly of the target rendering device. The DOM, or scripts within pages, should NEVER have to be exposed with values modified and remapped to integers by the renderer, they should just be created using the standard double precision as specified in ECMAScript, and they will produce consistant results on all browsers or renderers or target surface. In other words, they should retain their full double precision.
(In reply to comment #223) >. I assure you, SpiderMonkey's number representation system (and calculation methods and their implementation) does not overlap with Gecko's layout-dimension representation in any way. :-)
> this is the only world where "CSS pixels" are measured, independantly of the > target rendering device. Why are we imposing this arbitrary constraint? The target rendering device is very important. > should NEVER have to be exposed with values modified and remapped to integers > by the renderer Good luck. Every single renderer does this, for various good reasons (mostly to do with the fact that scripts tend to try to impose self-contradictory constraints; I suggest you google around for the long discussions that happened about this). > they should just be created using the standard double precision as specified > in ECMAScript We've considered using doubles for internal coordinates. There are issues that need to be worked out with rounding artifacts (which round-to-even does NOT address) and there is a noticeable memory/performance cost last it was measured. > and they will produce consistant results on all browsers No, they won't. It's much harder to get consistent results with floating point, in fact, because suddenly the order of arithmetic operations in the layout algorithms matters.
One other note. The most common way in which coordinates enter the system is via stylesheets and CSS does not use double-precision floating-point. It uses (due to limitations like finite memory) arbitrary rational coordinates represented via decimal expansions. In practice, renderers round these to some sort of integer unit instead of actually implementing infinite-precision rational arithmetic; we do it to 1/60 of a CSS pixel, while others round to integer CSS pixels or whatnot.
Isn't CSS directly exposed via OMG IDL for interoperability with scripting languages ? CSS defines two levels: - a syntaxic source language on the surface which uses finite decimal precision. That precision is effectively limited by source size. This syntaxic language however is not mandatory, this is just used for interchange of source code, rather than of binary objects. - a IDL interoperability level which is its runtime object representation: the CSS language will be anyway always complied to this object form. Scripts do not have to serialize or generate the language level, given that this can be done dynamically by serialization (basicall the generic Object::toString() method). CSS should interact cleanly with ECMAScript, without any losses or differences of precision, so it should allow correct rounding with IEEE-754 64-bit doubles, at least (but other scripting engines may support higher precisions), according to ECMAScript specs which define the precision of builtin mathematical operators and functions in the Math object with extreme care about the acceptable number of ups of error. Webpages (and other medias like SVG, PNG, JPEG2000, MPEG...) will be designed according to these precision requirements (even if some media types also use other numeric representations with lower precision), and SVG and some media types are currently being fully integrated within HTML 5. I can't imagine that HTML 5 will not require at least the standard IEEE-754 64-bit precision as documented in ECMAScript. Now you can do whatever you want in the renderer when mapping coordinates to device units, but this should ABSOLUTELY NOT affect what is perceived in scripts or through the IDL interface to the style member of HTMLElement objects. So the renderer MUST NOT modify these IDL-reflected values, even if it convertes them internally to device-specific integers after performing some device-specific transforms, notably rescaling to the device resolution, and pixel-grid alignments (and this is only after this coordinates transform has been done that correct rounding to integers of device units should occur).
> Isn't CSS directly exposed via OMG IDL No. > CSS defines two levels No, CSS itself, as defined by the CSS working group, only defines the first of the two things you list. A second group of people, who didn't actually understand CSS very well, then went and grafted an IDL API on top of it. And since they were basically building on top of the existing CSS syntax, IDL access to CSS is on a string level only; everything crossing that boundary from JS is first serialized to a string and then that string is parsed following the syntactic language rules. Now there are plans to create new CSS-related IDL APIs, but the above describes the state of things today. I seriously suggest looking into the actual APIs the CSSOM exposes instead of just assuming what they look like. I also suggest generally familiarizing yourself with things before making completely incorrect claims...
The second workgroup you're speaking about is also within the W3C. The CSS API level is described in SAC, and it is a FULL part of the W3C Style Activity, with bindings being defined for Java, C (extensible to C++), and Perl. and yes CSSOM is now part of DOM Level 2 (it should be mandatory for HTML 5), whose SAC describes concrete interfaces for a few languages. CSSOM is alread at the stage of Candidate Recommandation. You can't say that this was defined by a random separate group, given that the W3C is already adapting all its past specifications to define their interoperation in terms of OMG IDL Well, in terms of IDL, SAC (v.1.3) provides effectively the LexicalUnit object that has several properties: a string representation of the value, an extractable unit (from a small enumerated list), and two numeric value extractors: as an integer, or a a 32-bit float (it's strange that they did not include a 64-bit double, similar to what is used in ECMASCript Number, but ECMAScript will still be able to interact with the string serialization of the LexicalUnit object). My opinion is that a future version of SAC should extend this interface to use a more direct LexicalUnit::getDoubleValue() which would make it significantly faster for ECMAScript interoperability, and an even faster as well for processing in Java, C, C++, .Net, Perl and other engines, given that it already defines and supports a limited getFloat() value extractor and the lexicalUnitType() value extractor, in addition to the getStringValue() serialization plus two incrementing/decrementing operations (mostly for list counters). And anyway, in CSSOM, the CSSPrimitiveValue object also exposes the enumerated unitType value property (readonly) and the same floatValue which can be set or retrieved, or the combined stringValue property. That is a damn'ed slow interface if this is just the one exposed, and in fact if this is the only supported representation, it has a possibly infinite decimal precision that scripts will never be able to compute. Note also that there's not even any public constructor defined in the interface for LexicalUnit. The only way it can be instanciated is through the InputSource object constructors taking a string or characterStream in parameter, in order to compile it to a stylesheet containing lists of Selectors. It's really inefficient in practice, and actual implementations will necessarily add (and use) concrete constructors of objects implementing the LexicalUnit interface. In CSSOM at least there's a concrete isolated LexicalUnit object but CSSOM is just one concrete implementation of SAC and browsers with ECMAScript bindings for DOM Level 2 will implement SAC using a class implementing the LexicalUnit interface with a concrete constructor taking a Number and a unit type, even if they operate with other engines through the basic (readonly) interfaced accessors. Oh... well... CSS should really define a standard minimum supported precision for the lexical units. Anyway CSS 2.1 clearly states that computed valus are expected to be rounded when converted to device pixels, but it does not describe the operation, given that it does not fix their geometry (they are not necessarily in square grids or with a constant size, or arranged in horizontal rows and vertical columns. But if you implement this coordinates mapping, the deault IEEE 754 rounding mode should be used to avoid border effects.
SAC has nothing to do with the CSS working group, and I believe it's a dead project. (I'm a member of the CSS working group.) Also, this discussion doesn't belong on this bug, which was fixed a long time ago.
Someone moved my initial comment there, I actually posted in an open bug that was unified with this one (and the last post before mine refers to a bug redirected here. And SAC, even if it looks "dead", is still what CSSOM implements concretely. And both are supported by the W3C, which has not removed either of them. It remains valid as a basic interface and usable even when interoperating with the ongoing CSS 3 proposals. You can easily derive a SAC 3.0 interface from the SAC 1.3 interface, to add additional services for CSS 3.0 (and I wish that it defines a doubleValue property, implemented in CSSOM 3.0, for use in CSS 2.1 or 3.0 as well, and directly usable with ECMAScript Number values without using the slow interface with strings only). But even without this change, the dual exposition of the numeric value as a 32-bit int or as a 32-bit float means that it can represent at least all coordinates with 24-bit integers. This clearly limits the number of significant decimal digits supported in the string interface to 7 at most plus a sign. For higher precisions, only the slow string interface is interoperable); so yes, "CSS pixels" in most documents can't be more precise than 0.01px, if documents cannot be wider or higher than 5 digits (less than 100000px) if using the limited float precision (such limited precision means that zooming in the document at moderate scale 50:1 or higher will very easily exhibit the roundoff errors). And because of this insufficient precision of 32-bit floats, the effect of the rounding mode to choose when computing images or layouts is EVEN MORE CRITICALLY IMPORTANT !
And effectively this bug is not CLOSED given that its full resolution still depends on open bugs, and it is also blocking other open bugs. In its state, it is just temporarily FIXED, within some limits that do not fully cover all the open bugs on which it depends.
verdy_p, I can only make several recommendations: 1) I'm happy to discuss the politics and realities of W3C stuff (which you seem to be _very_ confused about) with you in the right forum (private mail is fine). This bug is not the right forum. 2) I'm not sure what makes you think you know better than the person who filed a bug whether that bug is fixed. | https://bugzilla.mozilla.org/show_bug.cgi?id=177805 | CC-MAIN-2019-18 | refinedweb | 21,576 | 63.39 |
I just installed the Mac OSX Alpha and wanted to get started trying to write a plugin but i'm getting this error in my console whenever I save anything after importing sublimeplugin"ImportError: No module named sublimeplugin"
Here are the contents of my file I'm doing just to test things out:
import sublime, sublimeplugin
print "plugin loaded"
class Test1Command(sublimeplugin.TextCommand):
def run(self, view, args):
print "command run"
Any ideas on what I may be doing wrong or missing?
I Sublime 2, the new name's "sublime_plugin". That should work.
that works perfectly now, thanks!
one more question, did
view.runCommand('test1')
change? that's the way the API docs specify to run it but i'm getting an AttributeError.
The api docs cover good old Sublime 1.x... Sublime 2 introduces a few changes to how the api works and looks... Most prominently, Sublime 2 api conforms to PEP8, which means camelCase becomes camel_case. However, your best bet is to dir() whatever object you're interested in in the python console (CTRL + ~) and try to find the new names. The old docs are still valid to a great extent, but new ones are badly needed.
Ah, I see now. Well last question I promise =)
Is there a list around somewhere of good plugins that can be run on Sublime 2 that I could use as an example to look at?
Thanks again for all the help. Much appreciated.
Sublime Text 2 has only recently been released, and the API is still a little incomplete, so there aren't a lot of plugins for it yet.If you check out sublimetext.info/, you can see a bunch of plugins for ST1, though, and most of the calls are identical in the way that Guillermo summarized.Feel free to keep asking questions here if you have them, though.
For example Sublime Text 2 plugins, you can have a look at the ones that ship with the editor, in the Packages/Default directory (accessible via Preferences/Browse Packages)
exactly what i needed. thanks!
Almost three years later now, and the old (Sublime 1) documents are still prominently featured on sublimetext.com/docs/plugin-examples.
I'm fairly new to Sublime, though I've been programming for decades. I have ended up wasting a couple days worth of my time, trying to figure out why I could not make any extensions work, even these simplest examples.
There is better Sublime 2 plugin documentation at net.tutsplus.com/tutorials/pytho ... -2-plugin/
Is there --any-- way that we could get the page to mention or mention this present thread or mention that those examples are applicable only to Sublime 1 or -- something -- to give the poor saps who get lost trying to follow those plugin-examples a bit of a clue?
(I realize that you are almost certainly not the one to ask this question of ... but I don't know who is ... perhaps I'll get lucky and someone will offer a clue as to who might be.)
Here are the two fixes I found were needed in.
Wrong sublimeplugin.TextCommandRight sublime_plugin.TextCommand
Wrong view.runCommand('hello')Right view.run_command('hello')
With these two fixes, the HelloCommand runs as 'hello' in both sublime 2 and 3, using the command:
view.runCommand('hello')
in the console (accessible via Ctrl+~).
Someone must have edit access to that page
The API documentation at sublimetext.com/docs/3/api_reference.html.
A naive user (well, at least this naive user) might think that sublimetext.com/docs/api-reference (based both on what the URL and the page state) is the (one and only) API for Sublime.
I would suggest that instead sublimetext.com/docs/api-reference, a short page, with essentially just three links, to
sublimetext.com/docs/1/api-referencesublimetext.com/docs/2/api-referencesublimetext.com/docs/3/api-reference
Once again, however, I have no clue to whom to properly address these comments.
It is aggravating that the examples are not even close to working with ST v3. Here is the working version of the "Events" example. Hopefully this will save someone from experiencing my two hours of frustration:
import sublime, sublime_plugin
class EventDump(sublime_plugin.EventListener):
def on_load(self, view):
print (view.file_name(), "just got loaded")
def on_pre_save(self, view):
print (view.file_name(), "is about to be saved")
def on_post_save(self, view):
print (view.file_name(), "just got saved")
def on_new(self, view):
print ("new file")
def on_modified(self, view):
print (view.file_name(), "modified")
def on_activated(self, view):
print (view.file_name(), "is now the active view")
def on_close(self, view):
print (view.file_name(), "is no more")
def on_clone(self, view):
print (view.file_name(), "just got cloned") | https://forum.sublimetext.com/t/importerror-no-module-named-sublimeplugin/1128 | CC-MAIN-2017-22 | refinedweb | 782 | 65.52 |
Azure Storage
Updated: July 16, 2015. For an overview of Azure Storage, see Introduction to Microsoft Azure Storage.
An Azure storage account is a secure account that gives you access to services in Azure Storage. Your storage account provides the unique namespace for your storage resources. There are two types of storage accounts:
- A standard storage account includes Blob, Table, Queue, and File storage. To create a standard storage account, see Create, manage, or delete a storage account.
- A premium storage account currently supports Azure Virtual Machine disks only. Azure Premium Storage is available by request via the Azure Preview page.
For an in-depth overview of Azure Premium Storage, see Premium Storage: High-Performance Storage for Azure Virtual Machine Workloads.
Before you can create a storage account, you must have an Azure subscription, which is a plan that gives you access to a variety of Azure services. You can create up to 100 uniquely named storage accounts with a single subscription. See Storage Pricing Details for information on volume pricing.
You can get started with Azure with a free trial. Once you decide to purchase a plan, you can choose from a variety of purchase options. If you’re an MSDN subscriber, you get free monthly credits that you can use with Azure services, including Azure Storage.
The Azure Storage services are Blob storage, Table Storage, Queue Storage, and File Storage:
- Blob Storage stores file data..
- File storage (Preview) offers shared storage for legacy applications using the standard SMB 2.1 protocol. Azure virtual machines and cloud services can share file data across application components via mounted shares, and on-premise applications can access file data in a share via the File service REST API. File storage is available by request via the Azure Preview page.
See Also | https://msdn.microsoft.com/en-us/library/azure/gg433040 | CC-MAIN-2015-48 | refinedweb | 299 | 64.2 |
Only if creating the world, you will be god.
I want to learn from this site.
whats dis actually means.........
Post your Comment
Actions in Struts
Actions in Struts Hello Sir,
Thanks for the solutions you have sent me.
i wanted examples on Struts DispatchAction,Forword Action ,Struts lookupDispatchAction,Struts mappingDispatchAction,Struts DynaActionform.please
Writing Actions
Writing Actions
The Action is responsible for controlling of data flow within an application.
You can make any java class as action. But struts have some built in interface
and class for making action easily.
To make an Action class
Actions -
Two or more actions in the same form
Two or more actions in the same form Can I have two or more actions in the same form
Struts2 Actions
Struts2 Actions
When... to the handler class
is defined by the Action interface.
Struts2 Action interface
All actions may Built-In Actions
Struts Built-In Actions
... actions shipped with Struts APIs. These
built-in utility actions provide different...;
Actions
Struts 2 Actions
in many different ways. Here are some of
the example.
In Struts2 actions can...
Struts 2 Actions
In this section we will learn about Struts 2 Actions, which is a fundamental
concept in most of the web
JSP Actions : Java Glossary
JSP Actions : Java Glossary
JSP actions are XML tags that forces the server to
directly use... Actions are made of a usually (XML-based) prefix of "jsp"
followed
Struts2 Actions
Struts2 Actions
When a client's request matches the action's
name, the framework uses the mapping from struts.xml file to process the request.
The mapping to an action
JSP Actions
JSP Actions
... JSP Action Tags in the JSP application.
What is JSP Actions?
Servlet container.... Programmers can use these functions
in JSP applications. The JSP Actions tags
Test Actions
Test Actions
An example of Testing a struts Action is given below using the junit. In this
example the execute method is being test.
HelloAction.java
package net.roseindia.action;
import com.opensymphony.xwork2.ActionSu
JSP Standard Actions 'jsp:setProperty' & 'jsp:getProperty'
JSP Standard Actions <jsp:setProperty> & <jsp:getProperty>
<jsp:setProperty>
It sets a property value or values in a Bean only if a new object was
instantiated, not if an existing one was found.
Syntax
Implementing Actions in Struts 2
Implementing Actions in Struts 2
Package com.opensymphony.xwork2 contains the many Action classes and
interface, if you want to make an action class for you.... Actions the contains the
execute() method. All the business logic is present
JSP Standard Actions
JSP Standard Actions
In this section, we will learn about JSP standard Action & their elements
with some examples.
JSP Standard action are predefined..., to generate a browser-specific code, etc. The
JSP standard actions affect the overall
Configuring Actions in Struts application
Configuring Actions in Struts Application
To Configure an action in struts application, at first write a simple Action
class such as
SimpleAction.java
package net.roseindia;
import com.opensymphony.xwork2.Action;
public class
STRUTS ACTION - AGGREGATING ACTIONS IN STRUTS
STRUTS ACTION - AGGREGATING ACTIONS IN STRUTS... using which you can aggregate a related set of actions into a single unified... actions.
DispatchAction: In this type
Action in Struts 2 Framework
Actions
Actions are the core basic unit of work in Struts2 framework. Each.... Actions are mostly associated with a HTTP request of User.
The action class... actions configuration matches
with a specific result that will be rendered
Use if statement with LOOP statement
Use if statement with LOOP statement
Conditional if statements in SQL are used to perform different
actions based on different conditions
JSP Interview Questions
;
Question: What do you understand by JSP
Actions?
Answer: JSP actions are XML tags that direct the server to use existing components or control the behavior of the JSP engine. JSP Actions consist of a typical
Mysql Switch CASE
Mysql Switch CASE
Mysql Switch Case perform one of several different actions based on several
different conditions.
Understand with Example
The Section of Tutorial
Mysql Else Condition
Mysql Else Condition
Mysql Else Condition evaluate the statements in database to perform different
actions based on different conditions.
Understand with Example
The Tutorial
TriggerListeners and JobListeners
TriggerListeners and JobListeners
To perform any action you create Listeners objects and
these actions are based on events occurring within the scheduler
Struts Forward Action Example
). The ForwardAction is one of the Built-in Actions
that is shipped with struts framework
engineerraymond July 24, 2012 at 11:28 AM
Only if creating the world, you will be god.
javamohammad minhazul hoque December 3, 2012 at 10:55 PM
I want to learn from this site.
javamohammad minhazul hoque December 3, 2012 at 10:56 PM
I want to learn from this site.
whats dis actually means.........
Post your Comment | http://roseindia.net/discussion/20979-Struts2-Actions.html | CC-MAIN-2015-40 | refinedweb | 792 | 56.66 |
# Various things in MetaPost
What is the best tool to use for drawing vector pictures? For me and probably for many others, the answer is pretty obvious: Illustrator, or, maybe, Inkscape. At least that's what I thought when I was asked to draw about eight hundred diagrams for a physics textbook. Nothing exceptional, just a bunch of black and white illustrations with spheres, springs, pulleys, lenses and so on. By that time it was already known that the book was going to be made in LaTeX and I was given a number of MS Word documents with embedded images. Some of them were scanned pictures from other books, some were pencil drawings. Picturing days and nights of inkscaping this stuff made me feel dizzy, so soon I found myself fantasizing about a more automated solution. For some reason [MetaPost](https://en.wikipedia.org/wiki/MetaPost) became the focus of these fantasies.
The main advantage of using MetaPost (or similar) solution is that every picture can be a sort of function of several variables. Such a picture can be quickly adjusted for any unforeseen circumstances of the layout, without disrupting important internal relationships of the illustration (something I was really concerned about), which is not easily achieved with more traditional tools. Also, recurring elements, all these spheres and springs, can be made more visually interesting than conventional tools would allow with the same time constraints.
I wanted to make pictures with some kind of hatching, not unlike what you encounter in old books.
First, I needed to be able to produce some curves of varying thickness. The main complication here is to construct a curve that follows the original curve at a varying distance. I used probably the most primitive working [method](https://math.stackexchange.com/questions/465782/control-points-of-offset-bezier-curve/467038#467038), which boils down to simply shifting the line segments connecting Bezier curve control points by a given distance, except this distance varies along the curve.
In most cases, it worked OK.
**Example code**From here on it is assumed that the library is [downloaded](https://github.com/jemmybutton/fiziko) and `input fiziko.mp;` is present in the MetaPost code. The fastest method is to use ConTeXt (then you don't need `beginfig` and `endfig`):
`\starttext
\startMPcode
input fiziko.mp;
% the code goes here
\stopMPcode
\stoptext`
or LuaLaTeX:
`\documentclass{article}
\usepackage{luamplib}
\begin{document}
\begin{mplibcode}
input fiziko.mp;
% the code goes here
\end{mplibcode}
\end{document}`
`beginfig(3);
path p, q; % MetaPost's syntax is reasonably readable, so I'll comment mostly on my stuff
p := (0,-1/4cm){dir(30)}..(5cm, 0)..{dir(30)}(10cm, 1/4cm);
q := offsetPath(p)(1cm*sin(offsetPathLength*pi)); % first argument is the path itself, second is a function of the position along this path (offsetPathLength changes from 0 to 1), which determines how far the outline is from the original line
draw p;
draw q dashed evenly;
endfig;`
Two outlines can be combined to make a contour line for a variable-thickness stroke.
**Example code**`beginfig(4);
path p, q[];
p := (0,-1/4cm){dir(30)}..(5cm, 0)..{dir(30)}(10cm, 1/4cm);
q1 := offsetPath(p)(1/2pt*(sin(offsetPathLength*pi)**2)); % outline on one side
q2 := offsetPath(p)(-1/2pt*(sin(offsetPathLength*pi)**2)); % and on the other
fill q1--reverse(q2)--cycle;
endfig;`
The line thickness should have some lower bound, otherwise, some lines are going to be too thin to be properly printed and this doesn't look great. One of the options (the one I chose) is to make the lines which are too thin dashed, so that the total amount of ink per unit length remains approximately the same as in the intended thinner line. In other words, instead of reducing the amount of ink on the sides of the line, the algorithm takes some ink from the line itself.
**Example code**`beginfig(5);
path p;
p := (0,-1/4cm){dir(30)}..(5cm, 0)..{dir(30)}(10cm, 1/4cm);
draw brush(p)(1pt*(sin(offsetPathLength*pi)**2)); % the arguments are the same as for the outline
endfig;`
Once you have working variable-thickness lines, you can draw spheres. A sphere can be depicted as a series of concentric circles with the line thicknesses varying according to the output of a function that calculates the lightness of a particular point on the sphere.
**Example code**`beginfig(6);
draw sphere.c(1.2cm);
draw sphere.c(2.4cm) shifted (2cm, 0);
endfig;`
Another convenient construction block is a “tube.” Roughly speaking it is a cylinder which you can bend. So long as the diameter is constant, it's pretty straightforward.
**Example code**`beginfig(7);
path p;
p := subpath (1,8) of fullcircle scaled 3cm;
draw tube.l(p)(1/2cm); % arguments are the path itself and the tube radius
endfig;`
If the diameter isn't constant, things become more complicated: the number of strokes should change according to the tube thickness in order to keep the amount of ink per unit area constant before taking the lights into account.
**Example code**`beginfig(8);
path p;
p := pathSubdivide(fullcircle, 2) scaled 3cm; % this thing splits every segment between the points of a path (here—fullcircle) into several parts (here—2)
draw tube.l(p)(1/2cm + 1/6cm*sin(offsetPathLength*10pi));
endfig;`
There are also tubes with transverse hatching. The problem of keeping the amount of ink constant turned out to be even trickier in this case, so ofttimes such tubes look a tad shaggy.
**Example code**`beginfig(9);
path p;
p := pathSubdivide(fullcircle, 2) scaled 3cm;
draw tube.t(p)(1/2cm + 1/6cm*sin(offsetPathLength*10pi));
endfig;`
Tubes can be used to construct a wide range of objects: from cones and cylinders to balusters.
**Example code**`beginfig(10);
draw tube.l ((0, 0) -- (0, 3cm))((1-offsetPathLength)*1cm) shifted (-3cm, 0); % a very simple cone
path p;
p := (-1/2cm, 0) {dir(175)} .. {dir(5)} (-1/2cm, 1/8cm) {dir(120)} .. (-2/5cm, 1/3cm) .. (-1/2cm, 3/4cm) {dir(90)} .. {dir(90)}(-1/4cm, 9/4cm){dir(175)} .. {dir(5)}(-1/4cm, 9/4cm + 1/5cm){dir(90)} .. (-2/5cm, 3cm); % baluster's envelope
p := pathSubdivide(p, 6);
draw p -- reverse(p xscaled -1) -- cycle;
tubeGenerateAlt(p, p xscaled -1, p rotated -90); % a more low-level stuff than tube.t, first two arguments are tube's sides and the third is the envelope. The envelope is basically a flattened out version of the outline, with line length along the x axis and the distance to line at the y. In the case of this baluster, it's simply its side rotated 90 degrees.
endfig;`
Some constructions which can be made from these primitives are included in the library. For example, the globe is basically a sphere.
**Example code**`beginfig(11);
draw globe(1cm, -15, 0) shifted (-6/2cm, 0); % radius, west longitude, north latitude, both decimal
draw globe(3/2cm, -30.280577, 59.939461);
draw globe(4/3cm, -140, -30) shifted (10/3cm, 0);
endfig;`
However, the hatching here is latitudinal and controlling line density is much more difficult than on the regular spheres with “concentric” hatching, so it's a different kind of sphere.
**Example code**`beginfig(12);
draw sphere.l(2cm, -60); % diameter and latitude
draw sphere.l(3cm, 45) shifted (3cm, 0);
endfig;`
A weight is a simple construction made from tubes of two types.
**Example code**`beginfig(13);
draw weight.s(1cm); % weight height
draw weight.s(2cm) shifted (2cm, 0);
endfig;`
There's also a tool to knot the tubes.
**Example code. For brevity's sake only one knot.**`beginfig(14);
path p;
p := (dir(90)*4/3cm) {dir(0)} .. tension 3/2 ..(dir(90 + 120)*4/3cm){dir(90 + 30)} .. tension 3/2 ..(dir(90 - 120)*4/3cm){dir(-90 - 30)} .. tension 3/2 .. cycle;
p := p scaled 6/5;
addStrandToKnot (primeOne) (p, 1/4cm, "l", "1, -1, 1"); % first, we add a strand of width 1/4cm going along the path p to the knot named primeOne. its intersections along the path go to layers "1, -1, 1" and the type of tube is going to be "l".
draw knotFromStrands (primeOne); % then the knot is drawn. you can add more than one strand.
endfig;`
Tubes in knots drop shadows on each other as they should. In theory, this feature can be used in other contexts, but since I had no plans to go deep into the third dimension, the user interface is somewhat lacking and shadows work properly only for some objects.
**Example code**`beginfig(15);
path shadowPath[];
boolean shadowsEnabled;
numeric numberOfShadows;
shadowsEnabled := true; % shadows need to be turned on
numberOfShadows := 1; % number of shadows should be specified
shadowPath0 := (-1cm, -2cm) -- (-1cm, 2cm) -- (-1cm +1/6cm, 2cm) -- (-1cm + 1/8cm, -2cm) -- cycle; % shadow-dropping object should be a closed path
shadowDepth0 := 4/3cm; % it's just this high above the object on which the shadow falls
shadowPath1 := shadowPath0 rotated -60;
shadowDepth1 := 4/3cm;
draw sphere.c(2.4cm); % shadows work ok only with sphere.c and tube.l with constant diameter
fill shadowPath0 withcolor white;
draw shadowPath0;
fill shadowPath1 withcolor white;
draw shadowPath1;
endfig;`
Certainly, you will need a wood texture (update: since the Russian version of this article was published, the first case of this library being used in a real project which I'm aware of [has occurred](https://tex.stackexchange.com/questions/423335/wood-pattern-with-metapost/448086?noredirect=1#comment1177927_448086), and it was the wood texture which came in handy, so this ended up not being a joke after all). How twigs and their growth affect the pattern of year rings is a topic for some serious study. The simplest working model I could come up with is as follows: year rings are parallel flat surfaces, distorted by growing twigs; thus the surface is modified by a series of not overly complex “twig functions” in different places and the surface's isolines are taken as the year ring pattern.
**Example code**`beginfig(16);
numeric w, b;
pair A, B, C, D, A', B', C', D';
w := 4cm;
b := 1/2cm;
A := (0, 0);
A' := (b, b);
B := (0, w);
B' := (b, w-b);
C := (w, w);
C' := (w-b, w-b);
D := (w, 0);
D' := (w-b, b);
draw woodenThing(A--A'--B'--B--cycle, 0); % a piece of wood inside the A--A'--B'--B--cycle path, with wood grain at 0 degrees
draw woodenThing(B--B'--C'--C--cycle, 90);
draw woodenThing(C--C'--D'--D--cycle, 0);
draw woodenThing(A--A'--D'--D--cycle, 90);
eyescale := 2/3cm; % scale for the eye
draw eye(150) shifted 1/2[A,C]; % the eye looks in 150 degree direction
endfig;`
The eye from the picture above opens wide or squints a bit and its pupil changes its size too. It may not make any practical sense, but mechanically similar eyes just look boring.
**Example code**`beginfig(17);
eyescale := 2/3cm; % 1/2cm by default
draw eye(0) shifted (0cm, 0);
draw eye(0) shifted (1cm, 0);
draw eye(0) shifted (2cm, 0);
draw eye(0) shifted (3cm, 0);
draw eye(0) shifted (4cm, 0);
endfig;`
Most of the time the illustrations weren't all that complex, but a more rigorous approach would require solving many of the problems in the textbook to illustrate them correctly. Say, L'Hôpital's pulley problem (it wasn't in that textbook, but anyway): on the rope with the length , suspended at the point  a pulley is hanging; it's hooked to another rope, suspended at the point  with the weight  on its end. The question is: where does the weight go if both the pulley and the ropes weigh nothing? Surprisingly, the solution and the construction for this problem are not that simple. But by playing with several variables you can make the picture look just right for the page while maintaining accuracy.
**Example code**`vardef lHopitalPulley (expr AB, l, m) = % distance AB between the suspension points of the ropes and their lengths l and m. “Why no units of length?”, you may ask. It's because some calculations inside can cause arithmetic overflow in MetaPost.
save A, B, C, D, E, o, a, x, y, d, w, h, support;
image(
pair A, B, C, D, E, o[];
path support;
numeric a, x[], y[], d[], w, h;
x1 := (l**2 + abs(l)*((sqrt(8)*AB)++l))/4AB; % the solution
y1 := l+-+x1; % second coordinate is trivial
y2 := m - ((AB-x1)++y1); % as well as the weight's position
A := (0, 0);
B := (AB*cm, 0);
D := (x1*cm, -y1*cm);
C := D shifted (0, -y2*cm);
d1 := 2/3cm; d2 := 1cm; d3 := 5/6d1; % diameters of the pulley, weight and the pulley wheel
w := 2/3cm; h := 1/3cm; % parameters of the wood block
o1 := (unitvector(C-D) rotated 90 scaled 1/2d3);
o2 := (unitvector(D-B) rotated 90 scaled 1/2d3);
E := whatever [D shifted o1, C shifted o1]
= whatever [D shifted o2, B shifted o2]; % pulley's center
a := angle(A-D);
support := A shifted (-w, 0) -- B shifted (w, 0) -- B shifted (w, h) -- A shifted (-w, h) -- cycle;
draw woodenThing(support, 0); % wood block everything is suspended from
draw pulley (d1, a - 90) shifted E; % the pulley
draw image(
draw A -- D -- B withpen thickpen;
draw D -- C withpen thickpen;
) maskedWith (pulleyOutline shifted E); % ropes should be covered with the pulley
draw sphere.c(d2) shifted C shifted (0, -1/2d2); % sphere as a weight
dotlabel.llft(btex $A$ etex, A);
dotlabel.lrt(btex $B$ etex, B);
dotlabel.ulft(btex $C$ etex, C);
label.llft(btex $l$ etex, 1/2[A, D]);
)
enddef;
beginfig(18);
draw lHopitalPulley (6, 2, 11/2); % now you can choose the right parameters
draw lHopitalPulley (3, 5/2, 3) shifted (8cm, 0);
endfig;`
And what about the textbook? Alas, when almost all the illustrations and the layout were ready, something happened and the textbook was canceled. Maybe because of that, I decided to rewrite most of the functions of the original library from scratch (I chose not to use any of the original code, which, although indirectly, I was paid for) and [put it on GitHub](https://github.com/jemmybutton/fiziko). Some things, present in the original library, such as functions for drawing cars and tractors, I didn't include there, some new features, e.g. knots, were added.
It doesn't run quickly: it takes about a minute to produce all the pictures for this article with LuaLaTeX on my laptop with i5-4200U 1.6 GHz. A pseudorandom number generator is used here and there, so no two similar pictures are absolutely identical (that's a feature) and every run produces slightly different pictures. To avoid surprises you can simply set `randomseed := some number` and enjoy the same results every run.
Many thanks to [dr ord](https://twitter.com/ordiology) and [Mikael Sundqvist](https://twitter.com/mpsmath) for their help with the English version of this text. | https://habr.com/ru/post/454376/ | null | null | 2,770 | 51.99 |
Every join point has a single set of modifiers - these include the standard Java modifiers such as public, private, static, abstract etc., any annotations, and the throws clauses of methods and constructors. These modifiers are the modifiers of the subject of the join point.
The following table defines the join point subject for each kind of join point.
For example, given the following types
public class X { @Foo protected void doIt() {...} } public class Y extends X { public void doIt() {...} }
Then the modifiers for a call to (Y y) y.doIt() are simply {public}. The modifiers for a call to (X x) x.doIt() are {@Foo,protected}. | https://www.eclipse.org/aspectj/doc/next/adk15notebook/join-point-modifiers.html | CC-MAIN-2021-43 | refinedweb | 106 | 69.48 |
Can you add energenie sockets as controlled devices.
I looked in list but it seems very limited.
Energenie
Can you add energenie sockets as controlled devices.
can you share a link to the device?
sure you can connect it. add a device using and then use energenie library to control it from the dashboard
wow. thanks . what a great response
you can create a pi plugin for energenie and help the cayenne community user to add engenie directly using pi agent and no need for coding.
i see energenie as an excellent add on to pi agent plugin list.
Thanks. Ive spent the last couple of hours looking through the information and trying things out. Im afraid its well out of my scope of experience.
I thought it would be easy to find the energenie pimote adapter and load it up but that doesnt seem to be the way it works.
I was also trying to add a dht11 sensor but again came unstuck
no issue here,
what is the issue you are facing?
if you can provide more detail, i can help solve it.
Thanks for your time on this its .much appreciated.
I cant see how to start and add the pimote board to your system.
same applies to the dht11 temp sensor.
I know both work as ive tested them with pyhon scripts.
Bob
First of all you need to create new device on cayenne: On your cayenne dashboard navigate to
add new--->devices/widget--->Bring your Own thing and you will get MQTT credential.
on you pi add this code and change the MQTT credential.
Next, run the code and you will see your device online on the dashboard. you will also see some green temporary widget, you can add them if you want.
Now you need a button to control the energenie device ON/OFF. For this add a custom button widget with channel 4 selected and add the below code in your code.
def on_message(message): print("message received: " + str(message)) if (message.channel == 4) and (message.value == "1"): switch_on(1) else: switch_on(0)
For DHT11 you can either use pi plugin or have a code to read DHT11 data and send it using the above code.
Hi,
I tried that info but get:
pi@raspberrypi:~/Desktop python cayenne.py Traceback (most recent call last): File "cayenne.py", line 2, in <module> import cayenne.client File "/home/pi/Desktop/cayenne.py", line 2, in <module> import cayenne.client ImportError: No module named client pi@raspberrypi:~/Desktop
I tried from the beginning but the same result.
Bob
Run this command on your pi:
git clone cd Cayenne-MQTT-Python python setup.py install
Thanks. Thats working now
I followed the dht11 sensor instructions but when i run
sudo pip3 install it just gives error.
you are missing the DOT
sudo pip3 install .
Thank you
That worked but now its the same message that it stops at the
import cayenne.client
Ive gone back and all the commands again but it comes back to halting at import cayenne.client
Bob
hold on, there are two separate method and seems like you using both. Use any one. either use python code or pi agent with plugin
Hi
I can see the message received and the value change from 0 to 1 etc but the code doesnt respond to the change.
can you share the code you are using?
Can you copy paste the code and the output here in place of sending the image as it is difficult to read it. | https://community.mydevices.com/t/energenie/11455 | CC-MAIN-2019-13 | refinedweb | 594 | 75 |
for generic
components:()
RGB, LED, back RyanteckRobot robot = RyanteckRobot()
curses module. This module requires that Python is
running in a terminal in order to work correctly, hence this recipe will
not work in environments like IDLE.
If you prefer a version that works under IDLE, the following recipe should
suffice, but will require that you install the evdev library with
sudo pip
install evdev first:
from gpiozero import RyanteckRobot from evdev import InputDevice, list_devices, ecodes robot = RyanteckRobot() devices = [InputDevice(device) for device in list_devices()] keyboard = devices[0] # this may vary keypress_actions = { ecodes.KEY_UP: robot.forward, ecodes.KEY_DOWN: robot.backward, ecodes.KEY_LEFT: robot.left, ecodes.KEY_RIGHT: robot.right, } for event in keyboard.read_loop(): if event.type == ecodes.EV_KEY: if event.value == 1: # key down keypress_actions[event.code]() if event.value == 0: # key up robot.stop() while True: with MCP3008(channel=0) as pot:) activity = LED(47)). | https://gpiozero.readthedocs.io/en/v1.2.0/recipes.html | CC-MAIN-2019-51 | refinedweb | 145 | 58.89 |
(367)
Hi, guys,
I'm Henrique, a 23 years old Backend Engineer from Blumenau, Brazil who has been developing software for 5 years.
I'm an architecture and backend enthusiast, mostly familiar with Java.
I look forward to keep on learning and keeping up with the changes in the tech world with you guys!
Heyyy 👋 , glad to have you aboard!
Hey!👋
hey!
Hello Henrique, I just joined the dev.to community! And came here to say Hi and hope we're gonna learn from each other.
great...
Hey!
hai sir
yo
Hi :D 👋
hi
Hey
bem vindo ao grupo meu amigo.
É tudo meu portugues 😄
Okay in all seriousness, hiya! Name’s Ray I live around Dallas, Texas and am a freshman attending high school! I have an interest in coding, and especially love Frontend but am also experimenting with Backend! Until the next time, peace!
PS: I love Furrets!
The places I could've been if I started coding in high school... Keep it up!
there's not much. no one takes you seriously when your in highschool and loves coding. it's very hard to get anything done with school and coding.
You just don't see it yet. When you get older and you tell people that you loved coding as a kid they will be impressed. Not to mention the kind of experience you have just from teaching yourself from such a young age.
But I feel like everyone does coding now days(including kids). I used to be happy thinking I was rare but look at repl.it. Most all of us are under 18(other than mods). I do think it is very good learning coding from a young age though(cause I absolutely love it).
Woah, hello RayhanADev... I think I know you from somewhere.... oh yes repl.it.
Can't wait to see what you make on dev.to (I just joined today myself)
Have fun with your repl api package! :D
LMAO I'm dead when seeing
import brain
Hey, a fellow Dallas dev - welcome!
Super creative - ❤️ it!
Sort of sad that totallyrealwebsite.lol doesn't exist yet...
Hoi Rayhan. Long time no talk. Join the org on GitHub. Go here. It seems like the whole repl community is here. I just joined myself.
Wait a sec this looks wrong, i created the org 🤦♂️🤦♂️
Ray! I didn't know you were on dev.to!! Nice to see you here
Hi, I joined today cause of Coder100's post. The whole repl community seems to be here.
Hello!
I am a mom of three little ones. I have been learning Frontend since right after my first child was born. In my spare time, I learn and practice web development as much as I can. I can't wait to get my first real job in Frontend development.
Welcome, Lauren to the Dev Community. Being a mother is a really tough job. Thumbs up for your courage to learn new things. Hope you and the kids are doing well. Love from India
Hello!
That's great that you have so many kids!
I have three kids, too. I have three sons.
I love Java.
Hello, I'm new here. I'm 24, from Nigeria. I just started out learning web development and I'm currently on the front end working with html, Css and javascript. My goal is to become a full-stack developer
hey!
Welcome Sulaiman 👋
Hola comunidad, entrando al desarrollo web, primeros pasos, html,css, python, js entre otros lenguajes. saludos cordiales! desde Buenos Aires Argentina.
Hello community, entering web development, first steps, html, css, python, js among other languages. best regards! from Buenos Aires Argentina.
Hi guys and gals,
My name is Katelynn Tenbrook and I have a passion for technology and what can be improved with it. I have been developing my coding skills for 2 years now. I'm proud to say that I know ASP. Net, C#, SQL, JavaScript and much more. I am currently developing my skills in the MERN stack and I'm super excited to start writing some articles centered around technology. I hope you drop me a follow :). Fun fact about me, my favorite food is tacos.
Hey!
How can you delete all series that you previously created but you don't need them anymore?
My OCD keeps screaming on that badge.
Hi! I am Holly. I have been here for a few days so I guess I should introduce myself. I am currently learning C#. Once my spring semester starts I will be adding Java to the mix. I am a career changer in progress after spending a lot of years in the dental field.
Hello guys!
My name is Alex, I'm from Campinas-SP-Brazil. I'm acting with IT for more than 5 years. I hope to learn a lot here and to contribute too. I'm like to write articles about technologies, architecture and C# development specifically. And in out of time, I'm like to see soccer and play it too.
Hi everyone!
I'm Kyle - hacker, software engineer, and most recently, co-founder of a SaaS startup that sends real-time application error notifications to code owners (shameless plug: codelighthouse.io).
I'm a big fan of Python, Node.js, and Golang, but have experience in frontend dev too. I love building with Linux and cloud technologies, especially serverless.
I'm looking forward to learning with y'all and contributing to tech discussions!
Hi everyone,
I am Riccardo 30yo from Munich, Germany. I have been professionally working on the business side in few tech startup, this year I have joined Le Wagon bootcamp and I have learned how to code.
In the past months, I have launched my startup project which is targeting the digital talents - people that spend many hours in front of the computer. For this reason I am about to launch a new eyewear brand called ÒCIO eyewear - @ocio.eyewear on Instagram. These glasses are all that you need: (a) high quality eco friendly materials, (b) blue light blocking glasses and (c) have a magnetic sunglass clip-on :)
Looking forward to learn and contribute on interesting tech discussions.
Cheers,
Riccardo
Hi 🙏 everyone,
I'm Pratik, a software engineer working on open source projects for 2 years now and that could sum up my total job experience.
I like Golang, Linux, Python, Kubernetes, Football and I'm from India so spicy food (by default)😉.
I want to start tech blogging. The main reason is I want to reinforce my learning and become a better techie. The best way to do this is by teaching and sharing. If someone can help me to get started in blogging that would be great.
Let's learn and evolve together.
AND TECH IS FUN! 😁
HAPPY LEARNING, HAPPY SHARING AND BE SAFE.
Hi there, my name is Akshaj. I'm a 15 year old front end developer and I am learning back-end developing. Currently I am having knowledge of HTML5, CSS and JavaScript. I am thinking to learn python and Java. I live in India and aims to join Google. And I will achieve it.😤
Welcome! It's nice to see someone so young! 👏
Thank you🤗
hi to all! i'm Vincenzo, i'm 38 and i'm a developer from italy. i work withi php and javascript. i was using Kohana framework becouse my agency asket it to me, now i'm switching in laravel. i read dev articles from month, i love them, so i decide to sign in
Hi Vincenzo 😃 Welcome to DEV.
thanks :)
Hello all,
I'm Bhavik, 20 years old backend engineer from Ahmedabad, India who is final year student in B.E.
I have designed & managed my college website and participated in 4 hackathons (1-winner). I have worked with PHP, Java, Node.js in my academic career.
Currently I am learning
Railsand looking forward to get knowledge and trends with this community.
Hi everyone,
My name is Daniel, I am 25 years old and I live in Paris.
I am a full-stack freelance developer and I mainly develop on React, React Native, TypeScript, Node.js, GraphQl.
I am very passionate about music. Outside of my dev time I compose with FL Studio. I have fun creating trap, edm, raggaeton and other genres
Hi everyone, my name is Anthony, I've always been interested in computers\ programming. A couple of years ago my business was hacked and I. Could not get help from anyone. I mean no one so for the last two years I have been learning as much as I can on my own but I am at the point where I need to learn from others and hope to do so here. Nice meeting you guys in advance
Hi Anthony, welcome! I'm a big fan of cybersecurity too (Just earned my OSCP cert). I'd love to chat about it if you need help!
Hello guys!
I'm Arvin, a 19 years old student, studying BS in Information Technology at Philippines.
I love learning and mostly passionate when I am highly motivated and productive.
I'm interested in Full-Stack Development, AI, Machine Learning and more. I may not know how, but I am interested learning about it.
With all that said, I am new to this community, I don't know much about it, but I am looking forward to know a lot of people and learn from them too. 😄
Hi, I am whippingdot, a kid who likes coding and who wants to learn Nodejs, JS, and CSS. I know C++, Python, and HTML.
Follow my GitHub here and see my repl account here. I am currently making the biggest project I have ever made, though I think it is only going to be 300 lines. C ya later, bye.
Hi all,
I'm Avinesh from India. I have 1 year of experience in IT. Recently I was introduced to someone on LinkedIn and this guy has started a web development course on Youtube. I started following it and I was encouraged to be a part of a community which helps you learn and grow. So here I am. I would be writing a couple of blogs on my understanding of a JavaScript book.
Looking forward to join you all in this journey.
Hi there, I'm Nick - aka RedbeardJunior
I'm a Father and a Developer, I love to learn new thing's im from the Netherlands and 27 years old i'm making software far over 5 years now still learning day by day.
Be save and I wish you all a good health and a happy 2021 !
Hi Nick, Welcome to Dev Community. Hope this community will help you to learn more day by day 😄.
Have a great day.
Hello world
I'm Stephen Mclin, a 17 year old tech enthusiast. I like everything about tech from programming to digital art. I currently have been more interested in web development and I'm on my journey to learn technologies like PHP, JavaScript and other web development frameworks.
I think it's gonna be fun working and learning with you guys 👍
Hey Stephen, love your enthusiasm! 🤩
Hey there,
I'm Oladipo Glory, an upcoming web designer/ developer. I am 20 years old, I just started learning not quite long and it's been interesting.
Anyways I am a student and I can't wait to get more than enough from you guys.
import {name, skills, passion} from './myself';
OUTPUT:
Hi Everyone!
I'm Goodnews Daniel from the United Kingdom. I love technology, though I started a bit late as I switched my career from Law to Software Engineering. So I'm here to learn how to develop web/mobile apps and hopefully connect with mentors and potential partners.
Hello Folks,
I'm Mohan, a 35 year old QA Engineer from US. I have been QA my whole life until now and looking to transition to development.
I've come across this website as I read a blog that explains the pitfalls of changing the roles and strategy to follow to avoid them. I've immediately created an account on DEV as it seems to have lot of authors and blogs that could help in my journey. Looking forward to share my experiences as I go...
Hi All,
My name is Salman Siddique, I am a Computer Science graduate and working as Tech Content Marketer at Cloudways.
I am married to Marketing and Tech is my BFF. I am a constant learner, currently polishing my PHP skills and getting into headless CMSs.
Hoping to learn and contribute a ton to dev.to the community. Cheers! <3
Dear All,
Lemme introduce myself. My name is Affandy Murad but you can call me Fandy. Currently working as iOS Developer at Otten Coffee, an Coffee online marketplace in Indonesia.
I'm working as mobile developer since 2016. My first experience as Android developer with Java language for 2 years, then carry on with Kotlin for 2 years, then become iOS developer using Swift until right now.
I look for new friends from another country to exchange our story and knowledge, either programming or culture. Nice to know you all!
hi guys,
I'm having issues installing SocialFish on my termux android followed the steps but the main software interface won't pop up.... having the error code that I haven't installed the necessary software to make it run
Hey everyone 👋!
My name is Kai, I'm an indie maker and web developer from Germany. I completed my studies with a Master of Science (Business Informatics) and just started blogging about web development and business-related stuff (kais.blog).
I'll try to regularly share my web development experience with you. Also, I hope you'll find it interesting to learn more about my life as an indie maker and some of my questionable life choices. 😏
Thank you everyone! Let's have a good time!
PS: I've just published my first (real) blog post. Until now I have only created a little tutorial series for this year's Advent of Code.
14 Awesome JavaScript Array Tips You Should Know About
Kai ・ Dec 9 ・ 8 min read | https://practicaldev-herokuapp-com.global.ssl.fastly.net/thepracticaldev/welcome-thread-v103-4dap | CC-MAIN-2022-27 | refinedweb | 2,379 | 76.32 |
#include <posix_status.hpp>
This class represents the status returned by a child process after it has terminated. It contains some methods not available in the status class that provide information only available in POSIX systems.
Creates a posix_status object from an existing status object.
Creates a new status object representing the exit status of a child process. The construction is done based on an existing status object which already contains all the available information: this class only provides controlled access to it.
If signaled, returns the terminating signal code.
If the process was signaled, returns the terminating signal code.
If signaled, returns whether the process dumped core.
If the process was signaled, returns whether the process produced a core dump. | http://www.highscore.de/boost/process/reference/classboost_1_1process_1_1posix__status.html | CC-MAIN-2018-05 | refinedweb | 120 | 58.08 |
Many,
Chris Boyd
Hi,
I’m currently assigned to a task to do some web parts for the new project server 2007.
Unfortunately as I haven’t even started yet while trying out calling the web services through the PSI I got stuck with what I can only guess are authentification problems. I’ve been going through every single post and sources I could find but so far it seems as no one had have this kind of a problem.
The problem itself starts with the simple set of a web reference to the (for example) Project web service (none of the rest work either, except for the loginForms and loginWindows). When trying to make a reference the studio can’t find the services unless not written like () But then the call for wsdl in runtime results in SOAP ex. “Possible SOAP version mismatch:….”.
If I try getting to the web service following the local path () the studio has no problems there but at runtime I get the 401 Unauthorized (even if the test class I did is run on the server machine where the user must have all the rights needed).
One more try I did using the “LoginDemo” example which comes with the SDK(where hoping that as it works fine can just add an extra button with simple call to some other web service which will prove that there is something I haven’t done in my previous tries). After I added the button and a call to the resource.asmx on_click
: System.Diagnostics.Trace.WriteLine(res.GetCurrentUserUid());
Where public static LoginDemo.WebSvcResource.Resource res =
new LoginDemo.WebSvcResource.Resource();
when I did try it out the login demo still worked fine but as for my call it threw a (System.Net.WebException' occurred in System.Web.Services.dll 401 unathorized ex. Again when the web reference was the local path for the service) and the (Possible SOAP version mismatch: Envelope namespace was unexpected. Expecting) when trying out the virtual path. I guessed the trick was hidden somewhere in the cookies or the credentials but after made sure they were alright nothing changed at all.
As I already said I’m a beginner and perhaps I have missed something quite simple and my problem might seem a silly one but I’ve spend over a week looking for a solution and a way to make a call to the web services that would work with no success so far. I’d appreciate any help or full sample codes that work fine to use as an example.
Just wondering has anyone had any problems of the sort?
HELP mates…getting desperate in here!
Thanks in advance
Cheers Hristiyan
Once again after trying so hard and at the final gave up and started whining for help the problem got a name and was solved but not before I made a big fuss about it. Sorry for bothering with a obvious mistake of mine:)
Hristiyan
I am happy that you got it working!
Can you please post how you solved the issue in case someone runs into the same problem?
Thanks,
Well in my case, for my great shame, the whole problem was from that I had missed the name of the Project Web Access instance I was connecting to. In other words I used “” where the correct one has the name “pwa” which refers to the project web access – “”.
Funny thing was that I could call the “loginForms” and “loginWindows” services without any problems whatsoever from where on I guessed I should not have any problems with the rest either but the fact is - it wasn’t exactly true and I found it the hard way.
Anyways I didn’t have the nerves to continue digging in the case “How can I call the services anyway without going through the pwa” but if I do it one day and find a "go around the authentication" way I will post it for sure.
Cheers
Hi Chris,
sorry for posting that as a comment here (guess not really the right place) but I seriosly need some help with this one and I do hope you'll noticed that faster from here and make a sugestion or give any kind of an advice!
I’m currently developing a web part which should work with the project’s timesheets but I’m stuck with a problem that has no solution so far.
What I’m trying to do is multiple new entries in a timesheet, for any given period for a project line for a specific date. The current MS Timesheet works in a way that let’s you have only one row where the projectUID, taskUid and the classLineUid relation must be unique and for that same row the corresponding values in the actuals are in one row where the altual’s lineUid (the lineUid column is the foreign key) in relation with the date entry is constrained to be unique as well.
Now, in order to make a new actual’s entry for the same date and still the 'TS_LINE_UID, TS_ACT_START_DATE' could be kept unique I’ve tried make entries for the same date but with difference in the time (as the date column has values of DateTime). The whole thing goes like that – it checks out if the line exists in the lines datatable and if it does gets the lineUid and creates a new row in the actuals datatable with lineUid, the value from the lines datatable. Calling the dataset.Actuals.AddActualsRow() method goes through fine. Than the next call to the timeSheetService.PrepareTimesheetLine() method goes without any errors as well but once I try to write the changes of the datatables in the database, done in a way:
Guid jobUid = Guid.NewGuid();
timeSheetService.QueueUpdateTimesheet(jobUid,
timesheetUid, dataset);
WaitForQueue(queueService, jobUid);
the WaitForQueue just waits for the job to get through the queue and shows the current state of the job from which I can see that my job goes from “sleeping” state to “proceeding” and eventually just before gets to “succeed” state it gets cancelled out without any particular reason.
Are there any examples available of the way the job’s queuing in the project server goes or has any reference to some documentation on that matter?
One more quite interesting error I get from time to time is the “GeneralQueueCorrelationBlocked” which means absolutely nothing to me and as I can’t find what might be the most probable cause of it I have no clue where to begin to look for a problem from.
If the above makes any sense to you (most probably you’ve deal with same things and eventually encountered some problems of the sort) PLEASE send suggestions or any references to examples that might help!
One more thing, the above is tested on BETA2 release and so far I can only hope the most probable cause of the problem is fixed in the final release but I won’t be able to find that out at least for another few days, so until that time I’ll be banging my head in the wall while trying to understand what the project server does and what is it that it has a problem with mine project!:)
Any help is welcomed!
Cheers
Hi Chris:
I've been trying your code on a Project Server 2007 enterprise installation, and I'm getting what I think is an authentication error (WebException with text "Unable to connect to the remote server") when calling the ReadProjectList method. The Web Service properly connects (I think) because I get the "projWS" variable properly instantiated.
My machine is on the domain, and my user is a PS Server Admin.
Could it be that the CredentialCache.DefaultCredentials is not providing the proper credentials to the server?
Any thoughts? Thanks in advance!!
Nelson
I've found the cause for the previous error! As it seems Mcafee got a little too restrictive with my app and was not allowing it to connect to the server. I disabled Access Protection and is all working now. Thanks anyway!!
Regards
I'm concerned with potential issues around scalability and performance of ReadProjectList and ReadProjectStatus with PSI customizations that need to process a small subset of projects against a PWA instance with a good amount of projects.
Your example here shows retrieval of all projects into a dataset:
DataTable projList = projWS.ReadProjectList().Tables[0];
foreach (DataRow dr in projList.Rows)
{
....
/////////////////////
//Retrieving Projects
I have noticed the filter xml capability (found in Calendars, Custom Fields, Lookup Tables, Resources, Assignments, etc.) is not a capability of retrieving Projects. Neither GetProjectList nor GetProjectStatus has the filter parameter and I do not see how one would, for example, request a list of projects without pulling every project over and looping through the resulting dataset. So if I had 1500 projects and only wanted those that belong to a particular department (maybe 15 projects), I need to pull over a list of 1500 projects to find the 15? This seems pretty expensive.
///////////////////////
// 1000 row limitation?
Also, I've seen mention in blogs of a limitation of a maximum of 1000 rows in the resulting Project dataset on these calls. This does not seem believable but is this true? I have not yet tested this but it concerns me as we have clients with 1000+ projects and I don't want my customizations failing the first time ReadProjectList() is encountered.
I hope neither case is true but if so, is there a workaround? Maybe a paging or filter mechanism in the SDK I'm not seeing? If this is true might it be wise for me to ponder the construction of a PSI extension to hit the working/published databases directly (read-only with no locks) to retrieve a filtered list of Projects UIDs that I could later call the single project methods on?
Paul Congdon
Chris,
Can I get you to sent me some code or direct me in the path of doing this with forms authentication?
thanks
Freddie
I'm transitioning from VB.net to C# and am having a problem with the above code. I get several errors during build with a common theme. Here is the first:
Error 1 The name 'conn' does not exist in the current context C:\Inetpub\MSPS_PSI_C\Default.aspx.cs 98 9 C:\Inetpub\MSPS_PSI_C\
Any ideas on why I can't set the variable conn?
thanks,
Alex
Entire code:;
}
public partial class _Default : System.Web.UI.Page
protected void Page_Load(object sender, EventArgs e)
private void txtURL_Leave(object sender, EventArgs e)
ddlProjects.Items.Clear();
conn = new PSIDemo.Connection(txtURL.Text);
projWS = (WSProject.Project)conn.GetWebService(PSIDemo.Connection.Project);
DataTable projList = projWS.ReadProjectList().Tables[0];
foreach (DataRow dr in projList.Rows)
ddlProjects.Items.Add(new ProjListItem(dr["Proj_Name"].ToString(), new Guid(dr[0].ToString())));
if (ddlProjects.Items.Count > 0)
ddlProjects.SelectedItem = ddlProjects.Items[0];
private void ddlProjects_SelectedIndexChanged(object sender, EventArgs e)
lstResources.Items.Clear();
WSProject.ProjectTeamDataSet pds;
ProjListItem projItem = (ProjListItem)ddlProjects.SelectedItem;
pds = projWS.ReadProjectTeam(projItem.getGuid());
DataTable dt = pds.Tables["ProjectTeam"];
foreach (DataRow dr in dt.Rows)
lstResources.Items.Add(dr["Res_Name"].ToString()); | http://blogs.msdn.com/b/project_programmability/archive/2006/10/16/getting-started-with-the-psi.aspx | CC-MAIN-2015-27 | refinedweb | 1,857 | 60.14 |
Control.FRPNow.Core
Description
Synopsis
- data Event a
- data Behavior a
- never :: Event a
- switch :: Behavior a -> Event (Behavior a) -> Behavior a
- whenJust :: Behavior (Maybe a) -> Behavior (Event a)
- futuristic :: Behavior (Event a) -> Behavior (Event a)
- data Now a
- async :: IO a -> Now (Event a)
- asyncOS :: IO a -> Now (Event a)
- callback :: Now (Event a, a -> IO ())
- sampleNow :: Behavior a -> Now a
- planNow :: Event (Now a) -> Now (Event a)
- sync :: IO a -> Now a
- runNowMaster :: Now (Event a) -> IO a
- initNow :: (IO (Maybe a) -> IO ()) -> Now (Event a) -> IO ()
Pure interface
The FRPNow interface is centered around behaviors, values that change over time, and events, value that are known from some point in time on.
What the pure part of the FRPNow interface does is made precise by denotation semantics, i.e. mathematical meaning. The denotational semantics of the pure interface are
type Event a = (Time+,a) never :: Event a never = (∞, undefined) instance Monad Event where return x = (-∞,x) (ta,a) >>= f = let (tb,b) = f a in (max ta tb, b) type Behavior a = Time -> a instance Monad Behavior where return x = λt -> x m >>= f = λt -> f (m t) t instance MonadFix Behavior where mfix f = λt -> let x = f x t in x switch :: Behavior a -> Event (Behavior a) -> Behavior a switch b (ts,s) = λn -> if n < ts then b n else s n whenJust :: Behavior (Maybe a) -> Behavior (Event a) whenJust b = λt -> let w = minSet { t' | t' >= t && isJust (b t') } in if w == ∞ then never else (w, fromJust (b w))
Where
Time is a set that is totally ordered set and has a least element, -∞.
For events, we also use
Time+ = Time ∪ ∞.
The notation
minSet x indicates the minimum element of the set
x, which is not valid Haskell, but is a valid denotation. Note that if there is no time at which the input behavior is
Just in the present or future, then
minSet will give the minimum element of the empty set, which is
∞.
The monad instance of events is denotationally a writer monad in time, whereas the monad instance of behaviors is denotationally a reader monad in time.
switch :: Behavior a -> Event (Behavior a) -> Behavior a Source
Introduce a change over time.
b `switch` e
Gives a behavior that acts as
b initially, and switches to the behavior inside
e as soon as
e occurs.
whenJust :: Behavior (Maybe a) -> Behavior (Event a) Source
Observe a change over time.
The behavior
whenJust b gives at any point in time the event that
the behavior
b is
Just at that time or afterwards.
As an example,
let getPos x | x > 0 = Just x | otherwise = Nothing in whenJust (getPos <$> b)
Gives gives the event that
the behavior
b is positive. If
b is currently positive
then the event will occur now, otherwise it
will be the first time that
b becomes positive in the future.
If
b never again is positive then the result is
never.
futuristic :: Behavior (Event a) -> Behavior (Event a) Source
Not typically needed, used for event streams.
If we have a behavior giving events, such that each time the behavior is sampled the obtained event is in the future, then this function ensures that we can use the event without inspecting it (i.e. before binding it).
If the implementation samples such an event and it turns out the event does actually occur at the time the behavior is sampled, an error is thrown.
IO interface
A monad that alows you to:
- Sample the current value of a behavior via
sampleNow
- Interact with the outside world via
async,
callbackand
sync.
- Plan to do Now actions later, via
planNow
All actions in the
Now monad are conceptually instantaneous, which entails it is guaranteed that for any behavior
b and Now action
m:
do x <- sample b; m ; y <- sample b; return (x,y) == do x <- sample b; m ; return (x,x)
Instances
async :: IO a -> Now (Event a) Source
Asynchronously execte an IO action, and obtain the event that it is done.
Starts a seperate thread for the IO action, and then immediatly returns the
event that the IO action is done. Since all actions in the
Now monad are instantaneous,
the resulting event is guaranteed to occur in the future (not now).
Use this for IO actions which might take a long time, such as waiting for a network message, reading a large file, or expensive computations.
Note:Use this only when using FRPNow with Gloss or something else that does not block haskell threads.
For use with GTK or other GUI libraries that do block Haskell threads, use
asyncOS instead.
asyncOS :: IO a -> Now (Event a) Source
callback :: Now (Event a, a -> IO ()) Source
Create an event that occurs when the callback is called.
The callback can be safely called from any thread. An error occurs if the callback is called more than once.
See
callbackStream for a callback that can be called repeatidly.
The event occurs strictly later than the time that the callback was created, even if the callback is called immediately.
planNow :: Event (Now a) -> Now (Event a) Source
sync :: IO a -> Now a Source
Synchronously execte an IO action.
Use this is for IO actions which do not take a long time, such as opening a file or creating a widget.
Entry point
runNowMaster :: Now (Event a) -> IO a Source
Run the FRP system in master mode.
Typically, you don't need this function, but instead use a function for whatever library you want to use FRPNow with such as
runNowGTK,
runNowGloss. This function can be used in case you are not interacting with any GUI library, only using FRPNow.
Runs the given
Now computation and the plans it makes until the ending event (given by the inital
Now computation) occurs. Returns the value of the ending event.
Arguments
General interface to interact with the FRP system.
Typically, you don't need this function, but instead use a specialized function for whatever library you want to use FRPNow with such as
runNowGTK or
runNowGloss, which themselves are implemented using this function. | https://hackage.haskell.org/package/frpnow-0.18/docs/Control-FRPNow-Core.html | CC-MAIN-2017-30 | refinedweb | 1,017 | 62.31 |
A one sample t-test is used to determine whether or not the mean of a population is equal to some value.
This tutorial explains how to conduct a one sample t-test in Python.
Example: One Sample t-Test in Python
Suppose a botanist wants to know if the mean height of a certain species of plant is equal to 15 inches. She collects a random sample of 12 plants and records each of their heights in inches.
Use the following steps to conduct a one sample t-test to determine if the mean height for this species of plant is actually equal to 15 inches.
Step 1: Create the data.
First, we’ll create an array to hold the measurements of the 12 plants:
data = [14, 14, 16, 13, 12, 17, 15, 14, 15, 13, 15, 14]
Step 2: Conduct a one sample t-test.
Next, we’ll use the ttest_1samp() function from the scipy.stats library to conduct a one sample t-test, which uses the following syntax:
ttest_1samp(a, popmean)
where:
- a: an array of sample observations
- popmean: the expected population mean
Here’s how to use this function in our specific example:
import scipy.stats as stats #perform one sample t-test stats.ttest_1samp(a=data, popmean=15) (statistic=-1.6848, pvalue=0.1201)
The t test statistic is -1.6848 and the corresponding two-sided p-value is 0.1201.
Step) is greater than alpha = 0.05, we fail to reject the null hypothesis of the test. We do not have sufficient evidence to say that the mean height for this particular species of plant is different from 15 inches.
Additional Resources
How to Conduct a Two Sample T-Test in Python
How to Conduct a Paired Samples T-Test in Python | https://www.statology.org/one-sample-t-test-python/ | CC-MAIN-2022-21 | refinedweb | 298 | 72.26 |
Design and manufacture systems with flash memory¶
The tools can be used to target xCORE devices that use SPI flash memory for booting and persistent storage. The xCORE flash format is shown in Flash format diagram.
The flash memory is logically split between a boot and data partition. The boot partition consists of a flash loader followed by a “factory image” and zero or more optional “upgrade images.” Each image starts with a descriptor that contains a unique version number, a header that contains a table of code/data segments for each tile used by the program and a CRC. By default, the flash loader boots the image with the highest version with a valid CRC.
Boot a program from flash memory¶
To load a program into an SPI flash memory device on your development board, start the tools and enter the following commands:
-
xflash -l
XFLASH prints an enumerated list of all JTAG adapters connected to your PC and the devices on each JTAG chain, in the form:
ID Name Adapter ID Devices
-- ---- ---------- -------
-
xflash --id ID *program*.xe
XFLASH generates an image in the xCORE flash format that contains a first stage loader and factory image comprising the binary and data segments from your compiled program. It then writes this image to flash memory using the xCORE device.
Caution
The XN file used to compile your program must define an SPI flash device and specify the four ports of the xCORE device to which it is connected.
Generate a flash image for manufacture¶
In manufacturing environments, the same program is typically programmed into multiple flash devices.
To generate an image file in the xCORE flash format, which can be subsequently programmed into flash devices, start the tools and enter the following command:
xflash *program*.xe -o image-file
XFLASH generates an image comprising a first stage loader and your program as the factory image, which it writes to the specified file.
Perform an in-field upgrade¶
The tools and the libflash library let you manage multiple firmware upgrades over the life cycle of your product. You can use XFLASH to create an upgrade image and, from within your program, use libflash to write this image to the boot partition. Using libflash, updates are robust against partially complete writes, for example due to power failure: if the CRC of the upgrade image fails during boot, the previous image is loaded instead.
Write a program that upgrades itself¶
The example program below uses the libflash library to upgrade itself.
#include <platform.h> #include <flash.h> #define MAX_PSIZE 256 /* initializers defined in XN file * and available via platform.h */ fl_SPIPorts SPI = { PORT_SPI_MISO, PORT_SPI_SS, PORT_SPI_CLK, PORT_SPI_MOSI, XS1_CLKBLK_1 }; int upgrade(chanend c, int usize) { /* obtain an upgrade image and write * it to flash memory * error checking omitted */ fl_BootImageInfo b; int page[MAX_PSIZE]; int psize; fl_connect(SPI); psize = fl_getPageSize(); fl_getFactoryImage(b); fl_getNextBootImage(b); while(fl_startImageReplace(b, usize)) ; for (int i=0; i page[j];) fl_writeImagePage(page); fl_endWriteImage(); fl_disconnect(); return 0; } int main() { /* main application - calls upgrade * to perform an in-field upgrade */ }
The call to
fl_connect opens a connection between the xCORE and SPI
devices, and the call to
fl_getPageSize determines the SPI device’s
page size. All read and write operations occur at the page level.
The first upgrade image is located by calling
fl_getFactoryImage and
then
getNextBootImage. Once located,
fl_startImageReplace
prepares this image for replacement by a new image with the specified
(maximum) size.
fl_startImageReplace must be called until it returns
0, signifying that the preparation is complete.
The function
fl_writeImagePage writes the next page of data to the
SPI device. Calls to this function return after the data is output to
the device but may return before the device has written the data to its
flash memory. This increases the amount of time available to the
processor to fetch the next page of data. The function
fl_endWriteImage waits for the SPI device to write the last page of
data to its flash memory. To simplify the writing operation, XFLASH adds
padding to the upgrade image to ensure that its size is a multiple of
the page size.
The call
fl_disconnect closes the connection between the xCORE and
SPI devices.
Build and deploy the upgrader¶
To build and deploy the first release of your program, start the tools and enter the following commands:
-
xcc *file*.xc -target=*boardname* -lflash -o first-release.xe
XCC compiles your program and links it against libflash. Alternatively add the option -lflash to your Makefile.
-
xflash first-release.xe -o manufacture-image
XFLASH generates an image in the xCORE flash format that contains a first stage loader and the first release of your program as the factory image.
To build and deploy an upgraded version of your program, enter the following commands:
-
xcc *file*.xc -target=*boardname* -lflash -o latest-release.xe
XCC compiles your program and links it against libflash.
-
xflash --upgrade *version* latest-release.xe --factory-version *tools-version* -o upgrade-image
XFLASH generates an upgrade image with the specified version number, which must be greater than 0. Your program should obtain this image to upgrade itself.
If the upgrade operation succeeds, upon resetting the device the loader boots the upgrade image, otherwise it boots the factory image. | https://www.xmos.ai/documentation/XM-014363-PC-4/html/tools-guide/tutorials/design-with-flash/flash.html | CC-MAIN-2022-21 | refinedweb | 872 | 51.28 |
Error Handling
Netpbm Programming Library Errors
As part of Netpbm’s mission to make writing graphics programs quick and
easy, Netpbm recognizes that no programmer likes to deal with error
conditions. Therefore, very few Netpbm programming library functions return
error information. There are no return codes to check. If for some reason a
function can’t do what was asked of it, it doesn’t return at all.
Netpbm’s response to encountering an error is called "throwing an error."
The typical way a Netpbm function throws an error (for example, when you
attempt to open a non-existent file with pm_openr()) is that the function
writes an error message to the Standard Error file and then causes the
program to terminate with an exit() system call. The function doesn’t do any
explicit cleanup, because everything a library function sets up gets cleaned
up by normal process termination.
In many cases, that simply isn’t acceptable. If you’re calling Netpbm
functions from inside a server program, you’d want the program to recognize
that the immediate task failed, but keep running to do other work.
So as an alternative, you can replace that program exit with a longjmp
instead. A longjmp is a classic Unix exception handling concept. See the
documentation of the standard C library setjmp() and longjmp() functions.
In short, you identify a point in your programs for execution to hyperjump
to from whatever depths of whatever functions it may be in at the time it
detects an exception. That hyperjump is called a longjmp. The longjmp
unwinds the stack and puts the program in the same state as if the
subroutines had returned all the way up to the function that contains the
jump point. A longjmp does not in itself undo things like memory
allocations. But when you have a Netpbm function do a longjmp, it also
cleans up everything it started.
To select this form of throwing an error, use the pm_setjmpbuf() function.
This alternative is not available before Netpbm 10.27 (March 2005).
Issuing of the error message is a separate thing. Regardless of whether a
library routine exits the program or executes a longjmp, it issues an error
message first.
You can customize the error message behavior too. By default, a Netpbm
function issues an error message by writing it to the Standard Error file,
formatted into a single line with the program name prefixed. But you can
register your own error message function to run instead with
pm_setErrorMsgFn().
pm_setjmpbuf()
pm_setjmpbuf() sets up the process so that when future calls to the Netpbm
programming library throw an error, they execute a longjmp instead of
causing the process to exit as they would by default.
This is not analogous to setjmp(). You do a setjmp() first, then tell the
Netpbm programming library with pm_setjmpbuf() to use the result.
Example:
#include <setjmp.h>
#include <pam.h>
jmp_buf jmpbuf;
int rc;
rc = setjmp(jmpbuf);
if (rc == 0) {
struct pam pam;
pm_setjmpbuf(&jmpbuf);
pnm_readpam(stdin, &pam, PAM_STRUCT_SIZE(tuple_type));
printf("pnm_readpam() succeeded!0);
} else {
printf("pnm_readpam() failed. You should have seen "
"messages to Standard Error telling you why.0);
}
This example should look really strange to you if you haven’t read the
documentation of setjmp(). Remember that there is a hyperjump such that the
program is executing the pnm_readpam() and then suddenly is returning a
second time from the setjmp()!
Even pm_error() works this way -- if you set up a longjmp with
pm_setjmpbuf() and then call pm_error(), pm_error() will, after issuing your
error message, execute the longjmp.
pm_setjmpbuf() was new in Netpbm 10.27 (March 2005). Before that, Netpbm
programming library functions always throw an error by exiting the program.
User Detected Errors
The Netpbm programming library provides a function for you to throw an error
explicitly: pm_error(). pm_error() does nothing but throw an error, and does
so the same way any Netpbm library function you call would. pm_error() is
more convenient than most standard C facilities for handling errors.
If you don’t want to throw an error, but just want to issue an error
message, use pm_errormsg(). It issues the message in the same way as
pm_error() but returns normally instead of longjmping or exiting the
program.
Note that libnetpbm distinguishes between an error message and an
informational message (use pm_errormsg() for the former; pm_message() for
the latter). The only practical difference is which user message function it
calls. So if you don’t register any user message function, you won’t see any
difference, but a program is still more maintainable and easier to read when
you use the appropriate one of these.
pm_error()
Overview
void pm_error( char * fmt, ... );
Example
if (argc-1 < 3)
pm_error("You must specify at least 3 arguments. "
"You specified" only %d", argc-1);
pm_error() is a printf() style routine that simply throws an error. It
issues an error message exactly like pm_errormsg() would in the process.
pm_errormsg()
Overview
void pm_errormsg( char * fmt, ... );
Example
if (rc = -1)
pm_errormsg("Could not open file. errno=%d", errno);
return -1;
pm_errormsg() is a printf() style routine that issues an error message. By
default, it writes the message to Standard Error, but you can register a
user error message routine to be called instead, and that might do something
such as write the message into a log file. See pm_setusererrormsgfn.
There is very little advantage to using this over traditional C services,
but it issues a message in the same way as libnetpbm library functions do,
so the common handling might be valuable.
Note that the arguments specify the message text, not any formatting of it.
Formatting is handled by pm_errormsg(). So don’t put any newlines or tabs in
it.
pm_setusererrormsgfn()
Overview
void pm_setusererrormsgfn(pm_usererrormsgfn * function);
Example
static pm_usererrormsgfn logfilewrite;
static void
logfilewrite(const char * const msg) {
fprintf(myerrorlog, "Netpbm error: %s", msg);
}
pm_setusererrormsgfn(&logfilewrite);
pm_errormsg("Message for the error log");
pm_setusererrormsg() registers a handler for error messages, called a user
error message routine. Any library function that wants to issue an error
message in the future will call that function with the message as an
argument.
The argument the user error message routine gets is English text designed
for human reading. It is just the text of the message; there is no attempt
at formatting in it (so you won’t see any newline or tab characters).
You can remove the user error message routine, so that the library issues
future error messages in its default way (write to Standard Error) by
specifying a null pointer for function.
The user error message routine does not handle informational messages. It
handles only error messages. See pm_setusermessagefn().
Error Handling In Netpbm Programs
Most Netpbm programs respond to encountering an error by issuing a message
describing the error to the Standard Error file and then exiting with exit
status 1.
Netpbm programs generally do not follow the Unix convention of very terse
error messages. Conventional Unix programs produce error messages as if they
had to pay by the word. Netpbm programs tend to give a complete description
of the problem in human-parseable English. These messages are often many
terminal lines long. | https://manpag.es/YDL61/1+error | CC-MAIN-2021-49 | refinedweb | 1,189 | 63.49 |
Hi there...
I am pretty much a newbie working with Flash, so I would really appreciate any kind of help.
I have created a game, with lot of "brain burn" and getting help from the web, however, when I tried to put an exit button, even if I copied the AS3 code, it did not work.
This is what I Have
import flash.events.MouseEvent;
import flash.media.Sound;
import flash.system.fscommand;
fscorea_txt.text = score.toString();
playAgainButton.addEventListener(MouseEvent.CLICK, playAgain);
exitBtn.addEventListener(MouseEvent.MOUSE_DOWN, exitGame);
function exitGame(event:MouseEvent):void {
fscommand("quit");
}
function playAgain(event:MouseEvent):void {
playAgainButton.removeEventListener(MouseEvent.CLICK, playAgain);
gotoAndStop(6);
}
I have also tried with MouseEvent.Click, exitGame but it did not work either, however, the other button works perfectly fine
Any help, please?
Thanks!!!!
Unless you are working with Adobe AIR there is no way to "exit" your game with ActionScript. You might be able to have JavaScript unload your object from from the DOM.
Another way you could do this, however, is have a button redirect the user to a different web page that does not have your .swf.
Yes, I am working with Adobe Air, actually, it suppose to be a cell phone game, so I want the user to have the choice to play again or exit the game
Ah okay. With Adobe AIR try using:
NativeApplication.nativeApplication.exit();
instead.
So, you mean this:
exitBtn.addEventListener(MouseEvent.MOUSE_DOWN, exitGame);
function exitGame(event:MouseEvent):void {
NativeApplication.nativeApplication.exit();
}
?
It works perfectly.
Thanks a lot!!!!!!! | https://forums.adobe.com/thread/1160167 | CC-MAIN-2017-30 | refinedweb | 253 | 50.73 |
@cholmes Thanks for the feedback!
Quick piece of background, we are in the process of creating a backup archive of about 110TB of historical aerial imagery and we are looking to store STAC metadata with it.
So I think for now we are looking to start using STAC internally until we can use it in anger for a bit to make sure it meets our needs then we could look at contributing it back as a STAC extension, Even if we have to go and remap all our fields we only have about 1M files so shouldn't take too long!
We will clean up some of our internal docs on what fields we have and where we are planning on storing them.
Another question: namespaces, Land Information New Zealand (LINZ) does a lot more than just historical imagery, we are thinking that our
linz: namespace might get quite polluted do you have any thoughts about subname spaces
Some possibilities we are thinking of
linz_aerial:
linz:aerial or even sub sub namespaces
linz:imagery:aerial | https://gitter.im/SpatioTemporal-Asset-Catalog/Lobby?at=5f96e36857fe0a4f3035b050 | CC-MAIN-2020-50 | refinedweb | 175 | 53.68 |
#include <vga.h>
void vga_setwritepage(int page);
vga_setreadpage(3) allows a different 64K memory chunk to be modified by reads from this memory window. This is a very useful feature for screen to screen pixmap copies. However, not all cards support this in all graphics modes. Check vga_modeinfo(3) for availability. Esp. as this feature can also not be used when in background mode.
Of course, this function must be used in conjunction with vga_setreadpage(3) but may not be used in conjunction with vga_setpage(3).
svgalib(7), vgagl(7), libvga.config(5), fun(6), vgatest(6), vga_getgraphmem(3), vga_claimvideomemory(3), vga_modeinfo(3) vga_setpage(3), vga_setreadpage. | http://www.makelinux.net/man/3/V/vga_setwritepage | CC-MAIN-2014-52 | refinedweb | 106 | 60.01 |
Windows CE kernel and storage technologies and system tools.
Posted by: Sue Loh
The Windows CE Monte Carlo profiler works with support from the BSP. All of our sample BSPs implement the profiler support, but a lot of OEMs seem hesitant to implement it. Perhaps it looks like too much work or is too complicated. Well I'm going to show how to do the easiest possible implementation.
In a nutshell, when you turn on the profiler, the kernel calls a routine in the OAL, OEMProfileTimerEnable(). This routine programs an interrupt to occur at the specified interval. When the interrupt happens, the OAL reports it to the kernel by calling ProfilerHit(). When you disable the profiler, the kernel calls the OAL routine OEMProfileTimerDisable(). So to support profiling, an OAL requires:
Usually we run the profiler at a 200us interval. But guess what? Your BSP already has an interrupt at a 1ms interval. Windows CE requires that already. That's only going to give you one-fifth the number of profiler hits, but it's not so different as to be unusable. So here is an easy profiler implementation.
// Keep track of whether the profiler is enabledBOOL g_IsProfilerEnabled = FALSE;
void OEMProfileTimerEnable (DWORD dwUSec) // dwUSec is ignored here{ g_IsProfilerEnabled = TRUE;}
void OEMProfileTimerDisable (void){ g_IsProfilerEnabled = FALSE;}
UINT32 OEMInterruptHandler (UINT32 ra){ // ... other code ...
if ( <this is a timer IRQ> ) { if (g_IsProfilerEnabled) {#ifdef ARM // This is the code you'd use on an ARM CPU ProfilerHit (ra);#else // This is the code you'd use on non-ARM CPUs ProfilerHit (GetEPC ());#endif } return SYSINTR_RESCHED; }}
The implementations of OEMProfileTimerEnable and OEMProfileTimerDisable are very simple. For the interrupt handler you have to find the cases where you return SYSINTR_RESCHEDULE, and call ProfilerHit there.
What's wrong with this implementation? Well if you implement variable-tick scheduling there are probably some cases where you'll miss data points due to idle, so you might want to skip the variable tick on profiling builds or while the profiler is running. You might want to check OEMIdle too, to make sure things happen the way you need when the profiler is enabled. Otherwise the main problem is you only get 1/5 as much data as normal. Make sure you run your test cases long enough that you can get a statistically significant number of samples. I've seen people try to draw conclusion out of profiler runs that only gathered 500 or so samples. That is far too few. Also, the profiler can badly under- or over-represent OS activity that occurs at a multiple of 1ms, since it will be occurring right on or between every timer interrupt.
I still recommend that you try for the full-blown implementation. Use a more frequent interrupt to get more information, and more accurate information. But at least this description will give you a starting point for understanding how to implement the full version.
Posted by: Sue Loh This material is drawn from a talk that Travis Hobrla gave at MEDC 2006 (thanks Travis!)... | http://blogs.msdn.com/ce_base/archive/2006/01/23/516637.aspx | crawl-002 | refinedweb | 503 | 54.93 |
Data.
An item that is part of a list is called datum, with the
plural being data, but data can also be used for singular. The group of items, or
data, that makes up a list is referred to as a set of data.
Data Set Creation
To support the creation and management of a set of data,
the .NET Framework provides the DataSet class, which is defined in the System.Data
namespace. Therefore, to create a list, you can start by declaring a variable of
type DataSet. To initialize a DataSet variable, the class is equipped with
four.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;");
}
} | http://www.functionx.com/vcsharp/dataset/Lesson01.htm | CC-MAIN-2013-20 | refinedweb | 110 | 68.16 |
Copyright © 2004 W3C® (MIT, ERCIM, Keio), All Rights Reserved. W3C liability, trademark, document use, and software licensing rules apply.
This is an internal working draft.
Last Modified: $Date: 2004/02/27 11:55:44 $
1 Motivation
2 Types of Metadata in XHTML Documents
3 Document Metadata
3.1 Top-level Metadata with meta and QNames
3.2 Statements About Top-level Metadata
3.3 Statements About Other Resources
3.4 String Literals and URIs
4 Markup Metadata
4.1 What Are We Trying To Represent?
4.2 Identifying Resources
4.3 Representing Type
4.4 Representing Value
5 Bibliography
We have two standards running parallel with each other;.
Our intention here is to make more of the information that is contained within HTML-family documents available to RDF tools, but without putting an unnecessary burden on authors familiar with HTML, but not with the subtleties of triples and statements. However, for our discussions on how best to do this, we do need to be familiar with at least the principles of RDF.
RDF is about statements and triples. There are a number of syntaxes which can be used to express these triples, such as N3 and RDF/XML. This document proposes ensuring that the metadata elements already proposed in XHTML 2.0 can be used to glean useful RDF-based information.
There are a number of sets of useful meta information that may be contained in, or relate to, an XHTML document:
There may be information about the document itself; who wrote it and when, the document's subject matter, and so on.
There may be further information about some of this document metadata; not only might we indicate the name of the author, but we might also mark-up metadata to indicate their contact information.
There may be information about words used in the text; an abbreviation may be marked-up with the full text version, or the word "yesterday" may be tagged as being a date with a value of "2004-01-01".
And finally, there may be information about items completely unrelated to the document; there may be meta information about a company or country.
To introduce the issues we'll go through some examples of each of these, using the RDF notion of triples.
metaand QNames
Our first scenario involves statements made about the document. This is to
some extent already catered for with the current use of
meta and
link. Statements can be made about the document, such as its
publication date, author, revision number, subjects, and so on.
An example of using
meta is provided in the latest XHTML 2.0
draft:
<html ... <head> <title>How to complete Memorandum cover sheets</title> <meta name="author">John Doe</meta> <meta name="copyright">© 1997 Acme Corp.</meta> <meta name="keywords">corporate,guidelines,cataloging</meta> <meta name="date">1994-11-06T08:49:37+00:00</meta> </head> ...
Whilst this provides useful metadata about the document, the lack of
qualification of the value in the
name attribute means that in
terms of RDF statements, it is weak. The technique usually used in previous
versions of HTML is to provide a prefix on the property name, or to use the
scheme attribute. Adopting this same technique in XHTML 2.0, our
markup would now look like this:
<meta name="DC.identifier"></meta>
From an RDF standpoint this doesn't help. The statement that the author intends - expressed as RDF triples - would be:
<> <> <> .
An easy way to resolve this would be to allow the author to use QNames
inside
name:
<html xmlns: <head> <title>How to complete Memorandum cover sheets</title> <meta name="dc:creator">John Doe</meta> <meta name="dc:rights">© 1997 Acme Corp.</meta> <meta name="dc:subject">corporate,guidelines,cataloging</meta> <meta name="dc:date">1994-11-06T08:49:37+00:00</meta> </head> ...
However, for reasons that will become clearer later, I would suggest that
we drop the use of
name and call it
property:
<html xmlns: <head> <title>How to complete Memorandum cover sheets</title> <meta property="dc:creator">John Doe</meta> <meta property="dc:rights">© 1997 Acme Corp.</meta> <meta property="dc:subject">corporate,guidelines,cataloging</meta> <meta property="dc:date">1994-11-06T08:49:37+00:00</meta> </head> ...
This would create the following statements:
@prefix dc: <> . <> dc:creator "John Doe" ; dc:rights "© 1997 Acme Corp." ; dc:subject "corporate,guidelines,cataloging" ; dc:date "1994-11-06T08:49:37+00:00" .
Our second scenario concerns statements made about the metadata that we have just created to describe the document. We might want to add further information about the author or publisher, for example.
The simplest way to do this would be to place the additional statements in an external RDF/XML document, and make reference to them. For example, the external reference could, under the current spec, be expressed like this:
<link rel="meta" href="JohnDoe.rdf" />
and the metadata in the document being linked to, might look like this:
<rdf:RDF xmlns: <con:Person rdf: <con:fullName>John Doe</con:fullName> <con:mailbox rdf: </con:Person> </rdf:RDF>
Whilst this makes the metadata available to an RDF/XML parser, this does imply that if this information were to be usable within the XHTML 2.0 browser, then it would need to incorporate an RDF/XML parser. Whilst this is certainly a desirable goal in the long term, it is probably an impractical requirement in the short. Anyway, there are other ways of structuring the data that may give the XHTML 2.0 browser access to the meta information, without requiring it to incorporate an RDF/XML parser.
Unlike previous HTML mark-up, XHTML 2.0 allows
meta to be
nested. The effect of this in RDF terms, is to create an anonymous resource
inside the containing
meta element, and then for nested
meta elements to be further statements about this anonymous
resource. This technique would therefore allow the previous information about
our document's author to be used by the browser, or indeed the author,
without requiring a full-fledged RDF parser. The structure might look
something like this (some statements from the external document are not
reflected):
<html xmlns: <head> <title>How to complete Memorandum cover sheets</title> <meta property="dc:creator"> <meta property="con:fullName">John Doe</meta> </meta> <meta property="dc:rights">© 1997 Acme Corp.</meta> <meta property="dc:subject">corporate,guidelines,cataloging</meta> <meta property="dc:date">1994-11-06T08:49:37+00:00</meta> </head> ...
This would be read in prose as "the document's author is called 'John
Doe'", and in terms of statements that "this document has a dc:creator
property, which is an anonymous object, which in turn has a con:fullName
property, which is 'John Doe'". Note that the implication here is that any
non-contained
meta element is referring to "this document". In
N3 notation it would be:
<> dc:creator [ con:fullName "John Doe" ] .
Sometimes we need to be able to make statements about resources outside of
our document. For example, we may know that the resource identified by has a
con:fullName of
"Tony Blair". Currently the only choices we have are either to embed RDF/XML
inside our XHTML document, or to link to an external RDF file, in the manner
that we saw above.
Since neither of these approches is suitable, I would suggest that we add
a new attribute
aboutthat can appear on the
meta
element. Our new markup might look like this:
<html ...> <head> <title>How to complete Memorandum cover sheets</title> <meta property="dc:creator"> <meta property="con:fullName">John Doe</meta> </meta> <meta about="p:TonyBlair" property="con:fullName">Tony Blair</meta> </head> ...
This time our triples contain one statement about our document (and a statement about that statement), followed by one about some external resource over which we have no control:
<> dc:creator [ con:fullName "John Doe" ] . p:TonyBlair con:fullName "Tony Blair" .
Another consequence of adding
about is that we can now make
statments about other parts of our document. For example, if we know that one
part of the document is actually attributable to someone else, we could use
the following markup:
<html ...> <head> <title>Some quotes</title> <meta about="#q1"> <meta property="dc:source"></meta> </meta> </head> <body> <blockquote id="q1"> > </body> </html>
The triple represented is as follows:
<#q1> dc:source "" .
Although allowing QNames in
property and
about
on
meta gets us a little closer to usable RDF statements, there
is still another problem; RDF allows statements to be made about two types of
objects - resources and string literals - but the syntax that we have built
up so far, only allows the object of a statement to be a string literal. The
consequence is that this:
<meta property="dc:identifier"></meta>
produces the following triples:
<> dc:identifier "" .
As you can see, the object of the statement is a string literal, even though we wanted it to be a resource. The statement we really wanted to make, was this:
<> dc:identifier <> .
Since XHTML 2.0 brings
link with it from HTML to express
relationships between the source document and some other document, then I
would suggest that we simply allow QNames in the
rel attribute
on
link. Our Dublin Core
identifier example would
then be expressed as follows:
<link rel="dc:identifier" href="" />
In addition we would allow
link to appear anywhere that
meta can, since we are saying that the only difference between
them is that one identifies a property of the document and assigns it a
string literal value, whilst the other identifies a property of the document
and assigns it a resource as the value. Our previous example (in which we
indicated the source of a quote), would then become this:
<meta about="#q1"> <link rel="dc:source" href="" /> </meta>
Note that wheras before the triples produced were these:
<#q1> dc:source "" .
now we correctly have these:
<#q1> dc:source <> .
There is no reason why a number of the attributes for
meta
and
link cannot appear on the same element. Our example could
therefore be abbreviated to:
<link about="#q1" rel="dc:source" href="" />
The next category of metadata concerns the qualification of text that appears in a document. This whole area is often discussed on the HTML mailing lists, and is the source of much confusiion. Almost every other day someone proposes new markup to capture the notion of weight, time, addresses, and so on. It's therefore worth looking at what the various proposals are trying to capture.
If we were simply concerned with how a document appeared then we would not need new markup. For example, if we have a corporate website, and on the 'contact us' page the address of the company headquarters is displayed, then the following would be sufficient:
<p class="address"> <span class="street">4 Pear Treet Court</span> <span class="city">London</span> <span class="country">United Kingdom</span> </p>
We could then add a stylesheet, and have the city shown in bold, red, loud, or whatever.
However, an automated process that was analysing this 'address' would be unlikely to derive any 'meaning' from this markup. To 'make sense' of this information, an indexing engine, content management system, e-commerce server, or whatever, would need to know in advance what 'street' and 'city' meant.
Of course, we could achieve this if we all agreed to reserve those words
in the
class attribute to mean only 'parts of an address', but
that would then break with people who want to use 'city' as a style class,
with no relation to its semantic meaning. It would also break for people who
want to use 'address' for an IP address. In other words, we would be imposing
a meaning onto those that don't want it. Far better then to try to create
'unique' versions of these identifiers.
What is usually proposed is to create some elements that capture the notions of addresses. The author's presentational requirements are not compromised since they are still able to style the new markup, and of course the elements capture the semantics:
<style> city { background-color: teal; } </style> <address> <street>4 Pear Treet Court</street> <city>London</city> <country>United Kingdom</country> </address>
Whilst the general approach is fine - the use of elements to clearly mark
metadata - the fact that those elements exist in the XHTML namespace causes a
number of problems. The main one is that it is not clear where we should draw
the line when adding new elements to XHTML; as mentioned earlier, on the
www-html list debates frequently run about whether
date/time/quantity/happiness elements should be added to XHTML. All of them
have worthy arguments for their inclusion, but the bigger issue is whether we
should even bother (since where do we stop), and should we instead focus on
providing mechanisms by which elements from other namespaces can be used?
The second problem with these elements appearing in the XHTML namespace concerns their actual 'meaning'. Since all of these elements exist in the XHTML 2.0 namespace, then to some metadata processor, what we actually have is an 'XHTML date' (or an 'XHTML address', or person, or whatever).
One solution to this might appear to be to use the techniques we described
above -
meta and
link:
<html xmlns: <head> <meta property="con:address"> <meta property="con:street">4 Pear Treet Court</meta> <meta property="con:city">London</meta> <meta property="con:country">United Kingdom</meta> </meta> </head> ...
The statements represented by this are
<> con:address [ con:street "4 Pear Tree Court" ; con:city "London" ; con:country "United Kingdom" ] .
Note however, that this indicates that it is the current document that 'has an address'. This is something that is often overlooked in the discussions about new markup to represent meta information - what exactly is it representing? To return to the type of example most often given:
<style> city { background-color: teal; } </style> <address> <street>4 Pear Treet Court</street> <city>London</city> <country>United Kingdom</country> </address>
what is it that is based in London? And more to the point, what exactly is London. Let's turn to these questions now.
When we are using
meta it is clear that this:
<meta property="dc:creator">John Doe</meta>
means:
"John Doe" is the creator of this document
or:
<> dc:creator "John Doe" .
As we know,
meta has an implied subject of the document that
contains it. But when we say:
<city>London</city>
What is it that 'has' a property of city? We are probably not saying that the document has a property of city, with a value of 'London'. In fact, the only thing that we can be saying is that the text string 'London' has a type of city.
And why might authors want to emphasise this? In the hope that when someone uses some search engine to find documents about 'the city called London', they will find this document, and when they search for documents about Jack London, they won't.
Currently there is no easy way to indicate in a document which of these
two queries you would like your document to repsond to. We could indicate the
subject of a document with
meta, by setting
property to "dc:subject":
<meta property="dc:subject">London</meta>
but there is nothing here to say whether we are writing about Jack London, or the city.
Solving this problem would also help internationalisation, since currently the only options to help French users to find this document would be to mark the document up like this:.
<meta property="dc:subject">London, Londres</meta>
or this:.
<meta xml:London</meta> <meta xml:Londres</meta>
neither of which are very practical.
So, we seem to be stuck; if we start at the top of the document, and say that
"London" is a subject of this document
we don't know if we mean Jack London, or London Town. But if we start at the bottom of the document and say that:
"London" is a type of city
we're none the wiser as to how this piece of information relates to anything else. Using RDF statements again, we can say that the top-down approach is represented by:
<> dc:subject "London" .
and the bottom-up, by:
[ :city "London" ] .
Our task then is to see if we can make the two meet in the middle, and our first step is to solve the problem of searching for 'London' versus 'Londres'.
To re-cap, the issue is that the type of markup that is often proposed, such as:
<city>London</city>
is not much use to us, since we do not know what the information relates
to. In this case, which city is it that has a string literal value of
"London"? The key to this is of course to use a unique value based on URIs.
If we had one defined for the city of London, in the UK, such as, and we were able to attach
this in some way to our text, then we could unambiguously identify what we're
talking about.
I propose therefore, that we add an attribute called
resource, which would allow us to do the following:
<city resource="city:london">London</city>
We have now uniquely identified that the document we're dealing with has a reference to the city of London, in the UK, regardless of the actual text used in the document:
I stayed in the <city resource="city:london">capital</city> for a week. Je voudrais visiter a <city resource="city:london">Londres</city>.
Now a search engine could search for all documents that contain a
reference to, which would
unambiguously identify documents relating to the city of London, in the UK.
And provided that the search server had a mapping between "Londres" and, French speakers could also
find this article. (And the indexing server may have deduced the mapping when
crawling documents.)
As we can see with the use of this technique to make documents available to searches in other languages, it is particularly powerful when the term being searched for is not present in the document. For example, with this document:
Tomorrow the Prime Minister is expected to fly to ...
Searching for "Tony Blair" would not find this article. However, if a
search server knows how to establish from a user that they are actually
searching for, then we could mark up
our document as follows:
Tomorrow the <span resource="p:TonyBlair">Prime Minister</span> is expected to fly to ...
A search for articles relating to Tony Blair would now yield this one, even though his name is not mentioned. This becomes particularly useful when individuals and places are identified in a number of ways, and the lexical description bears no relation to the actual meaning:
Today <span resource="p:DianaSpencer">Lady Diana</span> died in a car crash ... The <span resource="p:DianaSpencer">Princess of Wales</span> was killed today ...
One final example concerns distinguishing between two items of text that are the same, but represent something different. For example:
Yesterday in Parliament the <span resource="p:WinstonChurchill">Prime Minister</span> said that we will fight on the beaches ... Tomorrow the <span resource="p:TonyBlair">Prime Minister</span> is expected to fly to ...
In these two examples the string of characters "Prime Minister" is exactly the same, but they refer to two different individuals. It is now possible to search for "Winston Churchill" and still find the first of these two articles, but not the second. Indeed, where we to search for "Prime Minister", our search server could easily ask us whether we want to search for the words "Prime Minister", or for articles about the current or a previous prime minister.
One last thing on resources, we still haven't addressed where this
information is attached to. The easiest approach is to say that where any of
our new attributes appear on an element that is not
meta, then
there is an implied relationship to the containing resource (or the document)
of "xhtml2:reference". Our previous example therefore results in the
following triples:
<> xh2:reference p:TonyBlair .
This means simply that this document contains a reference to Tony Blair. Note that this is not just a fudge - we are saying that a search for articles and documents written by, reviewed by, criticised by, or summarised by Tony Blair, is very different to seeking out articles and documents that make reference to Tony Blair. This additional property allows us to make that distinction.
Our second issue was to be able to express the type of our
element. I would propose that
property is allowed on any
element, not just
meta. For example:
<span property="con:address"> <span property="con:street">4 Pear Tree Court</span> <span property="con:city">London</span> <span property="con:country">United Kingdom</span> </span>
This now opens up an unlimited supply of descriptors for our text, and so rather than trying to add elements such as address, weight, time and so on we can make use of established taxonomies, or devise our own.
Note also that the presence of
property creates opportunities
for avoiding repetition. For example, we often find ourselves creating our
metadata from information that also appears in the document itself:
<html xmlns: <head> <title>Prime Minister to Fly Out Tomorrow</title> <meta property="dc:creator">John Doe</meta> </meta> </head> <body> <h1>Prime Minister to Fly Out Tomorrow</h1> <span>By John Doe</span> <p> Tomorrow the <span resource="p:TonyBlair">Prime Minister</span> is expected to fly to ... </p> </body> </html>
This often seems like unnecessary duplication, and may add an additional
maintenance headache. However, specifying the
property
explicitly rather than using the default, we can abbreviate this document
to:
<html xmlns: <head> <title>Prime Minister to Fly Out Tomorrow</title> </head> <body> <h1>Prime Minister to Fly Out Tomorrow</h1> <span>By <span property="dc:creator">John Doe</span> </span> <p> Tomorrow the <span resource="p:TonyBlair">Prime Minister</span> is expected to fly to ... </p> </body> </html>
Just as we sometimes need to refer to a resource, so sometimes we need to indicate clearly what the vallue of something is. In this example:
Yesterday in Parliament the <span resource="p:WinstonChurchill">Prime Minister</span> said that we will fight on the beaches ...
we would like to indicate that the speech was made on June 4th, 1940,
since this would allow sophisticated metadata searches to be made. To do this
we use the attributes
val and
datatype:
<span datatype="xsd:date" val="1940-06-04">Yesterday</span> in Parliament the <span resource="p:WinstonChurchill">Prime Minister</span> said that we will fight on the beaches ...
This would create the following triples:
<> xhtml2:reference "1940-06-04"^^xsd:date , p:WinstonChurchill . | http://www.w3.org/MarkUp/2004/02/xhtml-rdf.html | crawl-001 | refinedweb | 3,798 | 50.57 |
import "github.com/grailbio/bigmachine".
bigmachine.go doc.go expvar.go local.go machine.go profile.go status.go supervisor.go system.go
RpcPrefix is the path prefix used to serve RPC requests.
Init initializes bigmachine. It should be called after flag parsing and global setup in bigmachine-based processes. Init is a no-op if the binary is not running as a bigmachine worker; if it is, Init never returns.
RegisterSystem is used by systems implementation to register a system implementation. RegisterSystem registers the implementation with gob, so that instances can be transmitted over the wire. It also registers the provided System instance as a default to use for the name to support bigmachine.Init.
B is a bigmachine instance. Bs are created by Start and, outside of testing situations, there is exactly one per process. }
Dial connects to the machine named by the provided address.
The returned machine is not owned: it is not kept alive as Start does.
HandleDebug registers diagnostic http endpoints on the provided ServeMux.
HandleDebugPrefix registers diagnostic http endpoints on the provided ServeMux under the provided prefix.
IsDriver is true if this is a driver instance (rather than a spawned machine).
Machines returns a snapshot of the current set machines known to this B.
Shutdown tears down resources associated with this B. It should be called by the driver to discard a session, usually in a defer:
b := bigmachine.Start() defer b.Shutdown() // driver code.
System returns this B's System implementation.
A DiskInfo describes system disk usage.
Environ is a machine parameter that amends the process environment of the machine. It is a slice of strings in the form "key=value"; later definitions override earlies ones.
An Expvar is a snapshot of an expvar.
Expvars is a collection of snapshotted expvars.
type Info struct { // Goos and Goarch are the operating system and architectures // as reported by the Go runtime. Goos, Goarch string // Digest is the fingerprint of the currently running binary on the machine. Digest digest.Digest }
Info contains system information about a machine.
LocalInfo returns system information for this process.
A LoadInfo describes system load..
Cancel cancels all pending operations on machine m. The machine is stopped with an error of context.Canceled.
DiskInfo returns the machine's disk usage information.
Err returns a machine's error. Err is only well-defined when the machine is in Stopped state.
Hostname returns the hostname portion of the machine's address.
KeepaliveReplyTimes returns a buffer up to the last numKeepaliveReplyTimes keepalive reply latencies, most recent first.
LoadInfo returns the machine's current load.
MemInfo returns the machine's memory usage information. Go runtime memory stats are read if readMemStats is true.
NextKeepalive returns the time at which the next keepalive request is due.
Owned tells whether this machine was created and is managed by this bigmachine instance.
func (m *Machine) RetryCall(ctx context.Context, serviceMethod string, arg, reply interface{}) error
RetryCall invokes Call, and retries on a temporary error.
State returns the machine's current state.
Wait returns a channel that is closed once the machine reaches the provided state or greater.
type MemInfo struct { System mem.VirtualMemoryStat Runtime runtime.MemStats }
A MemInfo describes system and Go runtime memory usage.
Option is an option that can be provided when starting a new B. It is a function that can modify the b that will be returned by Start.
Name is an option that will name the B. See B.name.
A Param is a machine parameter. Parameters customize machines before the are started.
Services is a machine parameter that specifies the set of services that should be served by the machine. Each machine should have at least one service. Multiple Services parameters may be passed. )
String returns a State's string.
Supervisor is the system service installed on every machine.
StartSupervisor starts a new supervisor based on the provided arguments.).
DiskInfo returns disk usage information on the disk where the temporary directory resides.
func (s *Supervisor) Exec(ctx context.Context, _ struct{}, _ *struct{}) error
Exec reads a new image from its argument and replaces the current process with it. As a consequence, the currently running machine will die. It is up to the caller to manage this interaction.
Expvars returns a snapshot of this machine's expvars.
func (s *Supervisor) GetBinary(ctx context.Context, _ struct{}, rc *io.ReadCloser) error
GetBinary retrieves the last binary uploaded via Setbinary.
Getpid returns the PID of the supervisor process.
Info returns the info struct for this machine..
LoadInfo returns system load information.
MemInfo returns system and Go runtime memory usage information. Go runtime stats are read if readMemStats is true.
Ping replies immediately with the sequence number provided.
func (s *Supervisor) Profile(ctx context.Context, req profileRequest, prof *io.ReadCloser) error
Profile returns the named pprof profile for the current process. The profile is returned in protocol buffer format.
func (s *Supervisor) Profiles(ctx context.Context, _ struct{}, profiles *[]profileStat) error
Profiles returns the set of available profiles and their counts.
func (s *Supervisor) Register(ctx context.Context, svc service, _ *struct{}) error
Register registers a new service with the machine (server) associated with this supervisor. After registration, the service is also initialized if it implements the method
Init(*B) error
Setargs sets the process' arguments. It should be used before Exec in order to invoke the new image with the appropriate arguments.
Setbinary uploads a new binary to replace the current binary when Supervisor.Exec is called. The two calls are separated so that different timeouts can be applied to upload and exec.
Setenv sets the processes' environment. It is applied to newly exec'd images, and should be called before Exec. The provided environment is appended to the default process environment: keys provided here override those that already exist in the environment.
func (s *Supervisor) Shutdown(ctx context.Context, req shutdownRequest, _ *struct{}) error
Shutdown will cause the process to exit asynchronously at a point in the future no sooner than the specified delay..
Local is a System that insantiates machines by creating new processes on the local machine.
Package bigmachine imports 51 packages (graph) and is imported by 8 packages. Updated 2020-08-27. Refresh now. Tools for package owners. | https://godoc.org/github.com/grailbio/bigmachine | CC-MAIN-2020-40 | refinedweb | 1,037 | 52.66 |
Happy Canada Day everyone! I’m back from a fabulous week in California and looking forward to taking a break from giving PowerPoint presentations.
Today, to follow up on my recent series on string concatenation, here’s a fairly easy little puzzle. I have a perfectly ordinary local variable:
string s = "";
Can you come up with some code that parses as a legal expression such that the statements
s = s + your_expression;
and
s += your_expression;
are both legal but produce completely different results in
s? Post your proposals in the comments and I’ll give my answer later this week.
Advertisements
I don’t know if this counts as completely different results, but “s = s + 5 + 5;” results in “55” while “s += 5 + 5;” results in “10”.
Shoot, that was easy…
I said it was easy!
Nice. And to think I was about to get crazy with GetHashCode() calls.
That’s exactly the example I had in mind.
s + 5 + 5 does not fit the structure you required, though. It is ((s + 5) + 5) which does not match the pattern s + exp.
You’ll note that I said that the code should “parse as a legal expression” but I didn’t say in what context it parses as a legal expression. I am tricksy!
I shall interpret your words more carefully!
Out of interest, the .NET 4.5 compiler converted s += 5 + 5 to s += 10 in the IL
“s = s + 5 + 5”
ldstr “”
stloc.0 s = 0
ldloc.0 s
ldc.i4.5
box int32 (object) 5
ldc.i4.5
box int32 (object) 5
call string [mscorlib]System.String::Concat(object, object, object)
“s += 5 + 5”
ldstr “”
stloc.0 s = 0
ldloc.0 s
ldc.i4.s 10
box int32 (object) 10
call string [mscorlib]System.String::Concat(object, object)
Looks like my operator precedence got bested by dtb’s operator associativity, but here’s my answer in 14 characters:
1is int?”a”:””
It only took me this long because ‘is’ isn’t listed on the operator precedence list, and I didn’t think of it for a while.
The solution I immediately thought of was to exploit the low precedence of the ‘as’ operator:
s = s + (object)1 as string; // results in “1”
s += (object)1 as string; // results in “”
I wanted to avoid such parse tree tricks; so I thought maybe we can distinguish the two statements using a custom class that has different user-defined conversions to string based on whether the conversion is explicit or implicit. However, this doesn’t work as the compound assignment only uses an explicit conversion if no implicit conversions exists, and only if the operator is built-in.
None of the built-in + operators have a return type that is explicitly but not implicitly convertible to string, so this semantic difference between the compound assignment and its expanded form cannot be used for a solution.
Neither can we exploit the ‘only evaluated once’ semantics, as ‘s’ is a simple variable.
That leaves no semantic difference between the two, so I conclude that any possible solution must involve some kind of parse tree tricks.
The documentation for += states that:
An expression using the += assignment operator, such as
x += y
is equivalent to
x = x + y
except that x is only evaluated once.
Since evaluation of s is irrelevant, that only leaves parse tree tricks.
That documentation is highly simplified; the rules in the C# specification are more complex.
Try this:
byte b = 0;
b += 1; // OK
b = b + 1; // CS0266: Cannot implicitly convert type ‘int’ to ‘byte’.
‘x += y’ is sometimes equivalent to ‘x = (T)(x + y)’ instead of ‘x = x + y’, and I was checking if that explicit conversion could be exploited to change the behavior.
That’s what I thought too, ambiguity of the ‘+’ operator:
s = s + null == null ? “Null” : “NotNull”;
s += null == null ? “Null” : “NotNull”;
Cute! Particularly nice given the subject of the previous episode.
My solution is equivalent to this.
Your_expression = (s = null).
In the first case, the result is null, whereas in the second it is an empty string.
Are you sure? I can’t repro that.
Ahh, evil type coercion. Visual Basic solved that problem in the 90’s when it introduced separate concatenation and addition operators. I was sorely disappointed when I first learned that C# didn’t learn from that lesson.
Unfortunately VB did not solve the problem; arguably it made it worse. In VB the “&” operator only concatenates strings but the “+” operator still works on strings; in VB, “123” + 123 is 246, but “123” + “123” is “123123”. This can be quite confusing. Also, string concatenation in VB6 and VBScript does not use the “hard vs soft” rule, but comparison does. See my article from 2004 on that subject for details.
VB.NET includes two languages: “Option Strict Off” is a weird goofy language which should never be used except in very limited contexts where dynamic binding is needed, or when it’s necessary to port horrible VB6 code that uses variable types in inconsistent fashion (e.g. a function that returns a string if it succeeds, or a numeric error code if it fails). I don’t know that anybody really likes the “Option Strict Off” language; it certainly shouldn’t have been the default.
In the “Option Strict On” language, which is what any real users of VB.NET are really talking about when discussing the language, (“123″+123) won’t compile, but (“123” & 123) will yield “123123” [as will, incidentally, (123 & 123), since the keyword “And”, rather than the ampersand operator, is used for the Boolean operation.]
null ?? “3”
Nice. This was actually the sort of expression that led me to write this puzzle.
Apart from playing tricks with the compiler you can also use old fashioned side effects:
class A
{
static int count = 0;
public override string ToString()
{
return count++.ToString();
}
}
static void Main(string[] args)
{
string lret = “”;
lret += new A();
lret = lret + new A();
Console.WriteLine(lret);
}
This will print 01 because every call to ToString will result in a different value. I know this is cheating because it has nothing to do with the used operators but it does produce the desired different output.
Seems like a lot of trouble to go to to produce a side-effected expression – there are dozens of ways to do that within the spirit of ‘just add an expression’, without the need to add a ton of boilerplate. Without leaving the System namespace, we can use “Guid.NewGuid()”, “new Random().Next()”, “DateTime.Now”, “Console.CursorLeft++”, “Environment.WorkingSet”…
You are right there are many more ways to do it. Just not DateTime.Now because it has only an accuracy of 15ms which will print (most) likely always the same two values if you call it directly in a row. Stopwatch.GetTimeStamp() would be a better alternative (if executed on recent Intel CPUs with a constant clock rate).
If you want to control exactly what the values for s are in these two scenarios, you can use the following:
class A
{
public static SPlusA operator +(string left, A right)
{
return null;
}
}
class B
{
public static string operator +(A left, B right)
{
return “two”;
}
}
class SPlusA
{
public static string operator +(SPlusA left, B right)
{
return “one”;
}
}
putting whatever you like in place of “one” and “two”.
With these types defined, “your_expression” can add any A to any B, e.g.:
new A() + new B()
Same basic trick that the earlier, simpler answers are using, of course… (Foolishly, when I set out to solve this, for some reason I had it in my head that we needed to be able to pick the results, which is how I ended up with this more complex variation on the theme.)
string str = null;
string s = “”;
s = s + str ?? “Hi”;
s+= str ?? “Hi”;
Sorry the same answer already given by Gilad Naaman…
Pingback: Beware operator precedence « Jim's Random Notes
I cheats… 🙂
string my_expression(string s) {
Thread.Sleep(new Random().Next(5, 10));
return s + DateTime.Now.Ticks;
}
void Main()
{
string s = “”;
s = s + my_expression(s);
s.Dump();
string t = “”;
t += my_expression(t);
t.Dump();
}
If you want an expression that is different every time, “Guid.NewGuid()” is a lot shorter.
Abusing the fact that << binds looser than +, but tighter than +=:
struct S {
public readonly string Value;
public S(string value) {
this.Value = value;
}
public static S operator +(string s, S b) {
return new S(string.Format("(({0}) + ({1}))", s, b.Value));
}
public static string operator <<(S s, int x) {
return new S(string.Format("(({0}) << ({1}))", s.Value, x));
}
public static implicit operator string(S b) {
return b.Value;
}
}
void Main() {
var a = "";
a += new S("_") << 1;
a.Dump(); // "((_) << (1))"
a = "";
a = a + new S("_") << 1;
a.Dump(); // "(((() + (_))) << (1))"
}
String s = “”;
Int32 i = 0;
s = s + ++i;
Console.WriteLine(s); // prints 1
i = 0;
s += ++i;
Console.WriteLine(s); // prints 11
Oops! Ignore. s still had 1 from the prev append. | https://ericlippert.com/2013/07/01/a-string-concatenation-puzzle/ | CC-MAIN-2019-39 | refinedweb | 1,497 | 64.61 |
Welcome to this review of the Pluralsight course Angular 2: Getting Started (Updated).
Introduction to Components
This module builds our first component, the “App component”.
What is a component?
An Angular component is made up of a:
- Template (for view layout, created with HTML, with bindings)
-: `
My 1st Component
`
})
A component is comprised of a template, a class, and metadata. Here the selector is metadata and the template is the value between the back ticks:
This is just beginner example. The back ticks are a new ES 2015 feature for multi-line strings. Instead of hard-coding all of your template html in a string, you can specify an html file to the templateUrl property.
This example contains the code for a component within the main app component:
app.component.ts
import { Component } from 'angular2/core';
import {FooComponent} from 'foo.component';
@Component({
selector: 'pm-app'
template: `
My 1st Component
directives: [FooComponent]
export class AppComponent { }
`
})
Bootstrapping Our App Component
In this lesson, Deborah begins by explaining the index.html file contains the main page for our application.
We have defined our selector in the component definition. This corresponds to the html tag that we use: <pm-app>
We see the other elements that go into our Angular app:
- Systemjs.config.js
- main.ts
- app.module.ts
- app.component.ts
This is a fair amount of work but Deborah says we only need to setup this bootstrapping process once.
Demo: Bootstrapping Our App Component
We create our index.html, including script tags with polyfills for older browsers, and scripts for SystemJS configuration within the head tag.
We run our app by typing “npm start”
index.html
We initially see this message, and then this is replaced when our component is loaded.
Deborah gives an introduction to the browser developer tools here (which you see by pressing F12).
Component Checklist
This is what we need to build a component:
Class -> Code
Decorator -> Metadata
Import what we need from 3rd party libraries, modules or Angular
Deborah provides checklists for each of these stages. She also gives a “Something’s Wrong! Checklist”.
If you are still having problems, there may be a solution on Deborah’s blog. Or you can post on the course discussion page. | https://zombiecodekill.com/2016/12/06/angular-2-getting-started-components/ | CC-MAIN-2022-21 | refinedweb | 371 | 55.95 |
Hello,
I have a user object that gets instantiated / initialized on login.
I have a javascript function that needs to utilize some of the properties of that user object.
In the HTML am trying to figure out how to take that session var / object called currentUser and get the FirstName property.
I tried: myFunction('<%= Session["currentUser"].FirstName %>');
... and I get this error:
CS0117: 'object' does not contain a definition for 'FirstName'
I know that the object exists in the session, the property exists in the object because I can write to a label from the code-behind without any trouble.
What else might be going on here?
tx
k
I think the object should be casted to the real type first before accessing the property. Assuming that the class for user is named UserInfo, then you should do this ((UserInfo)Session["currentUser"]).FirstName. You might need to supply the full name of the class, including the namespace..
Forum Rules
Development Centers
-- Android Development Center
-- Cloud Development Project Center
-- HTML5 Development Center
-- Windows Mobile Development Center | http://forums.devx.com/showthread.php?137570-getting-object-property-from-session-var-in-HTML | CC-MAIN-2015-18 | refinedweb | 175 | 54.32 |
current position：Home>10 minutes to learn how to play excel easily with Python
10 minutes to learn how to play excel easily with Python
2022-02-02 00:46:27 【Dream, killer】
Preface
When you need to be right every day
Excel Do a lot of repetitive operations , If you only rely on manual work, it will waste time , And very boring , Fortunately
Python It provides us with many operations
Excel Module , It can free us from tedious work .
Today I will share with you a quick processing
Excel Module
openpyxl, Its functions are more complete than other modules , Enough to deal with daily problems .
openpyxl install
Enter... Directly at the command prompt .
pip install openpyxl Copy code
Or use watercress image to install .
pip install -i openpyxl Copy code
After successful installation , Let's see how to use
open / Create Workbook
Sample Workbook
Worksheet 【 The first prize 】
Worksheet 【 The second prize 】
(1) Open local workbook
from openpyxl import load_workbook wb = load_workbook(' Winning list .xlsx') Copy code
(2) Create an empty workbook
from openpyxl import Workbook wb1 = Workbook() Copy code
Access worksheet
Create a new worksheet , The insertion position can be specified （
0: First place ,-1: At the end of ）.
'new_sheet', 0) <Worksheet "new_sheet"> Copy codewb.create_sheet(
Get all worksheets in the workbook .
'new_sheet', ' The first prize ', ' The second prize '] Copy codewb.sheetnames [
Use list derivation traversal to get all worksheet names .
for sheet in wb] ['new_sheet', ' The first prize ', ' The second prize '] Copy code[sheet.title
Use
wb[sheetname] To get a worksheet object
' The second prize '] <Worksheet " The second prize "> Copy codewb[
Get activity table ( That is, the worksheet that appears first when you open the workbook ).
" The first prize "> Copy codewb.active <Worksheet
Get worksheet row and column information .
' The first prize '] sheet1.max_column 7 sheet1.max_row 6 Copy codesheet1 = wb[
Get cell information
Access a cell
'D3'] <Cell ' The first prize '.D3> sheet1.cell(row=3, column=4) <Cell ' The first prize '.D3> Copy codesheet1[
If the access unit format is added
value Parameter will modify the value of the current cell .
3, 4).value ' be based on Spark、Python Information extraction and management of medical staff ' sheet1.cell(3, 4, value='Python') <Cell ' The first prize '.D3> sheet1.cell(3, 4).value 'Python' Copy codesheet1.cell(
Get cell value 、 coordinate 、 Row index 、 Column index .
'D3'].value 'Python' sheet1['D3'].coordinate 'D3' sheet1['D3'].row 3 sheet1['D3'].column 4 Copy codesheet1[
Access multiple cells
Use slices to access multiple cells , The slice here is different from the list slice , List slice is Before closed after opening ,Excel The slice in is Front closed and rear closed .
(1) selection
A1:B2 Cells in the range .
'A1':'B2'] ((<Cell ' The first prize '.A1>, <Cell ' The first prize '.B1>), (<Cell ' The first prize '.A2>, <Cell ' The first prize '.B2>)) Copy codesheet1[
Select a single column of data .
'D'] (<Cell ' The first prize '.D1>, ... <Cell ' The first prize '.D6>) Copy codesheet1[
selection B,C Column data .
>>> sheet1['B:C'] ((<Cell ' The first prize '.B1>, ... <Cell ' The first prize '.B6>), (<Cell ' The first prize '.C1>, ... <Cell ' The first prize '.C6>)) Copy code
Select single line data .
3] (<Cell ' The first prize '.A3>, <Cell ' The first prize '.B3>, ... <Cell ' The first prize '.F3>, <Cell ' The first prize '.G3>) Copy codesheet1[
Select the first 2,3 Row data .
2:3] ((<Cell ' The first prize '.A2>, ... <Cell ' The first prize '.G2>), (<Cell ' The first prize '.A3>, ... <Cell ' The first prize '.G3>)) Copy codesheet1[
Traverse to get data
Traverse the specified range by line （
B2:C3） data .
for row in sheet1.iter_rows(min_row=2, max_row=3, min_col=2, max_col=3): for cell in row: print(cell.coordinate) B2 C2 B3 C3 Copy code
Traverse the specified range by column （
B2:C3） data .
for col in sheet1.iter_cols(min_row=2, max_row=3, min_col=2, max_col=3): for cell in col: print(cell.coordinate) B2 B3 C2 C3 Copy code
If
iter_rows()/iter_cols() Specify parameters in
values_only=True, Then only the value of the cell will be returned
Traverse all data by row .
tuple(sheet1.rows) ((<Cell ' The first prize '.A1>, ... <Cell ' The first prize '.G1>), ... ... (<Cell ' The first prize '.A6>, ... <Cell ' The first prize '.G6>)) Copy code
Traverse all data by column .
tuple(sheet1.columns) ((<Cell ' The first prize '.A1>, ... <Cell ' The first prize '.A6>), ... ... (<Cell ' The first prize '.G1>, ... <Cell ' The first prize '.G6>)) Copy code
Modify worksheet
Cell assignment
Add a new column to calculate
author The number of .
for row_index in range(2, sheet1.max_row + 1): sheet1.cell(row_index, 8).value = len(sheet1.cell(row_index, 6).value.split(',')) Copy code
Assign a value to a cell using a formula ,
H7 Count the total number of authors .
'H7'] = '=SUM(H1:H6)' Copy codesheet1[
Append a row of data
Use the list to pass in values in order .
str(n) for n in range(6)]) Copy codesheet1.append([
Use the dictionary to specify
Column index : The column value .
'A':'1','C':'3'}) Copy codesheet1.append({
Insert blank line
Insert a blank line at the specified position ,
idx Row index , Insertion position ;
amount Insert the number of blank lines
2, amount=2) Copy codesheet1.insert_rows(idx=
Delete sheet
'new_sheet']) Copy codewb.remove(wb[
Save workbook
' Winning list V1.xlsx') Copy codewb.save(
Modify the style
typeface
Set up
B2 The cell font format is , Colors can be in hexadecimal color codes .
from openpyxl.styles import Font new_font = Font(name=' Microsoft YaHei ', size=20, color='3333CC', bold=True) sheet1['B2'].font = new_font Copy code
Cell background color
from openpyxl.styles import PatternFill, colors sheet1["A2"].fill = PatternFill("solid", fgColor=colors.BLUE) sheet1["A3"].fill = PatternFill("solid", fgColor='FF66CC') Copy code
Alignment mode
Set up
D2 Data in
Vertical center and
Horizontal center .
from openpyxl.styles import Alignment sheet1['D2'].alignment = Alignment(horizontal='center', vertical='center') Copy code
Row height / Column width
Set the first 2 Line height 40,C Column width is 20.
2].height = 40 sheet1.column_dimensions['C'].width = 20 Copy codesheet1.row_dimensions[
Merge / Split cells
To merge cells, you only need to specify the cell coordinates in the upper left corner and the lower right corner .
'A1:C3') Copy codesheet.merge_cells(
After splitting cells , The value of the merge range is assigned to the upper left cell A1.
'A1:C3') Copy codesheet.unmerge_cells(
For beginners
Python Or want to get started
Python Little buddy , You can go through the bottom Contact the author with a small card , | https://en.pythonmana.com/2022/02/202202020046263710.html | CC-MAIN-2022-27 | refinedweb | 1,074 | 69.68 |
How do I do that? I don't know how to place the text file in the same directory as the java file.
How do I do that? I don't know how to place the text file in the same directory as the java file.
Hello so for this problem i have to read a text file but for some reason it won't let me do it.
this is my code:
import java.io.BufferedReader;
import java.io.FileReader;
import...
my god i really dont know... anyone?
I tried it and still nothing. sigh... i want to die :(
I don't think changing the order does anything. I just tried and nothing. I dont know why the 3 stands are not adding together. what is wrong with the code there?
The problem is i'm not sure why the numbers for hot dog stands 1, 2, and 3 are not adding up. I know what i want to do, i just don't know how to get there.
So how would i fix it then? Do i just delete all the loops then?
Why isn't this adding up?
Ok so i finally got the program to output but the problem is the sum of the output is wrong. This is what i have:
class HotDogStand
{
private int id;
I'm having problems with this one question, i have been at it all day now and i still don't get it.
public class HotDogStand
{
private int id;
private int numSold;
private static int...
I'm using NetBeans too. How did you get that? Did you copy and pasted exactly what I posted and ran it or did you change anything to the code at all?
Oh I forgot to mention I also have this...
How do i do that?
Really? What IDE did you use and what does your output look like?
All it says is this:
run:
BUILD SUCCESSFUL (total time: 2 seconds)
What is it supposed to show?
Ok so I have a problem.
You operate several hot dog stands distributed throughout town. Define a class named HotDogStand that has a member variable for the hot dog stand's ID number and a member...
Can you please elaborate? What println statements and where should they go? Thank you.
No one knows the answer?
Hello, so I am still stumped on this 99 bottles problem. I got most of it working however there are still some issues. This is what i have so far.
public class BottleSong {
private int...
What number of bottles?
There is no error. It's just the wrong code.
Oh ok. Um I just need the output to say:
Ninety Nine bottles of beer on the wall,
Ninety Nine bottles of beer,
Take one down, pass it around,
Ninety eight bottles of beer on the wall.
...
Thank you. Better now?
That's the problem I don't know. I'm new to this forum and I'm not sure how to do it. So you just want me to quote my code I'm guessing?
Ok I tried something completely different. This is my completely different code.
public class BottleSong {
private int bottles =99;
public BottleSong(int n)
{
bottles = n;
}
I thought I already did do that. How do I create an instance of the class and call one of its methods?
So am I supposed to just add this: "System.out.println("\nTake one down, pass it around\n");" after the public static void main(String[] args) { ? I did and when I ran it it outputted Take one down,...
How do I do that? I'm sorry for asking these questions, I'm a beginner with this and really don't know... | http://www.javaprogrammingforums.com/search.php?s=2f1ced3be6d99ce062f1191cbf668bfb&searchid=258905 | CC-MAIN-2016-36 | refinedweb | 621 | 85.79 |
CURLOPT_HTTP09_ALLOWED - Man Page
allow HTTP/0.9 response
Synopsis
#include <curl/curl.h>
CURLcode curl_easy_setopt(CURL *handle, CURLOPT_HTTP09_ALLOWED, long allowed);
Description
Pass the long argument allowed set to 1L to allow HTTP/0.9 responses.
A HTTP/0.9 response is a server response entirely without headers and only a body. You can connect to lots of random TCP services and still get a response that curl might consider to be HTTP/0.9!
Default
curl allowed HTTP/0.9 responses by default before 7.66.0
Since 7.66.0, libcurl requires this option set to 1L to allow HTTP/0.9 responses.
Protocols
HTTP
Example
CURL *curl = curl_easy_init(); if(curl) { CURLcode ret; curl_easy_setopt(curl, CURLOPT_URL, ""); curl_easy_setopt(curl, CURLOPT_HTTP09_ALLOWED, 1L); ret = curl_easy_perform(curl); }
Availability
Option added in 7.64.0, present along with HTTP.
Return Value
Returns CURLE_OK if HTTP is supported, and CURLE_UNKNOWN_OPTION if not.
See Also
CURLOPT_SSLVERSION(3), CURLOPT_HTTP_VERSION(3),
Referenced By
curl_easy_setopt(3), CURLOPT_HTTP_VERSION(3).
November 04, 2020 libcurl 7.78.0 curl_easy_setopt options | https://www.mankier.com/3/CURLOPT_HTTP09_ALLOWED | CC-MAIN-2021-39 | refinedweb | 168 | 53.78 |
interface IBand { int ID {get;set;} string Name {get;set;} void GetStatus(); } public class Band : IBand { public int ID { get; set; } public string Name { get; set; } public Band() { GetStatus(); } public void GetStatus() { ID = 555; Name = "Ring"; Console.WriteLine("Base Class Called"); } } public class ABand : Band, IBand { public string Update { get; set; } public ABand() : base() {} public new void GetStatus() { ID = 655; Name = "TEsst"; Update = "ddd"; Console.WriteLine("Derived class Called"); } } class Program { static void Main(string[] args) { IBand band = new ABand(); band.GetStatus(); } }
We have 2 classes, Band and ABand which implement the IBand interface explicitly.
Now the call to band.GetStatus in Main will display "Derived class called". This will only happen if the Derived class (ABand) also implements the interface IBand explicitly (not in the literal sense in that it does not implement the methods using the IBand.GetStatus syntax)
If the ABand class does not implement the IBand interface explicitly (but implicitly since it inherits from Band which implements IBand) then the above code will output "Base class called"
However note one thing, that when the CTor for ABand is invoked it will invoke the ctor Band class and which will call GetStatus() method from within Band class hence this will output "Base class called" even though the ABand class has implemented the IBand interface explicilty.
Print | posted @ Monday, October 13, 2008 4:24 PM
©
Rohit Gupta
Key West theme by
Robb Allen. | http://geekswithblogs.net/rgupta/archive/2008/10/13/125816.aspx | crawl-002 | refinedweb | 235 | 54.66 |
(Photo by Perfect Snacks on Unsplash)
This time we shall start to see some practical applications of what we saw so far. I'm already excited, aren't you as well?
One note of method: we could build up the final project all at once, but then we will not see much improvements, we will only see what is needed to create the final result, and maybe, something similar. Useful, but not enough.
Instead we will incrementally go over some steps to show the building blocks of our project, and we will play with them. We will take more time to build the project, but we will learn more about our tools and methodologies.
So bear with me, but these tutorials are getting lengthier and lengthier.. and I have split them up already!!!
Part 1: Cleanup the mess and start all over
We start by cleaning up the mess we did so far. However, we will do so by copying the code or forking, the project I prepared as basis for a development environment for you and for me.
Fork it, or clone it, or download the zip and copy the files in your directory, whichever you think it more appropriate.
This is also a template project, so you could really also push the "Use this template" green button
For an easy follow on, I will show the passages to fork and rename the project.
- Fork the project and rename it (or use the "Use this template" function and assign it a name). In this case I called it "yew-weather"
- Open Cargo.toml and rename the field
nameto
name = "yew-weather". I left the
authorsunchanged but you should change it to your name
- Open main.js and change the first line to
import init, { run_app } from "./pkg/yew_weather.js";
- Change in the same file the first line of the
mainfunction to
await init("/pkg/yew_weather_bg.wasm");
Remember: to rename a project in github the field is under the
Settings tab
Now we init the whole project by:
./run
When prompted by the
npm init remember to put the correct name
package name: (yew-devenv) yew-weather version: (0.1.0) keywords: license: (MIT)
You should change the
"repository",
"author", and
"homepage" inside package.json
If all goes well, you should have a sample app running on
Now we can pass to the meatiest part of the tutorial.
Part 2: Maps and weather
We will be writing an interesting app today, I hope. Let's say we want to know what's the weather like. To know this info there are many different apps for the cellphone etc, but let's say we are writing an app for a camping location. The place is by the sea, there are the usual camping amenities, but also a small deck with boats to be rent for the day, and so on. We already built a nice landing page, with some nice pics of the place and a list of amenities, you know, the usual. What is really missing is a real time map to show the winds for sailing and windsurf, because the bay might seem calm, but further on the wind can get naughty...
Would you know how to build the map? An build it using Js and Rust together? How would you go about making it?
There are actually solutions already made for us.
For the base map we will use the free services provided by Mapbox to get the base tiles. More on this later.
For the weather part, there is is an API called OpenWeatherMap, which provides a key and a free tier to try the app. The service we will use is the Weather Map, which is the free-tier way to retrieve TMS tiles.
But what are these tiles that both Mapbox and OpenWeatherMap use? Trying to summarize, when we see a map there are really two implied components: the scale of the map (which is easily understood) and the projection. You know, the world is more like a sphere than a map, so we need to project a round portion to a square image, not easy at all. In time, there have been proposed and used several different types of projection.
(Mercator projection. Source: wikimedia)
To render maps in a website, TMS, or WebTMS, or the like, all adopt one such projection, the web Mercator, created ad hoc for the purpose, that can be used for the web. It is used with some variations by Google Maps and OpenStreetMap as well.
The format takes into account a zoom level, identified with a parameter
z, and then it renders the map as tiles that is, 256X256 pixel images, usually
.tiff,
.png, or
.jpg, identified by a position
x and
y on a Cartesian plane. At zoom level 0 all world is shown in a single tile. At zoom level 1 this tile is divided into four, so you need 4 tiles to represent the whole world, and so on.
The problem in this way, is to reference a point in the latitude, longitude coordinates to a certain zoom level, to know to which tile it belongs. And of course, given a tile, to know from which coordinates it spans, from upper left, to bottom right.
There are several tools to help (I myself have created long time ago a small Rust library to handle the needed conversions).
For example, there is a very useful JavaScript library, leaflet.js, used to visualize tilemaps in this web tile format (called also slippy maps).
Let's solve the problem with leaflet first, and then see how we can improve over it using yew.
However, before we can display our maps, we need to modify a little our Yew project
Code to follow this tutorial
The code has been tagged with the relative tutorial and part, in
yew-weather repo.
git clone cd yew-weather git checkout tags/v8p2
index.html
We are actually creating a separate entry-point for Yew and leaflet, that is we will mount a map in a
<div> and the wasm in another
<div>.
So far we have not seen how to actually achieve that. Indeed the documentation relative to mounting are very scanty, generally speaking they all seem to be happy to mount the output of the wasm directly into the
<body> of the html document... but it doesn't have to be that way We can do better.
So we start by creating two separate entry-points in the index.html
<body> <div id="my_map"></div> <div id="yewapp"></div> </body>
Since we are here, we'll add also the needed "imports" for lealfet in the
<head>
="" integrity="sha512-gZwIG9x3wUXg2hdXF6+rVkLF/0Vi9U8D2Ntg4Ga5I5BZpVkVxlJWbSQtXPSiUTtC0TjtGOmxa1AJPuV0CPthew==" crossorigin=""></script> <script src="/pkg/bundle.js" defer></script> <style> #my_map { height: 400px; } </style>
We added first the CSS for leaflet, then right after, the JS for it. This order is really important!
Then we add also our bundle.js
After that I added a custom style for the map
<div>, through its
id to set an hight. These are my settings, but if you want to give it a width, and center it, go ahead.
For the html page it's all. Let's move on to our lib.rs
src/lib.rs
I put it here in its entirety, then we will discuss the changes:
#![recursion_limit = "256"] mod app; mod components; use wasm_bindgen::prelude::*; #[wasm_bindgen(start)] pub fn run_app() -> Result<(), JsValue> { let window = web_sys::window().expect("no global `window` exists"); let document = window.document().expect("should have a document on window"); let body = document.body().expect("document should have a body"); let children = body.children(); let mounting_div= children.named_item("yewapp").expect("missing element with 'yewapp' id"); yew::App::<app::App>::new().mount(mounting_div); Ok(()) }
First of all we notice that we
use the whole prelude of
wasm_bindgen, and that the directive has changed from
#[wasm_bindgen] to
#[wasm_bindgen(start)].
However, the main changes we need is to walk through the DOM of the document, find the
<div> we need, and implant in it our entry-point.
We do so by first selecting the browser's
window (which is actually an abstraction, not the system's WM window). More references here.
We then select the document, then the body (the actual
<body> tag in index.html). As you can see, we have to walk through the whole DOM tree.
The
body, being an actual Html
Element, has got the
children() method, which we use to select our intended
<div> by its id, using the
named_item() method.
Once we have the selected
<div>, we mount the
yew::App directly to it.
What we were doing up to now, using
yew::start_app, is that we were just mounting the app in the
<body> of the document.
cargo.toml
Of course, we need a little upgrade to cargo.toml to get to use the above code, since the directives we used to walk the DOM are feature-gated in
web_sys
[dependencies] wasm-bindgen = "^0.2" yew = { git = "", features = ["web_sys"] } yewtil = { git = "", features = ["fetch"] } [dependencies.web-sys] version = "0.3.4" features = [ 'Document', 'Element', 'HtmlElement', 'HtmlCollection', 'Node', 'Window', ]
As you can see, we just added a
[dependencies.web-sys] section underneath the other dependencies, stating all the features wee need.
Time to run
Upon running you should see the space left for the map, and the Yew app mounted underneath it (for the screenshot, I actually reduced the
<div> space).
All nice and sweet. But we prepared only the space for the map: now we need to go ahead and get a map there!
Additional tools
We will need to create a
.env (pron. "dot env") file to save our credentials. We will use a rollup plugin to inject the variables we will save in the dot-env file inside the JavaScript source code. This will be done injecting them into the
bundle.js so that they do not need to be read again and again from the server.
If you are using
webpack or other, there are several equivalent packages to achieve the same end.
We will install rollup-plugin-inject-env:
npm i rollup-plugin-inject-env
Then we will modify our rollup.config.js and add also a .env file
rollup.config.js
We just need to add the new plug-in:
import livereload from "rollup-plugin-livereload"; import injectEnv from 'rollup-plugin-inject-env'; export default { input: "main.js", output: { file: "pkg/bundle.js", format: "iife", }, plugins: [ livereload('pkg'), injectEnv() ], };
We are set to use it
.env
Of course, we need a
.env file from where to get the "secrets"
We write it at first this way:
WEATHER_KEY=XXX BASEMAP_KEY=XXX
then, we will replace the
XXX with actual keys
Credentials for the base map
Time to point our browser to Mapbox home page, and get an API key.
Registration is quite easy: there is a 'call to action' button that reads Start mapping for free.
Upon pressing it, we should see a registration form. Follow all the steps.
After confirming the email, it will redirect you to your member area.
Get to the
Tokens page, and create a new Access Token (API key):
Now you can copy the access token and replace the
XXX for the
BASEMAP_KEY in the .env file.
Credentials for the weather map
Now for the weather map
We need to Sign in to openweathermap.org
After filling in your info, registering and confirming the email, you will be redirected for access to your member area.
From there you have to go to the API keys page, and generate a new API key (just enter a name and press the
Generate button)
Once done, you can copy the key and and replace the
XXX for the
WEATHER_KEY in the .env file.
js/map.js
Now we have all we need to create a map.
I will not delve too much into the JS code needed, but you can check out the official leaflet tutorials: they are well done.
We will create a map.js in a folder called JS at the root of the project, and add to it the following code:
var basemap = L.tileLayer('{id}/tiles/{z}/{x}/{y}?access_token={accessToken}', { attribution: 'Map data © <a href="">OpenStreetMap</a> contributors, <a href="">CC-BY-SA</a>, Imagery © <a href="">Mapbox</a>', maxZoom: 18, id: 'mapbox/streets-v11', tileSize: 512, zoomOffset: -1, accessToken: process.env.BASEMAP_KEY }); var weathermap = L.tileLayer('{layer}/{z}/{x}/{y}.png?appid={key}', { attribution: 'Weather data © <a href="openweathermap.org">OpenWeatherMap</a>', layer: 'temp_new', key: process.env.WEATHER_KEY }) var mymap = L.map('my_map', { center: [41.9028, 12.4964], zoom: 6, layers: [basemap, weathermap] }); var baseMaps = { "Terrain": basemap }; var overlayMaps = { "Heat map": weathermap }; L.control.layers(baseMaps, overlayMaps).addTo(mymap); export var map=mymap;
As a quick guide to the code:
- We create first the two layers, the base-map, and the weather map. For that we use the
L.tileLayer()to which we pass the URL of the API (with substitution variables) and a configuration object. In the configuration object we specify how to substitute the variables in the URL. Notice that for the API keys, we use the
process.env.KEYnotation, where
KEYis the name of the key set in the
.envfile. The plug-in rollup-plugin-inject-env has injected them for us as environment variables. Each layer is then saved in its own variable.
- One thing to notice: the URL of the API has many place-holder variables that the configuration object will substitute. However, there are 3 that are present in all tiled map services and that leaflet will produce for us, that is, the current zoom-level
{z}and a the tile coordinates
{x}and
{y}, that will be calculated from the geographic point we are watching, according to the zoom-level we are watching the map at, and how many of these tiles will be needed to fill the view of the map. Lots of math, but it is all done for us by leaflet.
- The next step is to create the actual map with
L.map(), to which we pass the
idof the
<div>where to inject the map, and a configuration object. The location (as latitude/longitude coordinates) and zoom level are set in this configuration object, together with the variables that make up its layers (the two we already created that is).
- For the layers to be correctly shown in the map, we create two objects, one representing the base maps (base maps can be switched one with the others, but only one at a time can be shown: think of them as a background), to which we assign our base-map; the other object representing the overlays of the map (many at a time can be shown: think of them as layers of added information).
- finally we add the layers objects to a
L.control.layers()object and add it to our map.
- we export the map thus composed
main.js
Before we can show our maps, we need to import the js/map.js into our main.js
import init, { run_app } from "./pkg/yew_weather.js"; import "./js/map.js"; async function main() { await init("/pkg/yew_weather_bg.wasm"); run_app(); } main();
Second run
Time to reap the fruits of our labor!
The layer control is top right as default, and we can expand and choose on hover
For now there is not much to choose, but we can exclude the weather layer, which gives a hue to the image according to the temperature.
It does so because we set it to the temperature map, with the variable called
layer: 'temp_new' in the configuration object of the weather layer.
All the weather maps available are shown in this page.
If you want, go ahead and substitute the layer with the name of the layer you want to show, for example
layer: 'clouds_new', for the clouds overlay.
You could also duplicate the
var weathermap and put another overlay, then add it to the controls object in overlays, for example, if you call it
weathermap2:
var overlayMaps = { "Heat map": weathermap, "Cloud map": weathermap2 };
Just remember that the more layers you add to the map, the more calls are made to the API.
Part 3: Stir the cauldron
All we did so far was integrating some JS in our project; but really, what we do have right now is two separate things: a map using JavaScript, and a counter app, using Webassembly made with Yew. We need to mix well the two in the cauldron for Magic to happen.
Code to follow this part
git checkout tags/v8p3
What we will do in this part is to create programmatically with Rust the layers to add to the map.
There are two problems in fact with the map we have created so far:
- Since it is not zoomed in to a precise location this is not so evident, but if you just try zooming to a coordinate at zoom level 12 to 17 (which are more useful to get a whole city, or parts, down to a neighborhood), you will find that the weather map just adds a single hue to the map: that is because it does have data only at city level; also it is not very common that the temperature change much from a neighborhood to the next in the same city, isn't it? so the whole area will be painted with the same color, which is actually more disturbing than useful. At that point, a numeric info on the whole city would be more useful. By the way, down to a certain zoom the tiles from OpenWeatherMap are not shown anymore in many locations.
- We are just fetching and showing data from APIs: we have no idea so far how to manipulate, and get useful information for each of our use-case.
Luckily we have almost all the info we need from our previous tutorials, we just do not know how to apply them.
The following part will be a digression based on theory, because, even if very practical, we will just be making toys. Don't worry: you'll feel the Magic very soon!
What we will do
We willmake Rust communicate with JS through a function and the use of the Session Storage.
You heard it right, we will pass info between Rust and JavaScript through the storage we learned about in Tutorial 4 (even though in that code we used the Local, not the Session storage, but now it makes sense not to want to store permanently our data). Then from Rust we will
wasm_bindgen a JS function that we will invoke to tell the map to update itself using the data we passed through the Storage.
Easy plot, isn't it?
We will use for the data, GeoJSON, a geospatial data interchange format based on JSON.
But, before we start, let's change a little the
<style> in the index.html to give a width to the map and center it:
<style> #my_map { height: 400px; width: 400px; margin: 0 auto; } #yewapp { width: 400px; margin: 0 auto; } </style>
Here we go, much happier, and consuming less tiles from the services :-)
We also added a style for yew, to center it along with the map.
js/map.js
map.js has changed a lot: let me first write it all down, then we will discuss it.
As a help I added lots of comments and whitespace to separate "concepts", so I hope it'll help:
var lat=process.env.LATITUDE; var lng=process.env.LONGITUDE; var mapbox_token = process.env.BASEMAP_KEY; var position = [lat, lng]; var mapboxurl = '{id}/tiles/{z}/{x}/{y}?access_token={accessToken}'; // Mapbox streetmap var basemap = L.tileLayer(mapboxurl, { attribution: 'Map data © <a href="">OpenStreetMap</a> contributors, <a href="">CC-BY-SA</a>, Imagery © <a href="">Mapbox</a>', maxZoom: 18, id: 'mapbox/streets-v11', tileSize: 512, zoomOffset: -1, accessToken: mapbox_token }); // Mapbox satellite var satmap = L.tileLayer(mapboxurl, { attribution: 'Map data © <a href="">OpenStreetMap</a> contributors, <a href="">CC-BY-SA</a>, Imagery © <a href="">Mapbox</a>', maxZoom: 18, id: 'mapbox/satellite-v9', tileSize: 512, zoomOffset: -1, accessToken: mapbox_token }); // Display popup if popupContent property // is present in the GeoJSON feature function onEachFeature(feature, layer) { if (feature.properties && feature.properties.popupContent) { layer.bindPopup(feature.properties.popupContent); } } // Get GeoJSON data from the sessionStorage } // Create a layer for GeoJSON data function get_datalayer () { var geojsonData = get_data(); return L.geoJSON(geojsonData, { onEachFeature: onEachFeature }) } var infolayer = get_datalayer(); // The map var mymap = L.map('my_map', { center: position, zoom: 18, layers: [basemap, satmap, infolayer] }); // Basemaps in Layer Control var baseMaps = { "Satellite": satmap, "Streets": basemap }; // Overlay maps in Layer Control var overlayMap = { "Info": infolayer }; // Layer Control var controls = L.control.layers(baseMaps, overlayMap).addTo(mymap); // Function to redraw the GeoJSON layer, and its control // connected then to an event;
The first thing we do now is to get the position, latitude and longitude coordinates, saved in the
.env file (go ahead and add:
LATITUDE=42.585000 LONGITUDE=14.089444
to the .env). Since the center of the map is of interest both for JS and Rust, I think this is the best idea.
var lat=process.env.LATITUDE; var lng=process.env.LONGITUDE; var mapbox_token = process.env.BASEMAP_KEY;
Since we are in there, getting variables from
.env, we extract also the Mapbox token in its own variable.
var position = [lat, lng]; var mapboxurl = '{id}/tiles/{z}/{x}/{y}?access_token={accessToken}';
We create the position as a array of
[latitude, longitude], and we extract also the Mapbox url, with its variables to fill in as well. This is to make atomic changes to code.
Then we build two Mapbox layers: one for the street view, the other for the satellite view, because... why not? (OK, it's there to show you how to do it, just in case it is needed).
The two layers are almost identical (See the code), except for the names, and a substitution variable:
{id}.
- for
basemapit is
id: 'mapbox/streets-v11'
satmapit is
id: 'mapbox/satellite-v9'
Next, we create a filter function. We will apply it to style the next layer we will build.
function onEachFeature(feature, layer) { if (feature.properties && feature.properties.popupContent) { layer.bindPopup(feature.properties.popupContent); } }
The function checks for the presence of the field "popupContent" in the "properties" of the feature, and attach to the representation of the feature a popup containing the content of
popupContent.
By the way, features are items in GeoJSON to be represented on a map. Later on we'll see some theory on these.
Next we make a function to retrieve the GeoJSON data from the session storage, so we check is there allright, otherwise we return an empty array. }
Remember, both operating in JS or Rust, the session storage works with strings, we have to parse the strings to JSON objects if we want to use them as such.
var geojsonData = JSON.parse(rawGeojsonData);
The line above, inside the function, does just that.
Next we prepare a function that returns a leaflet GeoJSON layer using the data we got from session storage. This function has to be invoked all the times the data in the session storage changes. In fact, we have not easy way to update the layer with GeoJSON objects, other than making diffs on the content. This is time consuming, so the best alternative is to remove each time the GeoJSON layer, parse the data in the session storage, and re-create the GeoJSON layer.
The only drawback of this method is that if we change the data in the session storage from Rust, we need to keep adding to what is there, unless we want to re-draw from the ground up. We are lucky, though: while Leaflet supports the whole GeoJSON specs, it supports also just passing arrays of features, without following too much the conventions. Thus we can just append to an array (Vec in Rust) the objects we want to draw, and transfer it in the session storage, and we are set to go.
// Create a layer for geoJSON data function get_datalayer () { var geojsonData = get_data(); return L.geoJSON(geojsonData, { onEachFeature: onEachFeature }) } var infolayer = get_datalayer();
Notice how we style the content with our filter, scanning for needed pop-ups.
We also create right away a layer ready to be added, even if we think that the storage would be empty.
In fact there can happen that the user refreshes the page, or navigates back and forth between pages, and the data will still be present in the storage, as long as the browser window, or tab, is not closed. For a more permanent storage, we should use the local storage though.
Then we create the map, and attach to it all the layers (refer to the above code).
After this, we create two objects containing layers: one for the base maps, that can be interchanged with each other, as we have seen, the other one for the overlay; in this case we deleted the weather map (don't worry, we will use the service again), and we are left with the GeoJSON layer only.
The last function we prepare is for a hook:;
This function, in sequence, removes the GeoJSON data layer from the control, removes it from the map, and re-creates it from the data; after this, re-attaches it first to the map, and then to the controller.
The function will be fired by the hook we create next.
These hooks must be attached to an event, but the
'submit' event is not in the official hooks. We use this, because in this way it will not be fired unintentionally by an event in the browser (the map is attached to a div, not a form, that normally fires the submit event).
As usual we export the map.
JS/wasm_bridge.js
We will use a separate JS module for
wasm_bindgen. The main reason is that it generates error to call the map.js both from JavaScript and Rust, because the map will be initialized two times. In fact calling
map.jst with
wasm_bindgen effectively imports again the same module from the Rust part, creating duplicates. Thus we need a separate module to store our bridge function.
The content of this module is pretty small though:
export function update_map() { window.mymap.fire('submit'); };
We impose to the map to fire the event
submit whenever the function
update_map() is called.
Notice that we call the map from the
window namespace; we will export it there in the main.js file.
main.js
In main.js we import the two new modules we created:
import init, { run_app } from "./pkg/yew_weather.js"; import "./js/wasm_bridge.js"; import mymap from "./js/map.js"; async function main() { await init("/pkg/yew_weather_bg.wasm"); run_app(); } // Export the Leaflet map window.mymap = mymap; main();
After this, we export
mymap to the
window, as we said, for an easier access. (when debugging with the console is also easily accessible, which is a pro, really!)
We have finished with the JavaScript. Let's go with Rust now
Cargo.toml
We add 4 crates for this part:
serde = "1.0" serde_json = "1.0" rand = { version = "0.7", features = ["wasm-bindgen"] } load-dotenv = "0.1.1"
Of course we will need to serialize and de-serialize GeoJSON, which is a form of JSON, so the first two entries were expected.
Rand is there to generate random numbers. Notice the feature
wasm-bindgen. Thumbs up for this crate that made the necessary adjustments to operate in wasm conditions!
There is a crate called dotenv to work with
.env in Rust. However, the intended use case for it is to load the .env file at run time. This means that the compiled wasm code will try to access the .env file in the environment of the browser. Apart from the various sandboxing restrictions, it would be useless to send over to the client, together with the wasm and js files, also the .env file. Moreover, we will need the info at compile time: if we needed them at run time, we would need to act differently, maybe taking advantages of cookies or the session storage.
For these reasons, we will use load-dotenv, a wrapper for dotenv. If I understand correctly, what it does is that it actually exposes a procedural macro to let us get the variables with
std::env!, but it loads them at compile time. It works exactly the same way as rollup-plugin-inject-env does for the JS counterpart.
src/data/mod.rs
We will create a new mod to handle the data for the project, so we cretae a data/ folder in src/ and we create in it a mod.rs. In it we will expose the entities we need.
pub mod geojson;
So now we need to create a geojson.rs inside data/ to interface our programs with GeoJSON.
src/data/geojson.rs
A little premise here: there is already a crate called geojson that does absolutely what it promises.
However, I find it formative to go through the steps to replicate an interface from scratch, especially one so simple as GeoJSON. In this case also, it's necessary that we understand well the data we are dealing with, so that we can know how to produce them or manipulate them. Should the project require more serious usage of GeoJSON, we will need to use the ready-made crate.
Moreover, we will not use the whole specs: as we have seen, leaflet supports passing just an array of features to draw, with no further ceremony. Thus we will need just two GeoJSON entities: Feature and Geometry.
We'll introduce them, and explain them as we go.
use serde::{Deserialize, Serialize}; use serde_json::{Value, Map};
Since it is a schema based on JSON, we'll use serde and serde_json.
This is a
Feature:
#[derive(Serialize, Deserialize, Debug, Clone)] pub struct Feature { pub r#type: String, pub properties: Option<Value>, pub geometry: Option<Geometry>, }
A GeoJSON Feature is just a JSON object, with a field called
type, that is a string whose value is
"Feature". Notice that we have to escape the word
type which is reserved in Rust. So in code we use the
r# prefix; when serializing to json, serde will automagically write the field as type instead.
A Feature can have an optional field called
properties, where we can have a JSON object with whatever data attached to it, as we want: a Feature is a way to connect a geographical entity with some metadata on it, and this is done through
properties.
The third field of a Feature is
geometry, and stores a
Geometry, the geographical entity to represent on a map.
impl Feature { pub fn new() -> Self { Feature { r#type: "Feature".to_string(), properties: None, geometry: None, } } pub fn add_property(&mut self, key: String, value: Value) { match &mut self.properties{ Some(v) => { v.as_object_mut().unwrap().insert(key, value); }, None => { let mut v = Map::new(); v.insert(key, value); let v: Value = v.into(); self.properties = Some(v); } }; } pub fn add_geomerty(&mut self, geometry: Option<Geometry>) { self.geometry = geometry; } }
Of course, the first thing that we impl is a
new() "constructor".
Then we impl a method to add a single property (key, value) to the properties field of the Feature, and one to add a whole
Geometry.
Of course we could add also the properties as en entire object, but then, this is a lightweight interface, otherwise we would have used the "official" crate geojson. In any case, each field of the struct is public, so we can always create a struct, coerce to a
serde_json::Value and assign it to a Feature's property, and go our merry way.
This is a
Geometry:
#[derive(Serialize, Deserialize, Debug, Clone)] pub struct Geometry { pub r#type: String, pub coordinates: Value, }
There are just two fields: a
type, for which we escape as for the
Feature, and the coordinates, which takes a JSON array, and gives the needed coordinates to build the geographic entity.
There are seven types of Geometry. but so far we will implement only a point, that has got a single array containing a longitude, a latitude, and optionally a third number with an elevation of the point
impl Geometry { pub fn new() -> Self { Geometry { r#type: "".to_string(), coordinates: Value::Null, } } pub fn new_point(coordinates: Value) -> Self { Geometry { r#type: "Point".to_string(), coordinates, } } }
Each geometry must have it own
type string, specifying the kind of Geometry it is; of course for a point, we need to mark the geometry
type as
"Point".
That said, we could already impl a constructor for each of the seven geometries, but for the sake of this tutorial a
Point is sufficient. We will implement some others very soon.
src/lib.rs
We will pass on to see the Yew app, but first we need to add the
mod data; to our src/lib.rs
mod app; mod components; mod data; use wasm_bindgen::prelude::*;
All the rest did not change, we are still mounting the yew app on its own
<div>.
src/app.rs
The app has undergone substantial rewrite and extension, so we will take our time to analyze it.
use crate::components::button::Button; use crate::data::geojson::*; use yew::prelude::*; use yew::format::Json; use yew::services::storage::Area; use yew::services::StorageService; use serde_json::Value; use wasm_bindgen::prelude::*; use rand::prelude::*; use rand::rngs::ThreadRng; use load_dotenv::load_dotenv;
Impressive list of things we need to
use!
We import first the
Button component, and the
geojson we just created.
For Yew, besides the prelude, we need the
Json format, and the imports to use the SessionStorage (both the Service and the storage area).
We need the prelude of the
wasm_bindgen as well, to call the JavaScript functions. From
rand we need both the prelude and the type
ThreadRng.
Lastly we need also the
load_dotenv, which we will use very soon:
const GEOJSON_KEY: &'static str = "geojsonData"; load_dotenv!();
In fact, we define here the key corresponding to the same one we used in the map.js to interchange our data. Then we invoke the procedural macro
load_dotenv!. From now on, we can access the variables inside the
.env file with
env!().
Next, we bind the JS function with the FFI:
#[wasm_bindgen(module = "/js/wasm_bridge.js")] extern "C" { fn update_map(); }
Notice how we specify where to find the module needed for the bind with
#[wasm_bindgen(module = "/js/wasm_bridge.js")].
The Msg has not changed at all:
pub enum Msg { AddOne, RemoveOne, }
while the struct App has increased considerably:
pub struct App { link: ComponentLink<Self>, counter: i32, storage: StorageService, geo_data: Vec<Feature>, position: Vec<f64>, rng: ThreadRng, }
After the
ComponentLink, as usual, and the
counter, we add the
StorageService, a
Vec for the
Features aptly called
geo_data, which will be used to store the features before transfering to the Session Storage, the position (as a Vec of
f64; we could have used a tuple as well), and the
ThreadRng that will be used by
rand to access the random number generator.
Now we can impl our App! We will analyze carefully the
create() function first:
impl Component for App { type Message = Msg; type Properties = (); fn create(_: Self::Properties, link: ComponentLink<Self>) -> Self { // Watchout! New: Now it returns a Result let storage = StorageService::new(Area::Session).expect("storage was disabled by the user"); let Json(geo_data) = storage.restore(GEOJSON_KEY); let geo_data = geo_data.unwrap_or_else(|_| Vec::new());
The first thing we do is to access the storage service, and restore its content to the
geo_data variable, just as we learned in the tutorial 4. However, since then the things have changed, and now
StorageService::new() returns a
Result. If you are following the tutorial 4 you should not have any problem, since we were using there an older version of Yew. But now we are using the new one, so we need to
expect or
unwrap the Return.
One thing for the use of browsers' dev tools: in case of
panic, Firefox shows in the console just that the
unreachable has been executed (wasm way to declare a panic). Chrome's console instead unwinds it a bit, so you can understand clearly that it is indeed a panic. However, in both cases, writing something through
expect does not have a clear advantage. Things have changed a little recently, and there is a way to take a peek at our
expects that I might show you very soon. In any case it's a good practice to write down our
expect as usual.
After this, we initialize the random generator "thread", and then we retrieve the center coordinates of the map, and we prepare them into a position:
let rng = thread_rng(); let lat = env!("LATITUDE","Cound not find LATITUDE in .env"); let lng = env!("LONGITUDE", "Cound not find LONGITUDE in .env"); let lat: f64 = str2f64(lat); let lng: f64 = str2f64(lng); // Longitude first! geoJSON and Leaflet take opposite conventions! let position = vec!(lng, lat); App { link: link, counter: 0, storage, geo_data, position, rng, } }
We use
str2f64 a small function I use to convert strings to f64. I put this function at the end of the file:
fn str2f64(s: &str) -> f64 { s.trim() .parse() .expect("Failed parsing a String to f64") }
This is one of the small functions to have handy as a Rust programmer, so that you remember to trim before parsing...
Before we go on, we have to notice that the GeoJSON standard interprets the first number in a position as longitude, while leaflet interprets the first as latitude. However, leaflet will interpret it correctly when importing GeoJSON.
Now we will take a look at the
update() function:
fn update(&mut self, msg: Self::Message) -> ShouldRender { match msg { Msg::AddOne => { self.counter += 1; let position: Vec<f64> = self.position.clone().into_iter() .map(|x: f64| { let d: f64 = self.rng.gen_range(0.00001, 0.0003); if random() { return x-d; } x+d }).collect(); let position: Value = position.into(); let point = Geometry::new_point(position); let mut feat = Feature::new(); feat.add_geomerty(Some(point)); feat.add_property("popupContent".into(), self.counter.to_string().into()); self.geo_data.push(feat); self.storage.store(GEOJSON_KEY, Json(&self.geo_data)); update_map(); } Msg::RemoveOne => { self.counter -= if self.counter == 0 { 0 } else { 1 }; let _ = self.geo_data.pop(); self.storage.store(GEOJSON_KEY, Json(&self.geo_data)); update_map(); } } true }
The first thing the
Msg::AddOne does is to increase the counter, as usual.
Then we make it clone the position and modify it, creating for each of the coordinates a random coefficient
d, between 0.00001 and 0.0003 (which is suitable for the zoom-level we are in now, 18).
To create a random number in a range (a, b) we use
rng.gen_range(a, b). After this we use
random() which is a convenience template function from the
rand prelude, to generate a
bool, by just slapping it after a
if:
if takes a
bool, so
random() will toss the coin for us: if
true the coefficient
d gets subtracted from the coordinate, otherwise its gets added.
In this way we obtain random positions nearby the map center. We coerce the new position into a JSON Value (an array, coming from a Vec), and we create a new Point with
Geometry::new_point, passing to it the position just created.
We then create a new feature and pass to it as geometry the one we just created, and we add a property with key
popupContent and as value a string containing the number in the counter. As we know when we will add the GeoJSON data as a layer we will style each feature with a filter that attaches to it a popup with the content taken from the value of the property
popupContent, if present.
We add the feature to the Vec of features in the
self.geo_data of the App structure.
We then sore the
geo_data in the Session Storage, and we call the JS function to update the map.
The
Msg::RemoveOne just decreases the counter, as well as calling
pop() on the
geo_data Vec. After this, it too synchronizes the Session Sotrage and calls a redraw of the map through the JS function.
That's it! The most is done.
We could leave all the rest as is, except for a little detail
fn change(&mut self, _props: Self::Properties) -> ShouldRender { false } fn view(&self) -> Html { html! { <> <Button onsignal=self.link.callback(|_| Msg::RemoveOne) <Button onsignal=self.link.callback(|_| Msg::AddOne) </> } } }
change() hasn't changed, we still need to return
false.
Instead we will take a look at the
view() function: we took out the
<h1> and wrapped the two buttons in a
<> and
</>. These are needed as a root for the DOM to be injected in the html, but in fact they will disappear once injected as an entity. Yet they are needed as the unique entry-point required by
html!.
As you can see in this image of Firefox inspector, once you run the app, the two buttons are injected inside the
<div>.
Let's roll
Upon running it and playing a little adding buttons and clicking on the positions markers:
Also moving to the sat view:
The black tile on the upper right corner is there because for the higher zoom-levels Mapobx does not have the sea tiles, so it renders them as black tiles. Zooming back we can see that the sea tiles are restored.
In the dev tools, we can see the session storage holding the GeoJSON of our data layer:
In the above image however, I excluded the data layer, just to show it is possible.
Conclusions
This is just the first part on this project, and it is already packed up with stuff.
I don't know if I should explain longer on the JavaScript part: I tried to balance the fact that we need it, with the fact that it is a series on Rust and Yew, not JS... but still I wanted to explain a little, not to throw code at you with no explanation, other than "trust me dude, it does work this way" (that is really a condescending attitude for me, a no-go).
I wanted to make a tutorial for each practical project, but writing it down I realized that it is just not possible: too many concepts, even if they are not totally new. The sheer length of this is scaring me for the proofreading already! Maybe I should have split it in three? Let me know what do you think of this format, and also how do you feel about this series, for those who are reading it: are you satisfied with the format, do you have any recommendation or request? Feedback is really appreciated.
Thank you for reading up to here, and stand ready and excited for the conclusion of this project in the next tutorial.
Discussion
Davide, I've cloned your yew-deven as a template and am now trying to modify it by adding a module I've created in rust to access a database with odbc. I got the db program to work independent from yew-deven and am now dropping it into yew-deven. After making all of the needed changes to incorporate it, I run it and get two warnings: Unresolved dependencies and Missing global variable name. Since the yew-deven ran without any Warnings or Errors prior to adding the new odbc module I suspect the problem lies somewhere with my new rust module. Any thoughts?
Can you show me your Cargo.toml and the actual error?
The first thing it comes to mind is to check whether the odbc driver interface crate you are using is compatible with webassembly and to which degree
Cargo.toml (dependencies)
[dependencies]
wasm-bindgen = "^0.2"
yew = { git = "github.com/yewstack/yew", features = ["web_sys"] }
yewtil = { git = "github.com/yewstack/yew", features = ["fetch", "pure"] }
odbc = "0.17.0"
odbc-safe = "0.5.0"
env_logger = "0.7.1"
log = "0.4.0"
[dependencies.web-sys]
version = "0.3.4"
features = [
'Document',
'Element',
'HtmlElement',
'HtmlCollection',
'Node',
'Window',
]
error msgs:
Serving "./index.html"
Listening on 0.0.0.0:8080/
main.js → pkg/bundle.js...
LiveReload enabled
(!) Unresolved dependencies
rollupjs.org/guide/en/#warning-tre...
env (imported by pkg/yew_template.js)
(!) Missing global variable name
Use output.globals to specify browser global variable names corresponding to external modules
env (guessing '__wbg_star0')
created pkg/bundle.js in 190ms
thanks
In reality this is a rollup.js error. It is searching for 'env'.
Did you install rollup-plugin-inject-env?
In case the command is:
npm i rollup-plugin-inject-env
At any rate, are you using odbc and odbc-crate compiled for wasm or as a serverside API? Since odbc wraps around a C driver I do not think it will compile to wasm. In case you succeed let me know 😃
Hi Davide, Thanks for all your help. This is looking pretty hopeless so far. What I'm looking to do is have a browser based app that accesses a local SQL Server running inside a Docker image on a Mac.I was hoping to access the db directly with out going through an API. I have one I created in React but was hoping for more efficiency by accessing the db directly.
Well something can be done... Warning: it is a method prone to malicious attacks if exposed.
Here's the gist of it: you are going to access a local db with a local webpage, right? So there's less concern for protection (less in not none)
Anyways, odbc has got a SOAP interface:
docs.microsoft.com/en-us/previous-...
You might access it with the soap crate in Rust or with javascript (better solution).
It is a hacky trick, but it should work. But it takes a lot of effort. An API wrapper can be hardened for security, and you could interface it with JSON which is more universal and simple than SOAP...
Another way, which is useful if all the structure is in the same machine, is a wrapper to odbc that saves data to Redis, then to use the Redis server to exchange data with the webpage... Also lot of work, but you can prefetch and manipulate data before storing it to Redis. The advantage of having an API is that all the pieces are there, plus if tomorrow you want to interface other databases, local or remote, it's much easier to add. Also, I do not know your needs, but once you provide an interface, needs always arises to make some manipulation of data serverside before presenting it to the final user's interface...
Davide, thanks for your reply and all of your effort. I think since I already have an api built in React that connects to MS SQL db, I'll just continue to use it and get Rust to communicate with it and then send the data needed by the React frontend. Again, thanks for your efforts and knowledge.
Hey Davide, this is really helping me - please continue. I don't mind if I have to read it twice.
It's coming, it's coming... but it will be a longer article than this (I know, I should chop it more, but I wanted to stick together things that belong together). An to continue, I got tons of ideas... just the hours of the day are too few :-( | https://practicaldev-herokuapp-com.global.ssl.fastly.net/davidedelpapa/yew-tutorial-07-dr-ferris-i-presume-web-geography-injected-with-rust-p-i-57g6 | CC-MAIN-2021-04 | refinedweb | 7,906 | 62.88 |
I'm trying to build a soundboard that plays samples on a button press. I have the sounds playing and stopping on the button press, however, I cant for the life of me figure out how to get the sounds to play on a continuous loop until the button is pressed again.
I've search google and the forums and figure it has something to do with the hold_repeat function on gpiozero but I don't have much coding experience and cant work out where to fit it in the code.
The code I'm using is:
Taken from here ... -music-box
Code: Select all
import pygame.mixer from pygame.mixer import Sound from gpiozero import Button from signal import pause pygame.mixer.init() sound_pins = { 2: Sound("samples/sound1.wav"), 3: Sound("samples/sound2.wav"), } print "pedal ready" buttons = [Button(pin) for pin in sound_pins] for button in buttons: sound = sound_pins[button.pin.number] button.when_pressed = sound.play button.when_released = sound.stop pause()
Any help would be greatly appreciated. | https://www.raspberrypi.org/forums/viewtopic.php?p=1422881 | CC-MAIN-2019-51 | refinedweb | 169 | 66.44 |
Hello,
trying to get some code converted from Visual C++ 6 to gcc.
I've got pretty far, but as a C++ newbie I'm little bit lost with templates.
It would be very nice if somebody could help me to get this working (with
explanation please so I can learn).
A possibility would also be to convert this template to a normal function
(if it would help to get this working)
Here is the template (table.h):
Thanks in advanceThanks in advanceCode:#ifndef _TABLE_H #define _TABLE_H #include <string> #include <vector> using namespace std; template <class T> class table: public vector<T> { public: T operator[] (const string& key); }; template <class T> T table<T>::operator[] (const string& key) { for (reverse_iterator i = rbegin(); i != rend(); i++) if (*i) if ((*i)->name ^ key) return (T)((**i)()); throw out_of_range("no elements matching key"); } #endif
efgee | https://cboard.cprogramming.com/cplusplus-programming/94187-convert-visualcplusplus-template-gcc.html | CC-MAIN-2018-05 | refinedweb | 142 | 57.61 |
Matplotlib is the "grandfather" data visualisation tool for data in Python - it offers unparalleled control over graphs and diagrams for data, and lets us annotate and customise figures to our heart's content. Matplotlib is built upon for other important modules we'll use later, such as Seaborn, which is more used for statistical visualisation.
All the documentation for Matplotlib can be found here
We can create quick and dirty graphs using the functional method as we've seen before. This method is simpler but won't allow us to customise our plots as much.
First, we need to import our modules:
import matplotlib.pyplot as plt import numpy as np
If we're running in a notebook, we can run the following line of code to save us from writing plt.show() to give us our graphs. Spyder automatically shows plots, so we don't need this line of code and everything will still work as normal.
%matplotlib inline
We can then graph any data we want using the plt.plot command. Here we will use the np.linspace function to generate a numpy array called x with 101 equally spaced points between 0 and 10, and another numpy array called y that is simply x squared. (Remember numpy arrays can be operated on and the operation will apply element-wise)
x = np.linspace(0,10,101) y = x**2
We can then plot the graph of x versus y by using the plt.plot() function:
plt.plot(x,y)
[<matplotlib.lines.Line2D at 0x1554ba8ed68>]
Great! From here we could add more plots, linestyles, x labels, y labels, titles and more using various methods:
plt.plot(x,x**2,x,x**3) plt.xlabel("x") plt.ylabel("y") plt.title("Graph of x squared and x cubed")
Text(0.5,1,'Graph of x squared and x cubed')
We can also create multiple plots on one output. These are called subplots, and to do this we use the plt.subplot() function - this takes 3 arguments - the number of rows, the number of columns, and then the plot number we're refering to. This may seem like an ugly way to do things, and we'll see a better way to do it later with the object orientated method.
plt.subplot(1,2,1) plt.plot(x,x**2) plt.subplot(1,2,2) plt.plot(x**2,x,'g')
[<matplotlib.lines.Line2D at 0x1554bfc94a8>]
The problem with the functional method is that it's a bit of a mess: Commands can be a bit hard to work with and understand at a glance, and problems often happen because of ambiguity in code. Python works on an "object orientetated" philosophy so it makes sense for us to use a similar methodology for graphing. The object orientated way of creating plots works by creating figure objects and calling methods on it.
To start, we need to create a figure:
fig = plt.figure()
<matplotlib.figure.Figure at 0x1554bbf1f98>
Figures are a blank canvas for us to work on. To add axis, we use the add.axes() method. We can always look at our figure by just running it as a line of code:
axes = fig.add_axes([0,0,1,1]) fig
Techinically, the "add_axes" method takes in one arguement - a list. This list has 4 elements - the x position with respect to the left of the axis (in percent), the y position with respect to the bottom of the axis (also in percent) - so here, 0,0 just means we don't want white space - and width and height (so here, 1,1 means width 1, height 1).
Now if we want to plot something on these axis, we call the "plot" method with respect to the axes:
axes.plot(x,y) fig
We can also add x labels, y lables, and more using the following methods:
axes.set_xlabel('X') axes.set_ylabel('Y') axes.set_title('x vs x squared') fig
Now let's see why the object orientated way of creating graphs is so much more powerful:
To start, we can create graphs within graphs:
fig = plt.figure() axes1 = fig.add_axes([0,0,1,1]) axes2 = fig.add_axes([0.1,0.6,0.4,0.3]) #Creating axes within our axes axes1.plot(x,y) axes2.plot(y,x) axes1.set_xlabel('X') axes1.set_ylabel('Y') axes1.set_title('x vs x squared') axes2.set_xlabel('X') axes2.set_ylabel('Y') axes2.set_title('Inverse')
Text(0.5,1,'Inverse')
We can also create subplots as before using a shortcut:
fig, axes = plt.subplots()
This combines the "fig" and "axes" call together assuming we want the large plot by using tuple unpacking. We can add arguements to the subplot method to create more plots:
fig, axes = plt.subplots(2,2)
Notice that all the numbers are a little bit squished together - we can fix this using the plt.tight_layout() function:
fig, axes = plt.subplots(2,2) plt.tight_layout()
Great! For now, to learn how to access these axes let's work with a 2x1 subplot grid.
If we call the "axes" object after using the subplot command, we can see that "axes" is just a list of axes objects:
fig, axes = plt.subplots(1,2)
axes #Notice the square brackets? It's a list (technically a numpy array)!
array([<matplotlib.axes._subplots.AxesSubplot object at 0x000001554D39DA58>, <matplotlib.axes._subplots.AxesSubplot object at 0x000001554D3E0B70>], dtype=object)
This means we can iterate over it, but more importantly, we can select what axes we want by using indexing:
axes[0].plot(x,x) axes[1].plot(x,x**2) fig
Say we don't want to use two axes and we'd rather use one set of axes with two lines. We already know how to do this:
fig, ax = plt.subplots() ax.plot(x,x) ax.plot(x,x**2) ax.plot(x,x**3)
[<matplotlib.lines.Line2D at 0x1554d4e09b0>]
How can we tell at a glance which plot is which? A legend would be helpful here - to do this we need to edit out code a little bit:
fig, ax = plt.subplots() ax.plot(x,x, label='Linear') ax.plot(x,x**2, label='Squared') ax.plot(x,x**3, label='Cubed') ax.legend()
<matplotlib.legend.Legend at 0x1554d541e10>
Notice how we need to label our plots within their respective methods, and then call the legend method. A similar method is used to change the color and linestyles of these plots:
fig, ax = plt.subplots() ax.plot(x,x, label='Linear', color = "purple", lw=3,ls=':') ax.plot(x,x**2, label='Squared', color = "pink", lw =3,ls='-.') ax.plot(x,x**3, label='Cubed', color = "blue", lw=3,ls='--') ax.legend()
<matplotlib.legend.Legend at 0x1554d5bbe10>
There are hundreds of different plot options out there, as well as the option for 3d plots, contour plots, log scaling and more! To check these out, take a look at the documentation here
Finally, it's worth noting that we can use the object orientated method for scatter graphs, boxplots and histograms.
To start, let's take a scatter plot looking at a (albeit fake) positive correlation.
fig, axes = plt.subplots() randomData = 0.4 * np.random.randn(101) + 0.5 #Nice way to fake a correlation - 0.4 is the slope, 0.5 is the intercept. axes.scatter(x,randomData + x)
<matplotlib.collections.PathCollection at 0x1554d62bd30>
Next, let's look at how to create a histogram representing a normal distribution:
fig, axes = plt.subplots() data = np.random.randn(1001) axes.hist(data)
(array([ 1., 15., 76., 156., 249., 267., 152., 61., 20., 4.]), array([-3.38632307, -2.71377809, -2.04123311, -1.36868813, -0.69614314, -0.02359816, 0.64894682, 1.3214918 , 1.99403678, 2.66658176, 3.33912675]), <a list of 10 Patch objects>)
And finally, let's look at boxplots:
fig, axes = plt.subplots() data = [np.random.normal(0,1,100),np.random.normal(0,2,100),np.random.normal(0,3,100)] axes.boxplot(data)
{'boxes': [<matplotlib.lines.Line2D at 0x1554c2757f0>, <matplotlib.lines.Line2D at 0x1554c2c3ef0>, <matplotlib.lines.Line2D at 0x1554c2c8160>], 'caps': [<matplotlib.lines.Line2D at 0x1554c275a90>, <matplotlib.lines.Line2D at 0x1554c2c38d0>, <matplotlib.lines.Line2D at 0x1554c1e66d8>, <matplotlib.lines.Line2D at 0x1554c2c8fd0>, <matplotlib.lines.Line2D at 0x1554c29d518>, <matplotlib.lines.Line2D at 0x1554c1ae6a0>], 'fliers': [<matplotlib.lines.Line2D at 0x1554c2c3588>, <matplotlib.lines.Line2D at 0x1554c2c8ba8>, <matplotlib.lines.Line2D at 0x1554c1ae748>], 'means': [], 'medians': [<matplotlib.lines.Line2D at 0x1554c2c34e0>, <matplotlib.lines.Line2D at 0x1554c2c8e48>, <matplotlib.lines.Line2D at 0x1554c1aefd0>], 'whiskers': [<matplotlib.lines.Line2D at 0x1554c275f28>, <matplotlib.lines.Line2D at 0x1554c2751d0>, <matplotlib.lines.Line2D at 0x1554c1e6470>, <matplotlib.lines.Line2D at 0x1554c1e6c50>, <matplotlib.lines.Line2D at 0x1554c29dcf8>, <matplotlib.lines.Line2D at 0x1554c29d198>]}
Notice how the boxplot and histogram methods also output a lot of other useful information about the plots to access later if we want to. If we don't want these, we can just run fig again in another cell, or omit %matplotlib inline:
fig
We're going to create a graph that looks at what happens to the graph of ex, 2x, 3x, x2 and x3. Soon we will be able to import data so we're not working the mathematical functions all the time!
import matplotlib.pyplot as plt import numpy as np %matplotlib inline x = np.linspace(0,10,101) fig = plt.figure() axes = fig.add_axes([0,0,1,1]) axes.plot(x, np.exp(x), label = "Exponential") axes.plot(x, 2**x, label = "2 Power") axes.plot(x, 3**x, label = "3 Power") axes.plot(x, x**2, label = "Squared") axes.plot(x, x**3, label = "Cubed") axes.set_ylim((0,1000)) axes.legend()
<matplotlib.legend.Legend at 0x1554c1a08d0>
Try graphing the sin(), cos() and tan() functions on one graph, adding a legend, title and axis. Then, try using the subplot command to put them all on different plots. | https://nbviewer.jupyter.org/github/HiPyLiv/HiPyProject/blob/master/%5BVISUALISATION%5D%20A%20Deeper%20Look%20Into%20Matplotlib/A%20Deeper%20Look%20Into%20Matplotlib.ipynb | CC-MAIN-2019-09 | refinedweb | 1,628 | 59.4 |
The information in this post is out of date.
EF6 RTM is now available.
Visit msdn.com/data/ef for the latest information on current and past releases of EF.
A couple of months back we released EF6 Alpha 2, since then we’ve been adding new features, polishing existing features and fixing bugs. Today we are pleased to announce the availability of Alpha 3. EF6 is being developed in an open source code base on CodePlex, see our open source announcement for more details.
We Want Your Feedback
You can help us make EF6 a great release by providing feedback and suggestions. You can provide feedback by commenting on this post, commenting on the feature specifications linked below or starting a discussion on our CodePlex site.
Support
This is a preview of features that will be available in future releases and is designed to allow you to provide feedback on the design of these features. It is not intended or licensed for use in production. The APIs and functionality included in Alpha 3 are likely to change prior to the final release of EF6.
If you need assistance using the new features, please post questions on Stack Overflow using the entity-framework tag.
Getting Started with Alpha 3
The Get It page provides instructions for installing the latest pre-release version of Entity Framework.
Note: In some cases you may need to update your EF5 code to work with EF6, see Updating Applications to use EF6.
What’s Changed Since Alpha 2
The following features and changes have been implemented since Alpha 2:
- Code First Mapping to Insert/Update/Delete Stored Procedures is now supported. The feature specification on our CodePlex site provides examples of using this new feature. This feature is still being implemented and does not include full Migrations support in Alpha 3.
- Connection Resiliency enables automatic recovery from transient connection failures. The feature specification on our CodePlex site shows how to enable this feature and how to create your own retry policies.
- We accepted a pull request from iceclow that allows you to create custom migrations operations and process them in a custom migrations SQL generator. This blog post provides an example of using this new feature.
- We accepted a pull request from UnaiZorrilla to provide a pluggable pluralization & singularization service.
- The new DbContext.Database.UseTransaction and DbContext.Database.BeginTransaction APIs enable scenarios where you need to manage your own transactions.
What Else is New in EF6
The following features and changes are included in Alpha 3 but have not changed significantly since Alpha 2:
- Async Query and Save adds support for the task-based asynchronous patterns that were introduced in .NET 4.5. We’ve put together a walkthrough that demonstrates this new feature. You can also view the feature specification on our CodePlex site for more detailed information.
- Custom Code First Conventions allow write your own conventions to help avoid repetitive configuration. We provide a simple API for lightweight conventions as well as some more complex building blocks to allow you to author more complicated conventions. We have a walkthough for this feature and the feature specification is on our CodePlex site.
- Dependency Resolution introduces support for the Service Locator pattern and we’ve factored out some pieces of functionality that can be replaced with custom implementations. The feature specification provides details about this pattern, and we’ve put together a list of services that can be injected.
- Code-Based Configuration – Configuration has traditionally been specified in a config file, EF6 also gives you the option of performing configuration in code. We’ve put together an overview with some examples and there is a feature specification with more details.
- Configurable Migrations History Table – Some database providers require the appropriate data types etc. to be specified for the Migrations History table to work correctly. The feature specification provides details about how to do this in EF6.
- Multiple Contexts per Database – In previous versions of EF you were limited to one Code First model per database when using Migrations or when Code First automatically created the database for you, this limitation is now removed. If you want to know more about how we enabled this, check out the feature specification on CodePlex.
- Updated Provider Model – In previous versions of EF some of the core components were a part of the .NET Framework. In EF6 we’ve moved all these components into our NuGet package, allowing us to develop and deliver more features in a shorter time frame. This move required some changes to our provider model. We’ve created a document that details the changes required by providers to support EF6, and provided a list of providers that we are aware of with EF6 support.
-.
- DbContext can now be created with a DbConnection that is already opened. Find out more about this change on the related work item on our CodePlex site.
- Improved performance of Enumerable.Contains in LINQ queries. Find out more about this change on the related work item on our CodePlex site.
- Default transaction isolation level is changed to READ_COMMITTED_SNAPSHOT for databases created using Code First, potentially allowing for more scalability and fewer deadlocks. Find out more about this change on the related work item on our CodePlex site.
- We accepted a pull request from AlirezaHaghshenas that provides significantly improved warm up time (view generation), especially for large models. View the discussion about this change on our CodePlex site for more information. We’re also working on some other changes to further improve warm up time.
- We accepted a pull request from UnaiZorrilla that adds a DbModelBuilder.Configurations.AddFromAssembly method. If you are using configuration classes with the Code First Fluent API, this method allows you to easily add all configuration classes defined in an assembly.
What’s after Alpha 3
Alpha 3 contains all the major features we are planning to implement for the runtime in the EF6 release. We’ll now turn our attention to polishing and completing these new features, implementing small improvements, fixing bugs and everything else to make EF6 a great release. We’re still accepting pull requests too.
We’ve also been getting the EF Designer code base updated for the EF6 release and we hope to have a preview of the EF6 designer available soon.
If you want to try out changes we’ve made since the last official pre-release, you can use the latest signed nightly build. You can also check out our Feature Specifications and Design Meeting Notes to stay up to date with what our team is working on.
Congrats,
You guys are a fantastic team.
Unai
And as for this bug has been resolved:
Store update, insert, or delete statement affected an unexpected number of rows (0). Entities may have been modified or deleted since entities were loaded. Refresh ObjectStateManager entries.
"Alpha 3 contains all the major features we are planning to implement for the runtime in the EF6 release"
Dies this mean there is no code first support for stored procedures and tvf as specified in the roadmap?
I just see, you removed the entries from the road map in the newest version. Too bad, that's the feature I wait most for, because without it I won't be able to introduce code first here. :-
Nice work, but there is no simple solution for bulk insert. I use entity framework to read data, and write small things. But to insert more data i use binding query… and i loose a part of the power of entity framework (detach, attach is a bit borring)
@Martin – We broke up stored procedure/function mapping into some more manageable sized features. Insert/Update/Delete was the highest priority since there was no way to do it (whereas you can select using SPROCs/TVFs using raw SQL via SqlQuery).
We do still have a few smaller features that we plan to do in the EF6 release (but haven't definitely committed to) – entityframework.codeplex.com/…/advanced. At the moment we are going to try and at least get function imports supported for Code First (entityframework.codeplex.com/…/818).
@dfiad77pro – We have a couple of features that we want to implement (though not in the EF6 release) around batch operations.
– Batching up insert/update/delete operations rather than sending individual commands entityframework.codeplex.com/…/53
– Bulk/set operations without loading data from the database entityframework.codeplex.com/…/52
Please add your votes to those… and feel free to comment with any scenarios we should take into account when we get to the features.
Great progress. However, (and I don't mean to sound like an ingrate…BUT…) support for arbitrary stored procedures is really important and it is a huge disappointment to see that it didn't make it in.
I'm also disappointed that function imports has been cancelled.
Anyway I like the EF Code-First approach.
Rowan,
Is there any solution for this? Store update, insert, or delete statement affected an unexpected number of rows (0). Entities may have been modified or deleted since entities were loaded. Refresh ObjectStateManager entries.
@CraigAJohnson & @ben – We are still trying to squeeze support in, but it's not a definite for EF6. One of the things that makes it lower priority is how easy it is to do with SqlQuery – db.Blogs.SqlQuery("dbo.GetBlogsWithTag @p0", "entity-framework").
@Veranildo Veras – There are many scenarios that can cause this error, if you are seeing it in a situation that you didn't in earlier versions of EF (or a situation that it shouldn't be occurring) can you file a new issue – entityframework.codeplex.com/…/Create. Be sure to include a code sample that reproduces the issue.
@Rowan Miller – thanks for the answer!
Yes, stored procedures are no problems to integrate, but TVF (e.g. fulltext search) is ugly to write in raw sql where there are many conditional fiters.
// Filter example:
if (countryId.HasValue)
query = query.Where(x => x.CountryId= countryId);
if (categoryId.HasValue)
query = query.Where(x => x.CategoryId = categoryId);
if (query != null)
{
query = from x in MyDbContext.ContainsTableWrapper(query)
join y in query on x.Key equals y.Key
select y;
}
// And so on….
Does anyone know if there are plans to improve insert/update performance? I noticed that EF seems to not use prepared commands for inserts/updates. This surprised me. I think this would help a lot with regard to bulk insert/update speed. It would also be great if there was a way to clear the DbContext of items to keep memory usage down.
What happens to System.Data.Entity.Design assembly?
How is it going to get released from now on?
I'm asking this because we build custom a model generator based on the design assemblies and we cannot migrate until they are in sync.
@Rowan Thank you for your response. 🙂 I'm looking forward to your progress. Keep up the good work!
Thanks for the great work you are doing!
One of the biggest problem we have right now is the startup time with a big model: 1100+ tables. With EF 6 aplha 2 the first query is taking about 25 sec. This is a great improvement over EF5 where without pre-generated views takes a lot more (tens of minutes!).
However it would be nice if EF6 offered API for pre-generated views. Any future plans on this?
It would greatly reduce those 25 sec on startup.
With EF 5 we are generating the views using Stystem.Data.Entity.Design/EntityViewGenerator class which calls System.Data.Entity.dll/System.Data.Mapping.StorageMappingItemCollection <== this was now improved and moved to EF 6. This means that EntityViewGenerator can no longer be used for EF6.
What would be nice is to also "move" the generators from Stystem.Data.Entity.Design to EF.
Thanks again!
@Osu – we have a work item for adding support for generating views entityframework.codeplex.com/…/369. We also have discussed this at our design meeting – you can find notes here: entityframework.codeplex.com/wikipage. I also created a T4 template for generating views for EF6 – you can find my blog post here: blog.3d-logic.com/…/entity-framework-6-and-pre-generated-views. The template is on VS Gallery (note I created it for EF6 Alpha1 and have not tried it with EF6 Alpha3 but hopefully it will work)
@Catalin Pop – We aren't planning to move System.Data.Entity.Design into the out-of-band releases (the EF5 version will remain in the framework though).
We are moving over bits of the functionality from System.Data.Entity.Design into EF6 so it would be great to here more about you scenario and which APIs you are using. You can either reply here or start a discussion on our CodePlex site – entityframework.codeplex.com/…/create.
@ben – You are right, TVFs are pretty ugly. We'll try and get support into EF6 (but we aren't definitely committed to it).
@Jon – We don't have any major improvements planned for EF6. We have talked about prepared commands before but we didn't have an item tracking it on CodePlex, so I created one – entityframework.codeplex.com/…/912. We also have a post-EF6 item to look at batching up statements during SaveChanges.
Hi mr Miller.
I appreciate you for the release 🙂
For the comment that you noticed for how to implement EF TVF usage in code-first, i don't actually
have an idea how to implement a Fully Linq-Enabled, Queryable (and/or Queryable<T>) implementation with SqlQuery<T>.
we need a Queryable<T> method to call our TVfs and need to pass parameters to those.
even, imagine if we could define a SelectSource and a TableSource for a POCO object in the [Table] attribute above that POCO, to distinguish between the physical source and View source of the POCO.
this will make object-population (materialization) very flexible and we also could define
some POCO fields as Databse Generated, so that they are populated from View-Like queries.
i mean fileds that are populated from fields in the query (or view) that are retrived via a join operetion
in the view/TVF.
I ensure This will have an impressive, and will soon be a developer-popular feature that whould never be in any other ORMs 🙂
Who will be better than EF, but EF ?! 😉
we are waiting for EF6, and we are counting for the features (or at least TVFs with Full Linq and Queryable support) i mentioned.
Thanks a lot mr Miller. Thanks for your support (Y)
Massoud Safari
@Rowan Thanks for creating the issue for prepared commands. I have seen the issue for batching previously. I was surprised that people were looking into that without first looking into prepared commands. It seems like prepared commands would be an obvious way to improve performance and easier to implement. It seems like there really needs to be a way of clearing the tracked objects as well. The only way I'm aware of to do this is create a new DbContext which seems to have a lot of overhead.
@Jon: We have been working on the problem you mentioned about the costs associated to create a new instance of DbContext. This improvement will be present in future versions of EF6, but it didn't make it in time for the alpha 3 build. You'll see a much lighter DbContext initialization time if you use the nightly build or the latest code from our git repository. For more details about this performance improvement refer to entityframework.codeplex.com/…/828.
You can implement the WhereOr for building dynamc query white EF6.
PLEASE !
@SeniorSafari – Being able to compose a query on the results of a TVF isn't supported by SqlQuery. The feature you are after is entityframework.codeplex.com/…/818 which we are going to try and get into EF6 (but it's not a definite at this stage).
@neoncyber – Could you provide some more information about the feature you are after. An example of the query you would like to write would be ideal.
Thanks mr Miller for the response to my comment.
I hope the feature will be added in the next release of EF6.
Sincerely
SeniorSafari
@Rowan Miller – Thanks for your reponse
My Example
var result = MyDbContext.Customers.Where(c => c.Name == "Toto");
if (boolIsTrue)
{
result = result.Where(c => SqlExpression.Or(c.Name == "Tata"));
}
And traduct in SQL with this :
If bool is true :
SELECT * FROM Customers WHERE Name == "Toto" OR Name == "Tata";
else
SELECT * FROM Customers WHERE Name == "Toto";
This is equivalent of Dynamic Linq but not use String or LinqKit with PredicatBuilder
but is not default include in EF.
@neoncyber – As you mentioned there are a few frameworks that help you build an expression that can be passed into the Where clause. Some folks on our team have thought about what something like this might look like built into EF. However, at this stage it's a lower priority because there are existing frameworks that will help you do it.
@Rowan Miller
To answer your question regarding System.Data.Entity.Design, our scenario is the following:
Environment:
– 1800+ Tables
– 20+ Schemas
– 3200+ Foreign Keys (Cross Schema)
(We can't use the designer due to sheer number of entities. As to why they are so many of them it's a separate discussion, but the simple explanation is that the business domain model is uncommonly large when modeled using 3NF)
Model Generation process:
– Use EntityStoreSchemaGenerator to generate SSDL
– Use EntityModelSchemaGenerator to generate CSDL
– Generate Mappings (MSL)
– Save EDMX in memory buffer.
– Read Metadata for entire Database model from a separate database
– Load the EDMX in a custom class model and rewrite it to achive the following:
– Remove a set of excluded columns (exluded using Metadata)
– Remove all undesired OneToMany relationships (based on Metadata; avoids having thousand of unwanted member collections in the generated code, IE all tables have foreign keys to the User table: remove all reverse relationships from user to all tables in the system as they are not needed and add a heavy performance tax when loading a related user)
– Rename navigation members based on foreign key name (IE: for Parent_ID: Person1 to Parent, for Neighbor_ID: Person2 to Neighbor, etc.)
– Perform other renames in the model (metadata driven)
– Mark entities with EDM metadata specifying what interfaces should be added by code generator to entities,
IE: IEntity(ID,Name), IAuditable(InsertDate, EditDate, InsertUser_ID, EditUser_ID), ILifeCycle(LifceCycle_ID), etc…
– Add EDM metadata to entities and properties for a set of validation attributes to be added to the model (IE: Readonly, Required, StringLenght, DisplayName, Visibility, UniqueConstraint, and many more very specific)
– Save modified EDMX.
– Generate C# classes for entities and 6 different contexts using a T4 generator.
– Generate pregenerated views using EntityViewGenerator.
Our biggest problems are:
– CSDL generation takes about 30 minutes;
– Creating Pregenerated Views takes about 45 minutes (post model cleaning, much more if all the relationships are kept).
Therefore until we will be able to programatically manipulate the EDMX and generate pregenerated views we cannot take full advantage of EF6 improvements.
With kind regards,
Catalin Pop
@David Thanks for the info. On an unrelated note, I noticed that EF now automatically creates indexes on foreign keys. Thanks for that. From what I remember, it didn't do that originally. It's nice to have those created automatically.
"CSDL generation takes about 30 minutes;"
Correction SSDL generation takes about 30 minutes.
Catalin Pop
Hi,
EF6 sounds very appealing and we are thinking of adopting it in our project even before the release. However from our tests it looks like there is a performance degradation from EF5 that is noticeable on first query execution (increase from 7 to 10 seconds ) and for certain queries (0.14 to 0.7 seconds).
We tested both EF by executing the same queries over the same model (a large one, 1100+ tables) and with pre-generated views.
Are there any known reasons for this? Can we expect some performance tuning before EF6 release?
@Pawel – I've tested your T4 views generator and it does work with alpha 3. Thanks.
Thanks
@Osu: there is a known performance issue with the alpha 3 release of EF6, as documented in entityframework.codeplex.com/…/902. The issue has been resolved and some additional improvements have been added to EF6 after the alpha 3 build was released.
Would it be possible for you to grab a nightly build and run your tests again? I'm curious to see if the performance issue persists, or if it is something we missed.
You can get the nightly build by following these instructions: entityframework.codeplex.com/wikipage
@Osu: Glad to hear that view gen templatate works with EF Alpha3. Thanks!
Hi,
a lack of functionality in CodeFest FileTable or FileStream a new version of EntityFramework 6
Hello,
There seems to be a major bug in EF6 alpha that is not present in EF5.
On short: Inner join generated where an outer join is expected. This happens when a join is made from T1 to T2 and then to T3 if the relation T1-T2 is optional but T2-T3 is mandatory.
More details here: entityframework.codeplex.com/…/960
This is currently keeping us from adopting EF6. Any chances of having this fixed soon?
Thanks!
@Osu: thanks for bringing the issue to our attention. We are investigating.
Loving the update so far as we've had some pretty brutal speed issues in EF 4.1 on .Net 4.0. I can't migrate the project to .Net 4.5 yet so I was excited when I saw that EF 6 added support for .Net 4.0. This might seem like a goofy question, but how does one go about adding enums to a model-first designer that was created in an old version of EF? I'd really prefer NOT to redo the entire EDMX as we modified objects to be better worded than the original underlying database tables; including primary keys. We have several templates built off these modifications and with a large number of tables it would be time consuming to update all of these with a new EDMX. Is it possible to add enums to existing EDMX files through the designer?
I should also mention that I'm still in VS 2010.
@Ryan H. EF6 does support all 3 schema versions supported by previous versions of EF. So you should be able to use your edmx file without changes. However if you generate code from your edmx this code won't work with EF6 without changes. Also, note that the designer in VS2010 did not support v3 schema so you won't be able to add enums using VS2010. (The designer in VS2010 supported only schema versions v2 and v1 while enums where introduced in v3). If you want enums on .NET Framework 4 in an existing edmx file you would need to update your edmx to v3 of the schema. You can do it manually by changing xml namespaces or you could try using VS2012 to do it for you (you would have to create a project targeting .NET Framework 4.5 and add the edmx file to the project and then the designer should give you the option to move to v3 (btw. v3 is backwards compatible with v2)). Unfortunately the designer in VS2012 does not support EF6. While you can somehow make it work (I blogged about this here blog.3d-logic.com/…/entity-framework-6-and-modeldatabase-first) it will probably not work for projects targeting .NET Framework 4. This means that in most cases you would have to edit your edmx separately from your project (or even manually) but then you may have problems with making all the customizations you have work. Just to let you know – we are working on a new version of the designer that will support EF6 which should solve these problems. Stay tuned.
@Catalin Pop Sever – For EF6 you will continue to be able to use the version of System.Data.Entity.Design that was included in .NET 4.5. This works because we haven't had to make and changes to the EDMX format to support new features.
@David – the time execution improved for some queries but the first query execution (start-up) is still taking longer than in EF5 (~10 seconds vs 7). Pre-generated views were used in both cases.
Thanks.javascript:WebForm_DoPostBackWithOptions(new WebForm_PostBackOptions("ctl00$content$ctl00$w_107850$ctl01$ctl00$ctl00$ctl05$bpCommentForm$ctl05$btnSubmit", "", true, "BlogPostCommentForm-ctl00_content_ctl00_w_107850_ctl01_ctl00", "", false, true))
and to this date still no built in second level caching …
SOS
There is a bug in this version:
Enum field types are being duplicated with a '1' after the original field name ! what should I do?!?
@Arash Masir: Can you create an issue on the Entity Framework codeplex site: entityframework.codeplex.com ? Make sure to provide a repro. | https://blogs.msdn.microsoft.com/adonet/2013/02/27/ef6-alpha-3-available-on-nuget/ | CC-MAIN-2019-35 | refinedweb | 4,156 | 55.13 |
Synopsis
Utility routines for ChIPS (CIAO contributed package)
Syntax
import chips_contrib.utils from chips_contrib.utils import * The module provides the following routines: xlabel(text) ylabel(text) title(text) xlog() ylog() xylog() xlin() ylin() xylin() xlim() ylim() xylim() add_ds9_contours()
Description
The chips_contrib.utils module provides utility routines for CIAO users, and is provided as part of the CIAO contributed scripts package.
Please see "ahelp add_ds9_contours" for help on the add_ds9_contours() routine; the other rountines are described below.
The module can be loaded into ChIPS, Sherpa, or Python scripts by saying one of the following:
from chips_contrib.utils import * from chips_contrib.all import *
where the second line will load in all the ChIPS contributed routines, not just the utils module.
Contents
The utils module currenly provides the routines:
The xlabel, ylabel and title routines
All three routines take a single argument, a string containing the text for the axis label. Use "" to remove the label. If the label contains LaTeX symbols that begin with a "\" character then prepend the letter r to the string. Examples of use are:
chips> xlabel("Energy (keV)") chips> ylabel(r"photon/cm^2/s/\AA") chips> title(r"Light curve: mean rate = 2.3 \pm 0.2 count s^{-1}")
These commands are short forms for the ChIPS set_plot_xlabel(), set_plot_ylabel() and set_plot_title() routines.
The xlog, ylog, and xylog routines
These routines take no arguments and change the current axis - or axes - to display with a logarithmic scale. They are short forms for the ChIPS log_scale() routine, and examples of use are show below:
chips> xlog() chips> ylog() chips> xylog()
The xlin, ylin, and xylin routines
These routines take no arguments and change the current axis - or axes - to display with a linear scale. They are short forms for the ChIPS lin_scale() routine, and examples of use are show below:
chips> xlin() chips> ylin() chips> xylin()
The xlim, ylim, and xylim routines
These routines take up to two arguments and change the limits of the current axis or axes. They are short forms for the ChIPS limits() routine, and examples of use are show below:
chips> xlim(10,100) chips> xlim(AUTO,100) chips> xlim(10,AUTO) chips> xlim(AUTO,AUTO) chips> ylim(1e-6,1e-2) chips> xylim(1, 100)
The xylim() command can only be used to change both axes to display the same range; the ChIPS limits() command should be used if the ranges for the X and Y axes are different.
Changes in the October 2011 Release
New routine
The add_ds9_contours() routine was added to the module..
Bugs
See the bugs pages on the CIAO website for an up-to-date listing of known bugs.
See Also
- limits
- limits, panto, pick_limits, zoom | https://cxc.cfa.harvard.edu/chips/ahelp/chips_utils.html | CC-MAIN-2020-10 | refinedweb | 448 | 59.94 |
Note: I'm aware another thread similar to this has been posted, however that one only describes mainly about configuring windows firewall. This guide gives more of a breif overview and a list of other things you can do to IMPROVE your protection with windows firewall.
With the release of SP2 for Windows XP, it came with a rather primitive firewall that the majority of the public classifies as horrible. However, it is still being used today by several people due to a few basic reasons:
-Extremely simplistic/easy to use.
-Very low memory usage.
-Comes with SP2 (on Windows XP) for free.
-Better than no protection at all.
Well if you really hate having those complex firewalls that consume a lot of resources and boggle your mind with configurations, here is a guide for you! I've done a bit of researching on a few relatively affective methods of "strengthening Windows XP's firewall", which basically involves the combination of a few simple programs and configurations. This guide is basically designed to give people a frew simple steps to setting things up to improve their defenses with Windows XP's Firewall. (Note: sorry if it seems like an advertisement [because it's not!! :mad: ] the links provided are used as resources.)
1.) Configuring Windows Firewall:
Seeing as how Windows Firewall was made to be simplistic, not a lot can be done with it. If you are planning to tightly secure your system with Windows Firewall, an easy thing to do is to simply turn it on and also check the "Don't allow exclusions" box (via Control Panel ->Windows Firewall). However, using that feature is quite blunt since you are very limited to what programs you can use. Another thing you can do is to basically manage your exceptions using rather a more of an advanced approach. Doing this can give you access to open specific ports, secure certain programs for access, ICMP access, etc. but since explaining this is relatively repetitive, been previously posted and quite in depth, you can find the entire ariticle here:
(A few good links are provided there as well. If you really want to go more in depth for Windows Firewall configuration only, then that is a good thread to look at)
The last possible thing you should do is turn on logging for windows firewall (it's defaultly turned off). Doing this is relatively simple and requires just a little browsing in the Windows Firewall configuration menu subheadings (Control Panel->Windows Firewall->Advanced->Security Logging Settings). Once you are there, check the box under "Log Dropped Packets" and the other if you want to closely monitor your network activities. Viewing these logs can be done by default with Windows' own seperate method, but it's rather primitive and inefficient. A better way to view the logs is by finding a third party log viewer. Many free ones can be found on the internet simply by using Google Search or browsing around this forum. Below I have provided a link if you'd rather not search:
2.) Getting the Advanced Features from other Firewalls:
If you want the advanced features that heavy firewalls tend to provide, there are combinations of relatively small programs you can use to substitute with Windows Firewall. They have been divided up into two sections below for easier browsing.
(A.)Outbound Protection:
One of the major issues Windows Firewall has is that it does not offer any outbound control. This makes your computer vulnerable once your defenses have been broken, since threats such as Trojans can freely transmit information back to a hacker. A few good programs out there that can substitute outbound protection are listed below (they also tend to not use a lot of memory):
-Prevx1 "R" (Beta version from Prevx, which is good, but still uses quite a bit of memory)
-AppDefend/GSS Beta (Made by Ghost Security and is low memory using and an effective solution.)
-ProcessGuard (Created by Diamond CS and is not really considered outbound protection. However, it still has several similar features that can aide you.)
Having programs such as these behind your Windows Firewall (or any other inbound only firewall) ensures a relatively strong outbound protection on your computer. All three programs listed have an easy-to-use interface, so setting it up isn't hard. Most of what you need to configure is given to you in the form of alerts as the programs ask for outbound access.
(B.) Other Features:
Almost all firewalls out there today that are paid offer several, rather random advanced features, such as cookie protection, script blocking, etc. Of course, Windows Firewall lacks such luxuries, so you must substitue by using a few good programs. Below I've listed a few programs that can help give you some idea of what you can use for the advanced firewall features you want.
-CCleaner (A very thorough, good cachecleaner. Used to remove all junk files, temporary files, and other old & unwanted files hanging around in your computer.)
Details:
-This is a very flexible program for removing 'residue/junk' files due to its several options and areas that are available for your use. Your scans can also be customized each time before running.
-SpywareBlaster (By far one of the best free substitutes. It offers cookie protection for Firefox and IE, a restricted sites list, and ActiveX protection for IE. Best of all it consumes basically no resources!)
Details:
-This is a very simplistic form of protection which I find great. All you basically do is install it, update and enable all protection and its done! You can selectively remove or add new threats to be blocked in the main menu's subheadings.
-MS Antispyware (This antispyware [along with any other ones that provide thorough real-time protection] provides features such as script blocking. Not only that, but they add a defensive layer against spyware/malware/adware too! Tools like these are great for performing several security tasks.)
Details:
-Well it's basically one of your big antispyware products out there that doesn't require as much configuring. Script-blocking is already enabled, but I would reccomend you allow it to block all scripts (Located at: Options->Settings->Real-Time Protection and check the box "Block all scripts"). You will still recieve some alerts asking your confirmation to block/allow programs/scripts after doing this, but just not as many without this option enabled. Also, there are a few handy extra tools added in, such as a browser restorer, a mini-cache cleaner, and a system explorer with decent flexibility. These can be accessed through the "Advanced Tools" button/section.
-HijackThis(Allows you to MANUALLY remove threats and has several convenient tools, such as the process manager, secure uninstall tool, etc.)
Details:
-This product's purpose is relatively used for scanning, detailed log creation, and malware removal. However, this requires advanced knowledge in order for malware removal and it is best that you post in a forum for help. Despite all this, HijackThis still remains very flexible and has several tools that a unknowledgeable person can use with ease. The other tools you seek can easily be found in the Misc. Tools Section (Accessed by the button "Misc. Tools" or through the "Config" button).
3.) The Remaining Basics:
Well, as I said previously Windows Firewall isn't that great, and even with all these combinations & configurations added onto it, you are never 100% safe. So remember, there are some things you should be careful of! Windows Firewall is a relatively easy in being overwhelmed, but still provides the basic protection you need (tested on GRC with perfect results).
There are a few things to keep in mind which really apply to computer security common sense, such as:
-Use a secure web browser, such as FireFox (FF) or Opera instead of Internet Explorer (IE).
-Be cautious on where you browse and try to avoid being a target for hackers (e.g your computer is found to contain valuable information).
-Get the basic defenses, such as antispyware, or antiviruses which are available free to the public from some companies.
-Keep Windows updated, as well as all other security software you have on your computer currently.
-Watch your network closely (especially if its wireless) as they are also targetted for attacks.
Conclusion: Well I hope this helps for the people who still want the extra "oomph" in their basic Windows Firewall protection. However, I myself have stopped using it quite a long time ago and moved on towards other firewalls. This is because they are better in a lot of ways UNLESS you still prefer lower memory usage and the Windows-simple style to protection. But as said before and by several others who agree, Windows Firewall is certainly better than nothing!:p
P.S Feel free to post any comments or additions into my guide. I will probably edit it if I find anything I've missed. (And sorry if I've missed anything relatively big.) | http://www.antionline.com/printthread.php?t=270227&pp=10&page=1 | CC-MAIN-2017-22 | refinedweb | 1,505 | 61.16 |
PhpStorm 6.0.1 EAP build 129.177/196
We continue to refine the IDE and work on plugin APIs. This build addresses both aspects by implementing a much requested feature purely relying on php-openapi. It required numerous adjustments – and much more are still pending, but we have a great progress..
The feature above is implemented using PHP open API. The introduced Extension Point also allows 3rd party plugin creators to provide type info automatically using framework’s established practices and configuration. The API is still highly unstable (and has already evolved since this build) but we will provide guidance. Check out tutorials on plugin development.
Other notable changes
- PHP type inference for variables should now correctly work with Fluent Interface style call chains – again removing the need for @var annotations.
- PHP completion for array indexes has been significantly improved
- PHP inspection got a couple of new ones – division by zero and invalid string operation
- Details on resolved tracker issues
Download PhpStorm 6.0.1 EAP for your platform and please report any bugs and feature request to out Issue Tracker. Automatic update via patch is also available.
Develop with pleasure!
-JetBrains Web IDE Team
24 Responses to PhpStorm 6.0.1 EAP build 129.177/196
Mark Badolato says:April 5, 2013
I posted this a few months back and it good some good reception, so I thought I’d post it again. We built a dark color theme that mimics the scheme used in the Symfony2 and Doctrine2 documentation code snippets. If anyone is interested, it’s available (with a few screenshots) at.
Jonathan Cardoso [JCM] says:April 5, 2013
It’s possible to have automatic metadata if the string used as argument is just some part of the namespace? Or even if the string itself is the class name. Like, the following:
$orm->getRepository( 'Entity\User' ); //return Entity\User instance.
and
$helper->getHelper( 'MyHelper' ); //return App\Namespace\MyHelper instance, so this method should return App\Namespace\{argument passed}
Alexey Gopachenko says:April 5, 2013
Not now, as solution is precisely targets, but we’ll go for quite soon.
Maciej says:April 5, 2013
After update i get a Message box saying
Plugin com.intellij failed to initialize and will be disabled:
Cyclic dependency: [class com.intellij.psi.stubs.StubIndexImpl]
Please restart JetBrains PhpStorm
this box shows up 2 times
and another one
Plugin com.intellij failed to initialize and will be disabled:
null
Please restart JetBrains PhpStorm
Stanislav Butsenko says:April 5, 2013
Crashes after update to this build, PHPStorm not run. Ubuntu 12.10 x64 generic 3.5.0-26. Logfile:
Heart says:April 5, 2013
same here… kill process worked
Rafi B. says:April 5, 2013
Yup. Lots of exceptions on first run, killed processes, rerun and works fine. (Xubuntu 12.10 x64)
Stanislav Butsenko says:April 5, 2013
Problem was solved by the closing of the previous PHPStorm process:
~~~
ps ax | grep phpstorm
kill
~~~
Alexey Gopachenko says:April 5, 2013
We might have a packaging issue. Please stand by. We give instructions / issue small patch ASAP.
Alexey Gopachenko says:April 5, 2013
Ok, packaging issue confirmed, patch-update will be available in a hour.
Zyava says:April 5, 2013
“Details on resolved tracker issues” link doesn’t work
Alexey Gopachenko says:April 5, 2013
I just renamed the build in tracker – link fixed in a minute.
Alexey Gopachenko says:April 5, 2013
Patches are UP. Full builds are also available.
In tracker its now EAP 129.177&196
Rafi B. says:April 5, 2013
Thanks!!
Adam Patterson says:April 6, 2013
This latest version for OSX is really really flaky.
The php structure won’t load.
When I open the app I get about 5 java errors ( Plugin com.intellij failed ).
Mike Schinkel says:April 7, 2013
1.) Can you explain these and when they are needed?
/** @noinspection PhpUnusedLocalVariableInspection */
/** @noinspection PhpIllegalArrayKeyTypeInspection */
2.) If the factory methods accept several addition parameters such as those for dependency injection can we just omit them for this purpose?
3.) Does it support PHP 5.3 array syntax, i.e. array(…) vs. […]?
4.) Why is PhpStorm indicating that all the class names are undefined?
5.) Would you be so kind as to verify that I understood correctly how this should be used: .phpstorm.meta.php?
Alexey Gopachenko says:April 7, 2013
1) To have a “green” file.
2) Add anything suitable there to have a green file.
3) Most likely, but its better to stick to current syntax. This file is for IDE only.
4) Note the PHPSTORM_META namespace. You should use FQNs.
5) see(4) Use FQNs for class references, i.e.
\RESTian_Application_Xml_Parser
Mike Schinkel says:April 8, 2013
Hi @Alexey,
Thanks.
1) What’s a “green” file?
2) Do you mean I need to add hypothetical parameters, i.e. like this?
\RESTIan::get_new_parser('', new \RESTian_Request(), new \RESTian_Response())
3) Since my projects are configured for PHP 5.3 this is what PhpStorm does when I use PHP 5.4 syntax:
4.) Ah, thanks. I’ve never used namespaces as they are not generally used for WordPress plugins yet given WordPress still supports 5.2.4.
Alexey Gopachenko says:April 8, 2013
1) Well, the file with no syntax and inspection errors. “Green” by the color of inspection status indicator on the top of vertical scrollbar. We want to have only real errors.
2) yes, or just (”, null, null) or anything that will keep file green.
3) I see. Then you can use 5.3 syntax of course. We, again, want the file green to be able to spot any real errors, i.e. unresolved method or class references.
Mike Schinkel says:April 8, 2013
Thanks!
Adrien Brault says:April 7, 2013
Hey, thanks for this!
I’ve started developping a symfony2 plugin. It currently can detect services types retrieved from the dependency injection container.
The plugin url is (screenshot/gif included :>)
Alexey Gopachenko says:April 7, 2013
Great! This is what we expect to be eventually done for all major frameworks/packages.
Quick note: try to stick to classes from php-openapi.jar only, i.e MethodReference instead of MethodReferenceImpl, etc. The naming convention is most obvious.
We’ll be providing more tutorials – and extension points in upcoming builds.
Alexey Gopachenko says:April 7, 2013
Your code won’t work with multiple projects open – the instance is a singleton, so you should either make it a project component or use element’s project’s user data to hold yours. Look at updated sample code.
Enrique Piatti says:April 8, 2013
I’ve added two new actions to Magicento, one for generating the PHPSTORM_META namespace with all the classes and factories (this actions is accessible everywhere with Alt+M like always), and another one for updating that file with the class for the factory over the current cursor (updating is faster than generating the whole file). | https://blog.jetbrains.com/webide/2013/04/phpstorm-6-0-1-eap-build-129-177/ | CC-MAIN-2021-17 | refinedweb | 1,148 | 67.76 |
:
import numpy and import matplotlib does work in the interactive python
shell.
thanks,
Archana.
On 3/31/07, Tommy Grav <tgrav@...> wrote:
>
> I do not immediately see why the error occurs. Hopefully someone else
> can add their input. Off the cuff it seems like matplotlib has not been
> installed properly. Can you confirm that import numpy and import
> matplotlib works in the interactive python shell.
>
> Cheers
> Tommy
>
> [tgrav@...] /Users/tgrav --> python
> Python 2.4.4 (#1, Oct 18 2006, 10:34:39)
> [GCC 4.0.1 (Apple Computer, Inc. build 5341)] on darwin
> Type "help", "copyright", "credits" or "license" for more information.
> >>> import numpy
> >>> import matplotlib
> >>> from matplotlib.numerix import *
> >>> random_array
> <module 'matplotlib.numerix.random_array' from
> '/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/site-packages/matplotlib/numerix/random_array/__init__.pyc'>
> >>> numpy.__version__
> '1.0.1.dev3436'
> >>> matplotlib.__version__
> '0.87.7'
> >>>
>
> so random_array works fine in my version
>
>
> On Mar 31, 2007, at 4:26 PM, Archana Ganesan wrote:
>
> Hi all,
>
> The exception I get is
> Traceback (most recent call last):
> File "App1.py", line 6, in ?
> File "Frame1.pyc", line 9, in ?
> File "Simulation.pyc", line 16, in ?
> File "pylab.pyc", line 1, in ?
> File "matplotlib\pylab.pyc", line 199, in ?
> File "matplotlib\cm.pyc", line 5, in ?
> File "matplotlib\colors.pyc", line 33, in ?
> File "matplotlib\numerix\__init__.pyc", line 147, in ?
> ImportError: No module named random_array
>
> I have numpy installed in site-packages. The setup.py that I am using is
> as follows:
>
> Thanks,
> Archana.
>
>
> from distutils.core import setup
> import os
> from os.path import join
> import shutil
>
> import glob
> import py2exe
> from py2exe.build_exe import py2exe
> import sys
>
> import matplotlib
> mpdir, mpfiles = matplotlib.get_py2exe_datafiles()
>
> # cleanup dist and build directory first (for new py2exe version)
> if os.path.exists("dist/prog"):
> shutil.rmtree("dist/prog")
>
> if os.path.exists("dist/lib"):
> shutil.rmtree ("dist/lib")
>
> if os.path.exists("build"):
> shutil.rmtree("build")
>
> options = {"py2exe": {"compressed": 1,
> "optimize": 2,
> "packages": ["encodings",
> ## "kinterbasdb",
> "pytz.zoneinfo.UTC",
> #" matplotlib.numerix",
>
> ##"numpy"
> ## "PIL",
> ],
> "excludes": ["MySQLdb", "Tkconstants", "Tkinter",
> "tcl",
> "orm.adapters.pgsql ", "
> orm.adapters.mysql"
> ],
> "dll_excludes": ["tcl84.dll", "tk84.dll",
> "wxmsw26uh_vc.dll"]
> }
> }
> zipfile = r"lib\library.zip"
>
> setup(
> classifiers = ["Copyright:: your name",
> "Development Status :: 5 Stable",
> "Intended Audience :: End User",
> "License :: Shareware",
> "Operating System :: Microsoft :: Windows 2000",
> "Operating System :: Microsoft :: Windows XP",
> "Operating System :: Microsoft :: Windows 9x",
> "Programming Language :: Python, wxPython",
> "Topic :: Home Use"
> "Natural Language :: German",
> "Natural Language :: French",
> "Natural Language :: English"],
> # windows = [wx_emb],
> #console = [twcb],
> options = options,
> zipfile = zipfile,
> data_files = [("lib\\matplotlibdata", mpfiles),
> matplotlib.get_py2exe_datafiles() # if you don't use
> the lib option
> #### ("prog\\amaradata", amaradata),
> #### ("prog\\amaradata\\Schemata", amaraschemata),
> #### ("prog\\", python4dll)
> ]
> )
>
>
> On 3/31/07, Tommy Grav <tgrav@...> wrote:
> >
> > It is hard to guess what exactly your problem is as you do not provide a
> > code
> > example or the traceback call of your exception. I would venture that
> > you are
> > trying to create a num_array without having Numerix, numpy or numarray
> > imported
> > or installed on your machine.
> >
> > Some more information about your troubles would be needed to really help
> > you out.
> >
> > Cheers
> > Tommy
> >
> >
> > On Mar 31, 2007, at 3:50 PM, Archana Ganesan wrote:
> >
> > Hi all,
> > I have a python application that uses matplotlib.I am trying to compile
> > it into an executable using py2exe. I am encountering "No module named
> > num_array problem". I do not know how to resolve this. I notice that "
> > matplotlib.numerix" is in the included package. Did you encounter this
> > problem. I am very new to this and I have to get it done by tomm. So I am
> > sorry if it is really silly.
> >
> > Thanks,
> >
> > Archana.
> >
> >
> > On 3/31/07, Archana Ganesan < archana.ganesan@... > wrote:
> > >
> > > Hi all,
> > >
> > > I have a python application that uses matplotlib. I want to compile it
> > > into an executable. I tried using py2exe but it returned some error
> > > w.rt matplotlib. Cpuld anyone please help me with this? Is there some
> > > other way to get it done?
> > >
> > > Thanks,
> > > Archana
> > >
> >
> > -------------------------------------------------------------------------
> >
> > | https://sourceforge.net/p/matplotlib/mailman/message/13646262/ | CC-MAIN-2017-04 | refinedweb | 654 | 61.22 |
dotCMS and the Scripting Worlds CollideAug 16, 2009
The Problem
Although Velocity is great and dotCMS has a bunch of rich web tooling, sometimes it is easier or more practical to use a script in another language. While it is possible to provide any kind of custom tooling or logic one might need through the use of Viewtools, Struts and other related tooling, it has become apparent that dotCMS web developers could benefit from more flexible options when it comes to implementing their logic. Our new plugin architecture aids greatly in deploying java+velocity stack kinds of solutions but all still require custom Java classes and of course a restart of the dotCMS to deploy. Additionally, most dotCMS web developers prefer to stay within the familiar confines of HTML, CSS, and scripting languages. Enter the dotCMS Scripting Plugin.
Download and Deploy your Plugin
To use scripting, first you will need to download the scripting plugin. There is a version for both the dotCMS trunk and dotCMS released code. Both can be checked out via SVN and you can download the released version of the plugin here.
After the download is complete, you must deploy your plugin. In order to deploy your plugin, unzip the download and copy resulting folder to the plugins directory and run "ant deploy-plugins" from the root of the dotCMS. In the future dotCMS will be providing binary plugins which won't require ANT but for now it is a minor inconvenience considering the power plugins bring to the dotCMS. If you need more help deploying your plugins see
Getting your System Ready to Script
Once your plugin is deployed and the dotCMS has been restarted you will see a new role in the dotCMS named "Scripting Developer". You need to add this role to any user who will be using the new scripting tools. The reason for this is that the scripting plugin checks for the latest modified user of the web asset with the scripting code to ensure not just anyone can deploy code to your dotCMS installation. This can be a problem as the scripting languages are really powerful and can even be used to call and execute Java code or any dotCMS java class.
Evaluating your First Script in dotCMS
There are two primary ways to interact with the dotCMS Scripting engine. First you can simply just evaluate an expression from Velocity. Consider the example below
Javascript
$scriptingUtil.evalJS("Math.round(4.3)")
Groovy
$scriptingUtil.evalGroovy("2 + 2")
Ruby
$scriptingUtil.evalRuby("2 * 2")
Python
$scriptingUtil.evalPython("4 * 1")
The $scriptingUtil is the Velocity tool that allows you to interact with with the dotCMS scripting engine. It has methods on it which, as seen above, will both evaluate and execute your scripts. In the example we asked the tool to evaluate a JavaScript expression for us. The result on the screen would be 4. Now lets add a little more to our test in the following example
#set($jsObject= $scriptingUtil.evalJS('
function printMe(num){
return "Printed from function"
}
'))
$scriptingUtil.callMethod($scriptingUtil.getJavaScriptLanguage(),$jsObject,'printMe',null)
The result would print "Printed from function" to the screen. As you can see, we defined a function and then called the JavaScript function we defined. The last arguement of the method, currently null, can be used to pass in parameters for the called function.
Setting Variables
Some people may be wondering if they can set variables to use within the scripting languages. Our "declare" method allows you to do just that. Let's alter our first example a little.
$scriptingUtil.declare("xyz", 4)
Javascript
$scriptingUtil.evalJS("Math.round(xyz)")
Groovy
$scriptingUtil.evalGroovy("xyz + 2")
Ruby
$scriptingUtil.evalRuby("$xyz * 2")
Python
$scriptingUtil.evalPython("xyz * 1")
As you can see we declared xyz and then used it within each scripting language. Notice a couple of things here.
- We only had to declare it once.
- Within the scripting languages it is handled as a global variable hence the $ in front of xyz in the Ruby example.
Keeping Scripts as dotCMS File Assets
You can also maintain your scripts within the dotCMS as files. This is beautiful because combined when combined with webdav you can get into some great RAD style programming. Need a quick Help Desk application or some custom form/report, no problem. Consider the following
Ruby Greeter program.
# The Greeter class
class Greeter
def initialize(name)
@name = name.capitalize
end
def salute
"Hello #{@name}! from ruby salute"
end
end
From a piece of content look at how we interact with our class
$scriptingUtil.execFile("/global/ruby/rubybasicgreeter.rb")
#set($robject=$scriptingUtil.evalExpression("/global/ruby/rubybasicgreeter.rb",'Greeter.new("Jason")'))
$scriptingUtil.callMethod($robject,"salute",null)
The key here is that you must first execute the file. This will execute any code within the file. In our case all we did was define a class but if we had logic outside of the class definition of after it called the class or a function within the script it would execute that code. Remember that before you can evaluate an expression on a file you must execute the file. From this point the file is loaded and is referenced with the file name as the key. In our expression Greeter.new("Jason") we constructed a new instance of the Greeter. After we got our greeter instance we called the salute method which printed o the screen for us.
Curl in PHP
Here is another example of how you can script - this time with php. This short script will curl (fetch) the contents of as a string and return them as a java string for you to parse to your hearts content.
$scriptingUtil.evalPHP('
$curl_handle=curl_init();
curl_setopt($curl_handle,CURLOPT_URL,"");
curl_setopt($curl_handle,CURLOPT_RETURNTRANSFER,1);
$buffer = curl_exec($curl_handle);
curl_close($curl_handle);
return $buffer;
')
As you can see, the possibilities are endless. There is much more to show and much more that can be done but as you can see this blows the doors wide open and brings tremendous power to what is already one of the most flexible wCMS frameworks on the market. And best of all it is all open source :-) | https://dotcms.com/blog/post/dotcms-and-the-scripting-worlds-collide | CC-MAIN-2021-43 | refinedweb | 1,017 | 64.2 |
2
Function Fundamentals
Written by Massimo Carli
In Chapter 1, “Why Functional Programming”, you learned that functional programming means programming with functions in the same way that object-oriented programming means programming with objects.
Objects and functions are very different beasts. Objects, at first sight, probably look closer to what you see and interact with every day. Think of your car, for instance. It has some properties like brand, color, engine and size. You use property values to distinguish a Ferrari from a Fiat 500. When you write your code, you model objects using classes. For instance, you define a
Car class like this:
class Car( val brand: String, val color: Color, val engine: Engine, val size: Int, var speed: Int ) { fun drive() { // Driving code } fun accelerate() { speed += 1 } }
The state of an object is the set of its properties’ values. Objects also interact, sending messages to each other. This interaction happens by invoking something called methods.
You can interact with a
Car instance by invoking the
accelerate method to increase its speed, which then changes its state.
Classes allow you to define what the methods are. The concept of a class is just the first tool you have to describe your application in terms of objects. Many other concepts and tools help define what object-oriented programming is, such as interface, inheritance, polymorphism and aggregation.
But what about functions? Are functions also close to what you see and interact with every day? What does it really mean to program with functions? Is there something similar to the concept of a class with functional programming?
Unfortunately, you can’t answer all these questions in a single chapter. To understand it, you need to really understand what a function is, which requires the knowledge of some fundamental concepts of category theory. This chapter will give you that knowledge! What you’ll learn in this chapter will give you a strong foundation for the rest of the book.
In this chapter, you’ll learn:
- What category theory is.
- How a category is related to the concept of a function.
- The useful concepts of initial and terminal objects, using logic as a fun example of a category.
- What the relationship is between category theory and functional programming.
- What the concept of type has to do with sets.
- Where the Kotlin
Nothingand
Unittypes come from.
- What it means for an object to be an element of a set in the context of category theory, and therefore, functional programming.
These are all very theoretical concepts, but you’ll see some practical examples using the Kotlin language.
Note: If you’re not confident with Kotlin, our Kotlin Apprentice book is the best place to start.
Note: The concepts in this chapter might look theoretical and, sometimes, not very intuitive. Don’t worry! It’s perfectly normal, and even expected, to read this chapter multiple times to digest the content that will help you to better understand many other concepts of functional programming in this book.
What is a function?
You might already have an initial idea about what a function is from the math you learned in school. You may remember that a function is a way to describe how to calculate an output given an input value, like in Figure 2.1:
For instance, the function:
fun twice(x: Int) = 2 * x
Returns an output value that’s double the value you pass in as an input. That’s a representation of a mathematical function.
In the context of functional programming, a function is something more abstract that’s not necessarily related to computation. A function is a way to map some values to others. The bunch of all the values a function accepts as input is the domain of the function. The bunch of all the values the function returns as output is the range of the function. Domain and range can also be the same bunch of values.
Note: It’s important at this point to emphasize the term bunch of values. This is because the term set, as you’ll see soon, is something that gives some kind of rules to that bunch. For instance, a set can include an object or not, and it can’t include the same object more than once.
Figure 2.2 better represents what a function is:
- For each value in the domain, there’s one and only one value in the range. This implies that the function always returns a value for inputs in its domain.
- The function
fcan map multiple values in the domain to the same value in the range.
- By definition, the function has meaning for each value in the domain, as you’ll see later. This might seem obvious, but it’s a crucial concept to understand.
Exercise 2.1: Can you write an example of a function mapping distinct values in the domain to non-distinct values in the range, like
f(b)and
f(c)in Figure 2.2?
Give it a try, and afterward, check the challenge project for a solution to see how you did. You can find hints and an explanation of the solution in Appendix B, “Chapter 2 Exercise & Challenge Solutions”.
Functions like
twice mapping distinct values in the domain to distinct values in the range have some interesting properties. As you’ll see later, the most significant is that they have an inverse function. An inverse function maps the values in the range back to the domain.
Exercise 2.2: Can you write the inverse function of
twice? What are the domain and range for the inverse function? Check out the challenge project and Appendix B for the solution.
Another simple and useful exercise is to define the domain and range for
twice in the previous example:
fun twice(x: Int) = 2 * x
In that case, you have what’s shown in Figure 2.3:
In this exercise, you’ve done something that’s not as obvious as it appears because you used the set of all the
Int values as the domain and range for the function
f. But what is
Int? What’s the relation of the
Int type with the bunch of things mentioned earlier?
To really understand all this, it’s time to introduce category theory.
Note: If you’ve heard category theory is intimidating, don’t be scared. This book is here to help. :] Soon, you’ll realize how understanding what a category is will help you assimilate many critical concepts about functional programming. Again, feel free to read this chapter multiple times to make this process easier. This book will go over only what you really need, and it’ll be a lot of unexpected fun.
Introduction to category theory
Category theory is one of the most abstract branches of mathematics, and it’s essential here because programming is one of its main applications. It’s also important to start with the definition of category.
A category is a bunch of objects with connections between them called morphisms. Categories follow three rules:
- Composition
- Associativity
- Identity
You can picture an object like a dot with no properties and no further way to be decomposed. A morphism is an arrow between two objects like in Figure 2.4:
In this image, you:
- Have objects named
a,
band
c.
- Define two morphisms between
aand
band one morphism between
band
c.
- Assign the names
f1and
f2to the morphisms between
aand
band the name
gto the morphism between
band
c.
Objects and morphisms are the primitive tools you can use to describe every possible concept about category theory.
As mentioned earlier, objects and morphisms need to follow some rules that will lead you to other fundamental functional programming concepts. It’s time to study them in detail.
Composition
A bunch of objects and arrows don’t make a category; they also need to follow some important rules, and composition is probably the most significant.
Look at Figure 2.5:
In this image, you have:
- The objects
a,
band
c.
- The morphism
ffrom
ato
b.
- The morphism
gfrom
bto
c.
The composition rule says that for every object
a,
b and
c and every morphism
f and
g, as shown in Figure 2.5, the category must have a morphism from
a to
c. You represent this as
g◦f between
a and
c, which is the composition of
f and
g. You read
g◦f as “g after f”, where the small circle
◦represents the composition.
For a Kotlin example, if
f is your
twice and
g is
triple, then
g◦f might look like:
val `g◦f`: (Int) -> Int = { triple(twice(it)) }
Note: An analogy with LEGO® makes the definition of the composition property for a category easier to understand. Imagine an object of a category is a LEGO® brick. Being able to attach a LEGO® brick to another is equivalent to having a morphism between them. In this case, composition means you can attach one LEGO® to another to get another LEGO® component that is the composition of the two bricks.
Associativity
The associativity rule is similar to the one you studied back in school about addition or multiplication. Look at the following image:
Here, you have:
- The objects
a,
b,
cand
d.
- The morphism
ffrom
ato
b.
- The morphism
gfrom
bto
c.
- The morphism
hfrom
cto
d.
The associativity rule says that for every object
a,
b,
c and
d and morphism
f,
g and
h (like in Figure 2.6), the following equivalence must be true:
This means that if you do the composition of
f,
g and
h, it doesn’t matter if you first compose
f to
g or
g to
h. Looking at Figure 2.6, this also means that, with those objects and morphisms, there must be a morphism from
a to
d, which you can obtain by either composing
f with
h◦g or composing
g◦f with
h.
Continuing with the previous example and quadrupling
h, this means:
twice(triple(quadruple(x))) == quadruple(twice(triple(x)))
Because these are equal, it conforms to associativity.
Note: The LEGO® analogy helps simplify the understanding of the associativity property too. Imagine you have three different LEGO® bricks you call A, B and C. If you attach A and B first and then C, you get a component that is exactly the same as what you’d get by attaching B and C first and then A. What you get is basically another LEGO® component you can use like the others.
Identity
Identity is the last rule, but it’s no less important than the other two. Consider the smiling diagram in Figure 2.8:
The identity property says that every object in a category must have at least one morphism to itself. This is why, in Figure 2.8, you have:
The objects
aand
b.
A morphism
ffrom
ato
b.
A morphism
iafrom
ato itself.
A morphism
ibfrom
bto itself.
This must be true for all the objects in the category.
An example of
ia might be
timesOne or
{ x * 1 }:
timesOne(x) == x
It’s interesting to use the identity with the previous properties. You can easily prove that, in a category, the following equivalence is true:
Note: Understanding why a category needs identity might not be very intuitive, and there’s not a plausible way to visualize identities using LEGO®. If a morphism in this analogy means to attach a LEGO® brick to another, it’s quite difficult to represent how to attach a piece to itself. On the other hand, you could think of the inverse of attaching two LEGO® pieces as the action of detaching them. In this case, the composition of attaching and detaching leads you to the initial piece.
The concept of isomorphism at the end of this chapter will probably help, but don’t worry — everything will be clearer when you get to Chapter 11, “Functors”, and Chapter 12, “Monoids & Semigroups”.
Now that you know what a category is, it’s time for some fun — you’ll give some meaning to objects and morphisms. Always remember that what’s true for a category will also be true when you give objects and morphisms specific meanings.
Category and logic
As mentioned, a category is a very abstract concept, but giving objects and morphisms some specific meanings makes everything closer to what you use every day.
In the context of logic, assume that objects are propositions and morphisms are entailments. Consider, then, Figure 2.10:
Using the symbol ⇒ to represent entailment, you have:
- The propositions
A,
Band
C.
- An arrow from
Ato
B, meaning that
A⇒
B.
- Another arrow from
Bto
C, meaning that
B⇒
C.
To prove it’s a category, you need to verify the three important rules:
- Composition
- Associativity
- Identity
It’s time to have some fun!
Proving composition
As you know, composition is the property that allows you to compose a morphism from
A to
B and one from
B to
C into a single morphism from
A to
C. In this case, if
A ⇒
B and
B ⇒
C, is it also true that
A ⇒
C? Try to use the following propositions:
A= Alice drives to work every day.
B= Alice has a driver’s license.
C= Alice is at least 18 years old.
The fact that Alice drives to work every day entails she has a driver’s license. That Alice has a driver’s license entails she’s at least 18. You then need to prove the following: Is the fact that Alice drives to work every day entailing she’s at least 18 years old? This is true, and this proves composition.
Note: In this example, you’re assuming Alice is in a place where you need to be 18 years old to drive a car and vote. You might also object that this works just for the previous propositions. In this case, you have two options: You can just believe it or find a case where this isn’t true. :]
Proving associativity
To prove associativity, you need a new proposition like:
D= Alice can vote.
In this case, referring to Figure 2.10, you can use the following entailments:
f=
A⇒
B= Alice drives to work every day. ⇒ Alice has a driver’s license.
g=
B⇒
C= Alice has a driver’s license. ⇒ Alice is at least 18 years old.
h=
C⇒
D= Alice is at least 18 years old. ⇒ Alice can vote.
From the definition of category, to prove associativity, you need to prove that:
(h◦g)◦f = h◦(g◦f)
You can break it down like this:
(h◦g)= Alice has a driver’s license. ⇒ Alice can vote.
(g◦f)= Alice drives to work every day. ⇒ Alice is at least 18 years old.
(h◦g)◦f= Alice drives to work every day. ⇒ Alice can vote.
h◦(g◦f)= Alice drives to work every day. ⇒ Alice can vote.
Which proves the hypothesis.
Proving identity
The final property to prove is very interesting and funny. In logic, you basically need to prove that for every proposition
A, you can say that
A ⇒
A. This is equivalent to saying that:
- Alice has a driver’s license. ⇒ Alice has a driver’s license.
This actually has a name: tautology. :] This proves that logic is a category, and you’ll use it to introduce two more crucial concepts: initial and terminal objects.
Category theory and the real world
Before introducing initial and terminal objects, it’s valuable to stop for a second and think about an important aspect of category theory. In the introduction for this chapter, you read that object-oriented programming allows you to model your code using objects, which might seem closer to what you see and interact with every day. Can you say the same thing about a category?
A category has composition as a fundamental property. You can even say that category theory is the science of composition. Isn’t decomposing concepts into smaller ones what you do every day? Humans’ brains continuously decompose concepts to make them simpler to understand and memorize, and then recompose them. This might seem somewhat philosophical, but it’s proof that category theory isn’t something completely unrelated to reality.
At this point, an example might help. Suppose somebody asks you to solve the following addition problem:
7 + 10
Of course, you answer
17. When you do that, are you getting the result from your memory, or is your brain actually computing the addition of
7 with
10? Frankly, it could be either. With this simple addition, your brain has probably memorized the answer somewhere and is giving it as the answer.
Now, imagine somebody asks you to solve the following subtraction problem:
42 - 8
In this case, you probably don’t have the answer memorized, so your brain needs to do some “computation”. Because it’s not an obvious answer, your brain might do something like this:
42 - 2 = 40 40 - (8-2) = 40 - 6 = 34
Putting it all together, you might mentally compute:
42 - 2 - (8 - 2) = 34
Your brain has decomposed the simple subtraction into multiple operations that are probably easier to calculate and then composed the results back into a single value. This is an example of composition!
Initial and terminal objects
You already learned that category theory explains everything in terms of objects and morphisms. Not all the objects and morphisms are the same, though. For instance, not all the objects are somehow connected with a morphism. For logic, this means that a proposition doesn’t necessarily entail all the others. For the same reason, not all the propositions are entailments of another.
Note: Using the LEGO® analogy, you represent this concept saying that not all the LEGO® pieces can be attached to any other piece.
To understand why this is, you need the definition of uniqueness. In this context, you can say that the morphism
f between the objects
A and
B is unique if any other morphism
g between the same objects cannot be different from
f. In short, if this happens,
f and
g must be equal.
With this definition in mind, you can define an initial object as an object with a unique outgoing morphism to every other object in the category. A terminal object is an object with a unique incoming morphism from any other object. Figure 2.11 gives an idea of these concepts:
Not all categories have initial and terminal objects but, if they do, they are unique. In logic, the initial object has the name False and the terminal object True.
Category properties have funny implications on these objects:
- Each object has at least one identity morphism. This means that what is false is false, and what is true is true.
- Because there’s always a morphism starting from the initial object to any other objects, there’s also a morphism from the initial object to the terminal one. This means that a false assumption entails everything. Therefore, “Tom has wings” entails “Tom can fly”, is a counterfactual implication.
You’re probably wondering what all this has to do with functional programming — and Kotlin.
The good news is that anything you see that’s true for a generic category is also true for a specific one. What’s true in logic is also true in programming when you give a specific meaning to objects and morphisms. It’s now time to be more pragmatic and start studying the most critical category for an engineer — spoiler alert — the category of types and functions.
Exercise 2.3: Can you prove that using
Sets as objects and “is a subset of” as morphisms results in a category? In other words, a morphism from set
Ato set
Bwould mean that
Ais a subset of
B. In that case, what are the initial and terminal objects?
Don’t forget to check out Appendix B for hints and solutions.
Category of types and functions
So far, you’ve learned what a category is, and you also had some fun with the category of propositions and entailments. That allowed you to introduce the properties of a category and define some significant concepts like initial and terminal objects. That’s all good — but this is a book about programming, and you need something more pragmatic. You basically need to answer the following questions:
- What happens if objects are types and morphisms are functions?
- What’s the meaning of composition, associativity and identity in the category of types and functions?
- What’s the meaning of initial and terminal objects?
In the following paragraphs, you’ll answer these questions using the Kotlin language, and you’ll have some revelations about some of the most important Kotlin standard types. Here, using Kotlin, you’ll:
- Prove that using types as objects and functions as morphisms, you define a category.
- See what initial and terminal objects mean when dealing with types and functions.
- Find out where the Kotlin types
Unitand
Nothingcome from.
As mentioned, you’ll do this using Kotlin.
Do types and functions define a category?
As a first step, you need to prove that by using types as objects and functions as morphisms, what you get is actually a category. You need to prove:
- Composition
- Associativity
- Identity
Proving each of the category properties is also a good exercise to review some interesting Kotlin concepts.
In the material for this chapter, you’ll find the starter project with some empty files you’ll fill in along the way. Start by opening the Aliases.kt file and write the following definition:
typealias Fun<A, B> = (A) -> B
This is a type alias that represents all the functions from a type
A to
B. If
A and
B are two objects of the category of types and functions,
Fun<A, B> is a morphism.
This simple definition allows you to prove each of the properties for a category.
Proving composition
To prove composition, you need to prove that for any function
f from
A to
B and any function
g from
B to
C, there’s an equivalent function:
g◦f from
A to
C, which is the composition of the two.
This is nothing new to you because you’re probably composing functions every day. Consider, for instance, the following code you can write in the Main.kt file:
fun twice(a: Int): Int = a * 2 // 1 fun format(b: Int): String = "Result is $b" // 2
These are very simple functions that:
- Double the
Intvalue it gets as input. The type of
twiceis
Fun<Int, Int>.
- Format the
Intit gets as input to a
String. The type is
Fun<Int, String>.
You can compose
twice and
format in a very simple way, like this:
fun formatAfterTwice(x: Int) = format(twice(x))
You can prove this by adding the
formatAfterTwice definition in the Main.kt file along with the following:
fun main() { println(format(twice(37))) println(formatAfterTwice(37)) }
When you run this code, you get the following output:
Result is 74 Result is 74
This is just an example, but proving that this works for any type and function requires something more.
Open the initially empty Category.kt file in the project for this chapter, and add the following definition:
inline infix fun <A, B, C> Fun<B, C>.after(crossinline f: Fun<A, B>): Fun<A, C> = { a: A -> this(f(a)) // HERE }
Note: In the previous code you use the Kotlin keywords
inline,
infixand
crossinline. In this case,
infixallows you to use a syntax like
g after finstead of
g.after(f). The
inlinekeyword, as you’ll see in Chapter 4, “Expression Evaluation, Laziness & More About Functions”, allows you to basically replace every
afterinvocation with the expression it represents. Using
inlinethen requires the use of
crossinlinefor the input parameter,
f, in order to allow no-local returns from the function
fitself.
For a full description of these keywords, please refer to Kotlin Apprentice, which is the best place to start.
This is a code representation of the
(g◦f) notation you were using before, where
g represents
this and
f represents
f.
The definition of
after has many interesting things to note. It:
- Is a generic function of the parameters types
A,
Band
C.
- Creates an extension function for the
Fun<B, C>type.
- Uses the
infixand
inlinekeywords.
- Accepts a function
Fun<A, B>as an input parameter and returns a function
Fun<A, C>as output.
Note: The last point asserts that
afteris a higher-order function because it accepts a function as an input parameter and returns a function as output. Don’t worry — you’ll learn about higher-order functions in Chapter 5, “Higher-Order Functions”.
The definition of
after looks more complicated than it actually is. Looking at its body, it does exactly what you’ve done with
formatAfterTwice but in a more generic way. It:
- Returns a function with an input parameter
aof type
A.
- Uses the parameter
aas input for the function
f.
- Passes the result of
f(a)as an input parameter for the function you use as the receiver, which in this case has type
Fun<B, C>.
You can now use
after with the previous example. Just add the following code to
main in Main.kt with:
main() { // ... val f: Fun<Int, Int> = ::twice // 1 val g: Fun<Int, String> = ::format // 2 val formatTwice = g after f // 3 println(formatTwice(37)) // 4 // ... }
Note: In the previous code, you used
::to reference a function using its name without calling it. For instance,
::twiceis a reference to
twice.
Here, you:
- Define
fas a reference to
::twiceof type
Fun<Int, Int>.
- Initialize
gas a reference to
::formatof type
Fun<Int, String>.
- Create
formatTwiceas a reference of type
Fun<Int, String>to
g◦f.
- Invoke
formatTwice, passing
37as a value.
Build and run
main, and you get the following additional output:
Result is 74
Exercise 2.4: In this section, you defined
after, which allows you to write expressions like:
val formatTwice = g after f
Can you write
composeinstead, which would allow you to implement the same expression as:
val formatTwice = f compose g
Again, give it a try and check the challenge project and Appendix B to see how you did.
It’s fundamental to note here the fact that
after compiles is proof that composition works for every type
A,
B and
C and every function
F<A, B> and
F<B, C>.
Proving associativity
To prove that the category of types and functions follows the associativity property, you basically need to prove that:
(h after g) after f == h after (g after f)
Open Main.kt and add the following function:
fun length(s: String): Int = s.length
Note: Here, you defined
lengthexplicitly, but you could use
String::lengthinstead.
It’s a simple function that returns the length of the
String you pass as an input parameter. Its type is
Fun<String, Int>.
In the same Main.kt file, add the following code in
main:
fun main() { // ... val h: Fun<String, Int> = ::length // 1 val leftSide: Fun<String, Int> = (h after g) after f // 2 val rightSide: Fun<String, Int> = h after (g after f) // 3 println(leftSide(37) == rightSide(37)) // 4 }
In this code, you:
- Define
has the reference to
lengthas a function of type
Fun<String, Int>.
- Create
leftSideas the left side of the equation you want to prove.
- Define
rightSideas the right side of the equation you want to prove.
- Check that the two members are equal.
When you run
main, you get the following additional output:
true
You might object again that this specific example doesn’t prove anything — and you’re right! What actually proves that associativity works for the category of types and functions is the successful compilation of
after.
This is because it means that, given the types
A,
B and
C and the functions
f and
g, you can always create a function that is their composition.
Another way to prove it is to replace the definition of
after with its implementation. In this case, the left side of the equation is:
(h after g) after f = ({ b: B -> h(g(b))}) after f = { a: A -> { h(g(f(a)))}}
The right side is:
h after (g after f) = h after ({ a: A -> g(f(a))}) = { a: A -> { h(g(f(a)))}}
The two members are exactly the same.
Note: In the last proof, you actually applied fundamental tools you’ll learn about in detail in Chapter 3, “Functional Programming Concepts”.
Proving identity
What does it mean to prove identity for the category of types and functions? It means to prove that, for every type
A, there’s always at least one function of type
Fun<A, A>. To create such a function, open the Category.kt file and add this code:
fun <A> identity(value: A) = value
Although this function proves that identity exists, it’s important to mention that it’s not the only one. The function
twice you created earlier, for instance, is an identity function because of type
Fun<A, A>.
At this point, there’s something interesting to say about identity. As mentioned earlier,
twice is an example of a function that maps distinct values in the domain to distinct values in the range. This means that
twice has an inverse function you can call
half and define like this:
fun half(a: Int): Int = a / 2
But what does it mean to say that
twice is the inverse of
half? This means that:
half after twice = twice after half = identity
Note: In this case, if you half a value and then double it, you get the same value. The same is true if you first double and then halve it.
When you have a function and an inverse function, you say that the functions are isomorphisms. This definition gives a first reason for the existence of the identity property. Of course, this isn’t true for all the functions from a type
A to a type
B. Some functions don’t have an inverse.
Anyway, it’s noteworthy that an isomorphic function somehow makes identity obsolete because it would just be the composition of the function with its inverse.
Exercise 2.5: Can you write an example of an isomorphic function
fand its inverse
gand prove they always compose to
identity? When you’re done, look at the solution in Appendix B.
This concludes the proof that types and functions creates a category. What, then, is the meaning of starting and terminal objects for the category of types and functions?
Initial and terminal objects
At this point, it’s interesting to see what the meaning of initial and terminal object is in the category of types and functions.
The initial object
As you learned earlier, the starting point is an object with a unique morphism compared to all other objects in the category. With types and functions, it means that the initial point is a type you can use as input to a unique function to any other type. Said in other words, any function that you call with this initial type must always have the same result.
In Kotlin, this initial object is
Nothing. You might argue that you could always write a function from
Int to any other type, and this is true. The problem is that the function must be unique.
If
Int were a starting point, the following two different functions of type
Fun<Int, String>:
val f: Fun<Int, String> = { a: Int -> "2 * $a = ${2*a}"} val g: Fun<Int, String> = { a: Int -> "$a * $a = ${a*a}"}}
Would be the same function, but they aren’t! Instead, for any type
A you define:
fun <A> absurd(a: Nothing): A = a as A
This function’s name is
absurd because, if you want to invoke it, you need a value of type
Nothing, which is impossible. IntelliJ gives you a hint for this:
Note: You might wonder why some of the functions have names like
absurdand others are anonymous, like
fand
gin the last example. The reason is that anonymous functions cannot be generic.
You might try to invoke it like this, giving the Kotlin compiler some hint about the generic type, but the result would be the same.
More importantly, because
absurd is a function you can’t invoke, it never returns, and this makes it unique. Any other function starting from
Nothing to any other type
A would do exactly the same, so it’d be the same as
absurd.
The terminal object
In the previous section, you learned where the famous
Nothing Kotlin type comes from. In the category of logic, you also learned that the terminal object is the counterpart of the initial object. A terminal object, if it exists, is an object with a unique incoming morphism from all other objects. In other words, all functions that return this type will return the same object.
In the category of types and functions, this means a terminal object is a type output of a unique function accepting an input parameter of any other type,
Nothing included. In Kotlin, this is the
Unit type.
For any type
A, you can create the function:
fun <A> unit(a: A): Unit = Unit
Also, in this case, there are many significant things to note:
- The
unitfunction is generic with the type
A, and it always returns something you represent with
Unit.
- For any type
A, the function
unitis unique.
Unitisn’t just a type; it’s also the only value of that type.
To disprove uniqueness, you might try to create the following function as a possible alternative. This wouldn’t work:
fun <A> unit2(a: A): Unit { println("I'm different") return Unit }
At the beginning of the chapter, you learned that the concept of function isn’t strongly related to the concept of computation. A function is a way to map values in the domain to values in the range. In terms of mapping,
unit and
unit2 are the same function.
Note:
unit2is actually a bad function because it hides a side effect, which is another fundamental concept you’ll learn about in the next chapter.
The last point is also essential. You learned that with category theory, you can only use objects and morphisms. What does it mean to say that
Unit is the only value for the
Unit type? What does it mean that
a is a value of type
A? In general, what’s the relationship between the concept of type and the probably more familiar concept of set?
Types and sets
In the previous section, you learned some interesting properties of the category where objects are types and morphisms are functions. But what is a type? What do you mean when you define a variable like this?
var a: Int = 10
You usually read this as “
a is an
Int”, but what do you mean by
Int? If you think about this,
Int is just a name to represent the set of all the integer values.
a is of type
Int means that you can assign to
a only values that are in the set you call
Int.
Note: To be accurate,
Intrepresents the set of all the possible integers that you can represent using 4 bytes, so the whole numbers between -2,147,483,648 (-2³¹) and 2,147,483,647 (2³¹-1).
In the same way, consider the following expression:
var s: String = "Hello"
In this case, you’re defining the variable
s of type
String, which is a way to represent the set of all possible
Strings. Again, you can assign to
s only elements of the set you call
String.
Thinking in terms of sets makes things easier. Consider the function
twice that you’ve already met:
fun twice(a: Int): Int = a * 2
This is basically a way to map elements in the set of
Int to other elements in the same set.
The function
format is a way to map elements in the set of
Int to elements in the set of
String.
fun format(b: Int): String = "Result is $b"
The following image gives a more intuitive representation of the previous functions:
Figure 2.14 can be a bit misleading, though, because it represents relations between elements of two
sets, which in this case are both
Int. But what does it mean to be an element of a set? How can you represent this using just objects and morphisms?
Definition of element
As you learned earlier, objects in category theory are like dots; they cannot be decomposed. This also means the morphisms are the only tool you can use to distinguish one object from the others.
In particular, category theory introduces the concept of structure, which you define using the incoming morphism. Because of this, the initial object has no structure. The terminal object, on the other hand, has a very clear and unique structure because there’s exactly one and only one morphism from any other object. This property makes the terminal object unique.
Nobody said that a terminal object can’t have outgoing morphisms. On the contrary, from the terminal object, you might have many morphisms to a given object
A, like in Figure 2.15:
More importantly, you can give a different name to each morphism, like
x,
y or
z.
You need to understand what this means for the category of types and functions. This means that for a given type
A, there are many functions of type
Fun<Unit,A>, one for each value of type
A.
As a simple example, consider the type
Int. You can create a different function of each value of
Int, like these you can add to the Main.kt file:
fun one(u: Unit): Int = 1 fun two(u: Unit): Int = 2 fun minusThree(u: Unit): Int = -3 // ...
A function for each element of the set you associate to the type
Int and the name you give to that function is an element of the set
A represents.
To see how to describe a simple invocation of a function using composition, just add the following code to
main:
fun main() { // ... // twice 2 val twiceTwo = ::twice after ::two // 1 println(twiceTwo(Unit)) // 2 // ... }
In that code, you:
- Define
twiceTwoas the composition of
twoand
twice.
- Invoke
twiceTwo, passing
Unitas a parameter.
When you run that code, you’ll get the expected output:
4
Initial and terminal objects using sets
Thinking in terms of sets makes things a little bit easier. The initial object, by definition, doesn’t have any incoming morphisms. It’s the type for a set without any elements.
Nothing is like the empty set. If you consider the concept of subtype related to the concept of inheritance, you also understand why
Nothing is a subtype of every other type in Kotlin. This is because the empty set is a subset of every other set. If you look at the source code for the
Nothing type, you’ll find this:
public class Nothing private constructor()
Nothing has a private constructor. This means that
Nothing is a type, but you can’t create any instances of it.
What about the terminal object? By definition, there’s a unique morphism from any other type. This means there’s a unique function from any other type. If you consider the terminal type to be a set, this means it’s a set containing a single value that you know, in Kotlin, has the name
Unit. Look at its source code, and you’ll find this:
object Unit { override fun toString() = "kotlin.Unit" }
This means that
Unit is a type, but it’s also the name for its unique instance.
Function types
In one of the previous paragraphs, you created the following definition:
typealias Fun<A, B> = (A) -> B
With
Fun<A, B>, you wanted to represent the set of all the functions from
A to
B.
According to what you learned previously, this is exactly the concept of type of a function. You’ve already learned how the category of types and functions works. If the objects of that category are types of functions, what would the morphisms be? This is something you’ll learn in the following chapter, and in particular, with the concept of functor.
Challenges
In this chapter, you learned some fundamental concepts about category theory and functional programming. It’s now time to solve some challenges and have some fun. You’ll find the solutions in Appendix B, and the code in the challenge project for this chapter.
Challenge 1: Functions and sets
How would you represent a specific
Set using a function? For instance, how would you represent the set of even numbers with a function? After that, how would you print all the values in the set?
Challenge 2: Functions and sets again
How would you represent the intersection and union of two sets using functions? The intersection is the set of objects that belong to set A and set B, and the union is the set of all objects that belong to set A or set B.
Challenge 3: The right domain
Consider the following function:
fun oneOver(x: Int): Double = 1.0 / x
What’s the domain and the range for this function? If you invoke
oneOver(0), you get an exception. How can you be sure you only pass values in the domain as an input parameter?
Key points
- A function is a way to map things from a domain to things in a range.
- Category theory is necessary for understanding the fundamental concepts of functional programming.
- You define a category using objects and morphisms, which must follow the fundamental rules of composition, associativity and identity.
- Logic is a category that makes some concepts closer to the way humans think.
- The category of types and functions is the most important for a software engineer because it explains functional programming concepts.
- The initial object in a category has outgoing unique morphisms to all other objects.
- The terminal object in a category has incoming unique morphisms from all other objects.
- Not all categories have a terminal object.
- Initial and terminal objects explain the meaning of the Kotlin standard types
Nothingand
Unit.
- It’s useful to think of a type for a variable as a way to represent the set of all the values that variable can reference.
- Understanding the relationship between types and sets simplifies some of the most critical functional programming concepts.
Where to go from here?
Congratulations! At this point, you have an idea of what a category is and why category theory is important in understanding functional programming. It might have been tough, but you don’t have to worry if something isn’t completely clear at this point. In the following chapters, you’ll have the opportunity to see these concepts again in a more practical context. In the next chapter, you’ll start to focus on pure functions.
Note: The inspiration for this chapter comes from the excellent work of Bartosz Milewski in his YouTube video series Category Theory for Programmers. | https://www.raywenderlich.com/books/functional-programming-in-kotlin-by-tutorials/v1.0/chapters/2-function-fundamentals | CC-MAIN-2022-21 | refinedweb | 7,187 | 63.39 |
Whenever I run the program, I get the following error:
NameError at /table/
name 'models' is not defined
from django.shortcuts import render
def display(request):
return render(request, 'template.tmpl', {'obj': models.Book.objects.all()})
from django.db import models
class Book(models.Model):
author = models.CharField(max_length = 20)
title = models.CharField(max_length = 40)
publication_year = models.IntegerField()
from django.conf.urls import url
from . import views
urlpatterns = [
# /table/
url(r'^$', views.display, name='display'),
]
You are referencing
models.Book in your view, but you have not imported
models. In your views.py you need to do
from myapp import models. Or you can do
from myapp.models import Book and change it in your view function to just
Book.objects.all(). | https://codedump.io/share/OPWOTFJ0Bwwl/1/why-am-i-getting-the-following-error-nameerror-name-39models39-is-not-defined | CC-MAIN-2017-17 | refinedweb | 122 | 57.23 |
Disclaimer: The information in this article & source code are published in accordance with the final [V1] bits of the .NET Framework
This sample shows how to use Windows XP Windows Image Acquisition (WIA) Scripting with .NET and C#. It is useful for integrating with scanners, digital cameras, webcams and still-video.
Note, this article doesn't save you from reading the detailed WIA documentation!
WIA is a standardized Win32 API for acquiring digital images from devices that are primarily used to capture still images, and for managing these devices. WIA was introduced with Windows Millennium and updated for Windows XP. Today, most digital imaging devices are supported on XP by built-in or manufacturer provided WIA drivers.
The API is exposed as COM interfaces, in two flavors:
A seamless and low-effort integration with .NET is (currently) only possible with WIA Scripting.
The WIA Scripting Model is described in the WIA Scripting reference on MSDN Platform SDK. The typical steps needed to acquire an image are as follows:
WIA Manager
WIA manager
WIA root device item
WIA image items
With pseudo code this looks as simple as:
manager = new Wia
root = manager.Create
collection = root.GetItemsFromUI
collection[0..n].Transfer
WIA provides its own common dialogs to select a device:
and e.g. dialogs specific to a scanner device, ... or specific to a photo camera device:
Note, some tasks are also possible without user (GUI) interaction. One such method is Item.TakePicture but you should check the documentation for a complete list.
Item.TakePicture
The WIA Scripting Model organizes all known items (devices, folders, pictures,...) in a hierarchical structure, e.g.: WIA Camera Tree. An advanced application can recursively enumerate this tree using the Item.Children property:
Item.Children
root e.g. Camera Device
+root.Children
item1
item2 e.g. Folder
+item2.Children
item21 e.g. Picture1
item22 e.g. Picture2
To use WIA Scripting in your Visual Studio .NET project, add a reference to the component "Microsoft Windows Image Acquisition 1.01 Type Library" (wiascr.dll)
Without VS.NET, you will have to use the TLBIMP tool.
Now you can add the WIA Scripting namespace at the top of your C# code file:
using System.IO;
using System.Runtime.InteropServices;
// namespace of imported WIA Scripting COM component
using WIALib;
This imported library namespace provides mapping of the WIA Scripting types to .NET wrapper classes:
Wia
WiaClass
DeviceInfo
DeviceInfoClass
Item
ItemClass
CollectionClass
and now you can write code like this simplified sample to acquire pictures:
// create COM instance of WIA manager
wiaManager = new WiaClass();
object selectUsingUI = System.Reflection.Missing.Value;
// let user select device
wiaRoot = (ItemClass) wiaManager.Create(
ref selectUsingUI );
// this call shows the common WIA dialog to let
// the user select a picture:
wiaPics = wiaRoot.GetItemsFromUI( WiaFlag.SingleImage,
WiaIntent.ImageTypeColor ) as CollectionClass;
// enumerate all the pictures the user selected
foreach( object wiaObj in wiaPics )
{
wiaItem = (ItemClass) Marshal.CreateWrapperOfType(
wiaObj, typeof(ItemClass) );
// create temporary file for image
imageFileName = Path.GetTempFileName();
// transfer picture to our temporary file
wiaItem.Transfer( imageFileName, false );
}
For more information on COM interop and WIA debugging read these two articles:
Some devices, especially photo-cameras on serial COM ports, are slow! So it will take many seconds to transfer pictures. WIA Scripting solves this issue with an asynchronous flag when using Item.Transfer:
Item.Transfer
// asynchronously transfer picture to file
wiaItem.Transfer( imageFileName, true );
To notify the application about the completed transfer and more, the WIA manager exposes three events:
OnTransferComplete
OnDeviceDisconnected
OnDeviceConnected
These events nicely map on to the .NET event/delegate model, so we add a handler function like so:
// delegate member variable
private _IWiaEvents_OnTransferCompleteEventHandler
wiaEvtTransfer;
...
// create event delegate
wiaEvtTransfer = new _IWiaEvents_OnTransferCompleteEventHandler(
this.wia_OnTransferComplete );
wiaManager.OnTransferComplete += wiaEvtTransfer;
...
// event handler function (callback from WIA!)
public void wia_OnTransferComplete(
WIALib.Item item, string path )
{
... // logic to handle completed transfer
}
For web cams or other video devices, WIA on Windows XP has a great new feature: live video stream overlay! It uses DirectShow to draw the overlay on the graphics card. The IWiaVideo interface can be accessed by importing another COM type library: "WiaVideo 1.0 Type Library" (wiavideo.dll). Unfortunately, the embedded TLB has a bug for methods passing a window handle. I used ILDASM to get the IL code of the interop assembly, then I changed all incorrect occurances of 'valuetype _RemotableHandle&' to 'native int', then finally compiled back to an assembly with ILASM. This repaired DLL is included in the download as WIAVIDEOLib.dll.
DirectShow
valuetype _RemotableHandle&
native int
The code to show a real-time and live video stream in a window could be as simple as these two steps:
wiaVideo = new WiaVideoClass();
wiaVideo.CreateVideoByWiaDevID(
wiaDeviceID, window.Handle, 0, 1 );
and to take a snapshot jpeg image from the current video stream is possible with one single method:
// this string will get the filename of the picture...
string jpgFile;
// call IWiaVideo::TakePicture
wiaVideo.TakePicture( out jpgFile );
Check the included video sample for more!
Full source code with Visual Studio .NET projects are provided for these samples:
Item.Thumbnail. | http://www.codeproject.com/Articles/2303/WIA-Scripting-and-NET?fid=3908&df=90&mpp=10&sort=Position&spc=None&tid=3977846 | CC-MAIN-2014-52 | refinedweb | 833 | 51.04 |
I'm using two gertbot h-bridge boards and a Raspberry pi 2 model B to control 3 stepper motors with python 3. I've freed up the uart and can control my motors by using Gert's GUI. I can also control the motors with my own python code BUT this only works if I've first run Gert's GUI and hit the 'connect' button.
Looking at the Gerbot.py functions that I've been using I can see that it is the 'read_uart' call that sometimes returns -1 for the first gertbot board and always returns -1 for the second gertbot board. If I remove the try, except blocks I see that the specific error is 'OSError: [Errno 11] Resource temporarily unavailable'. If I have first connected to the boards with Gert's GUI (without using this to set motor modes etc, just simply clicking connect and then closing the GUI) my python code works fine. It appears that the connect button is doing something that my python code isn't. I've started looking at the C++ source code for the GUI but haven't yet found anything that fixes this problem.
Gert (or someone) do you have a solution or any suggestions?
Below is an example of my python code (works fine after first connecting with GUI) but motor_error is always [x, y, -1] (where x and y are sometimes 0 sometimes -1) if I haven't first connected with the GUI:
Any advice would be much appreciated!
Code: Select all
import gertbot as gb MOTOR_MODE = 8 #Step gray off STEPS_MM = 25 #Steps per mm of stepper motors GERTBOT_IDS = [0, 0, 1] #Gertbot IDs for each MOTOR MOTOR_IDS = [0, 2, 0] #Motor IDs for each GERTBOT ID STEP_FREQUENCY = 600 #Stepper motor frequency in Hz def configure_gertbots(gertbot_ids, motor_ids, motor_mode, step_frequency): try: gb.open_uart(0) # Open serial port to talk to Gertbot motor_error = [0]*len(gertbot_ids) number_of_posts = len(gertbot_ids) for i in range(0, number_of_posts): # Setup as steppers gb.set_mode(gertbot_ids[i],motor_ids[i],motor_mode) # Set the step frequency gb.freq_stepper(gertbot_ids[i],motor_ids[i],step_frequency) # Set endstops gb.set_endstop(gertbot_ids[i],motor_ids[i],0,2) # set short mode gb.set_short(gertbot_ids[i],motor_ids[i],1) # test for errors for i in range(0, number_of_posts): motor_error[i] = gb.read_error_status(gertbot_ids[i]) print(motor_error) return sum(motor_error) except: return -1 configure_gertbots(GERTBOT_IDS, MOTOR_IDS, MOTOR_MODE, STEP_FREQUENCY)
Jochem | https://www.raspberrypi.org/forums/viewtopic.php?t=162756 | CC-MAIN-2019-18 | refinedweb | 399 | 51.99 |
Deploy FastAPI on Deta¶
Warning
The current page still doesn't have a translation for this language.
But you can help translating it: Contributing.
In this section you will learn how to easily deploy a FastAPI application on Deta using the free plan. 🎁
It will take you about 10 minutes.
A basic FastAPI app¶
- Create a directory for your app, for example
./fastapideta/and enter in it.
FastAPI code¶
- Create a
main.pyfile with:
from fastapi import FastAPI app = FastAPI() @app.get("/") def read_root(): return {"Hello": "World"} @app.get("/items/{item_id}") def read_item(item_id: int): return {"item_id": item_id}
Requirements¶
Now, in the same directory create a file
requirements.txt with:
fastapi
Tip
You don't need to install Uvicorn to deploy on Deta, although you would probably want to install it locally to test your app.
Directory structure¶
You will now have one directory
./fastapideta/ with two files:
. └── main.py └── requirements.txt
Create a free Deta account¶
Now create a free account on Deta, you just need an email and password.
You don't even need a credit card.
Install the CLI¶
Once you have your account, install the Deta CLI:
$ curl -fsSL | sh
$ iwr -useb | iex
After installing it, open a new terminal so that the installed CLI is detected.
In a new terminal, confirm that it was correctly installed with:
$ deta --help Deta command line interface for managing deta micros. Complete documentation available at Usage: deta [flags] deta [command] Available Commands: auth Change auth settings for a deta micro ...
Tip
If you have problems installing the CLI, check the official Deta docs.
Login with the CLI¶
Now login to Deta from the CLI with:
$ deta login Please, log in from the web page. Waiting.. Logged in successfully.
This will open a web browser and authenticate automatically.
Deploy with Deta¶
Next, deploy your application with the Deta CLI:
$ deta new Successfully created a new micro // Notice the "endpoint" 🔍 { "name": "fastapideta", "runtime": "python3.7", "endpoint": "", "visor": "enabled", "http_auth": "enabled" } Adding dependencies... ---> 100% Successfully installed fastapi-0.61.1 pydantic-1.7.2 starlette-0.13.6
You will see a JSON message similar to:
{ "name": "fastapideta", "runtime": "python3.7", "endpoint": "", "visor": "enabled", "http_auth": "enabled" }
Tip
Your deployment will have a different
"endpoint" URL.
Check it¶
Now open your browser in your
endpoint URL. In the example above it was, but yours will be different.
You will see the JSON response from your FastAPI app:
{ "Hello": "World" }
And now go to the
/docs for your API, in the example above it would be.
It will show your docs like:
Enable public access¶
By default, Deta will handle authentication using cookies for your account.
But once you are ready, you can make it public with:
$ deta auth disable Successfully disabled http auth
Now you can share that URL with anyone and they will be able to access your API. 🚀
HTTPS¶
Congrats! You deployed your FastAPI app to Deta! 🎉 🍰
Also notice that Deta correctly handles HTTPS for you, so you don't have to take care of that and can be sure that your clients will have a secure encrypted connection. ✅ 🔒
Check the Visor¶
From your docs UI (they will be in a URL like) send a request to your path operation
/items/{item_id}.
For example with ID
5.
Now go to.
You will see there's a section to the left called "Micros" with each of your apps.
You will see a tab with "Details", and also a tab "Visor", go to the tab "Visor".
In there you can inspect the recent requests sent to your app.
You can also edit them and re-play them.
Learn more¶
At some point you will probably want to store some data for your app in a way that persists through time. For that you can use Deta Base, it also has a generous free tier.
You can also read more in the Deta Docs. | https://fastapi.tiangolo.com/tr/deployment/deta/ | CC-MAIN-2021-17 | refinedweb | 650 | 74.08 |
Wikiversity:Subpages
Wikiversity allows creation of subpages for articles in the main namespace. Subpages are given names such as "Page/Subpage". Subpages are used to help organise individual projects such as courses, seminars, and research projects.
Example of a main namespace page with a subpage.
Contents
Courses[edit]
Courses are learning projects that can be joined at any time. Courses can have subpages like:
- Source materials (texts and articles)
- Lists
- Subtopics
- Seminars (see #Seminars)
- Help desk (for list of "tutors")
- Participants (students or "scholars")
- Quizzing and testing materials
- Papers, etc.
- Papers by students
- Essays by students or experts (including a course instructor) that include opinion, that need not be neutral, or that include original research or personal experience.
Classes[edit]
Classes are guided by instructors. These take place over a certain time frame. (Maybe not needed)
Classes on Wikiversity could use subpages for some or all of the following:
- Source materials
- Instructors
- Rosters
- Syllabi
- Quizzes
- Tests
- Study guides
- Papers
- Papers by students
- Reflections by students
- Assignments by students
- Whatever else you might find useful. Please be creative and innovative. Wikiversity is somewhat of an experiment!
For example you could use a page like Introduction to Engineering/Student848/Reflection 1, Introduction to Engineering/Roster, Introduction to Engineering/Quiz 1, Introduction to Engineering/Exam 1 study guide/Answers, etc. Please use subpages so long as they are useful!
Seminars[edit]
Seminars are egalitarian. If part of a course, student-led, though experts may participate. These may be split into projects, which may or may not have time limits.
- Instructors
- Projects
- Source materials
- Other materials (for art projects, etc.)
- Submission and critique
- Discussion
Research[edit]
- Brainstorming
- Different sections of research (results, methods, introductions, etc.)
- Publishing papers
- Allowing for peer review
- Start a peer reviewed journal at Wikiversity
Userpages[edit]
(learning) Projects[edit]
- Organization, organization, organization
Please note that Lunar Boom Town and CisLunarFreighter makes copious use of subpages.
Once you have an account here at Wikiversity for a few days or so, the move function can be a useful tool! It is usually better to move a page than to copy the page content under a new page name. This keeps the history intact.
(Page moves can sometimes be difficult to undo. A page move ordinarily leaves a redirect in place, and "double redirects," where a redirect links to a redirect, don't work, so some cleanup may be required. If the redirect has been edited, the move cannot be simply undone, a custodian is required. So be careful when moving pages.)
Subpage(s) to this project page[edit]
What could be more appropriate for a page on subpages than an illustration of the topic with a subpage of its own?
Forking and organizing (permalink) is an essay as written by one author (as you will see at the top of the page). The permalink is a "permanent version." That version cannot be changed, though it can be hidden by a custodian. The current version is shown by /Forking and organizing. As the essay is not attributed (except in page history), any user may edit it.
See w:Help:Permanent link for information about how to find and use permanent links. | https://en.wikiversity.org/wiki/Wikiversity:Subpages | CC-MAIN-2017-13 | refinedweb | 528 | 56.35 |
Download Images from a Web Page using Python
Want to share your content on python-bloggers? click here.
In this article we will discuss how to download images from a web page using Python.
Table of Contents
- Introduction
- Get HTML content from URL
- Finding and extracting image links from HTML
- Downloading images from URL
- Complete Object-Oriented Programming Example
- Conclusion
Introduction
text here
Let’s see how we can quickly build our own image scraper using Python.
To continue following this tutorial we will need the following Python libraries: httplib2, bs4 and urllib.
If you don’t have them installed, please open “Command Prompt” (on Windows) and install them using the following code:
pip install httplib2 pip install bs4 pip install urllib
Get HTML content from URL using Python
To begin this part, let’s first import some of the libraries we just installed:
import httplib2 from bs4 import BeautifulSoup, SoupStrainer
Now, let’s decide on the URL that we would like to extract the images from. As an example, I will extract the images from the one of the articles of this blog:
url = ''
Next, we will create an instance of a class that represents a client HTTP interface:
http = httplib2.Http()
We will need this instance in order to perform HTTP requests to the URLs we would like to extract images from.
Now we will need to perform the following HTTP request:
response, content = http.request(url)
An important note is that .request() method returns a tuple, the first being an instance of a Response class, and the second being the content of the body of the URL we are working with.
Now, we will only need to use the content component of the tuple, being the actual HTML content of the webpage, which contains the entity of the body in a string format.
Finding and extracting image links from HTML using Python
At this point we have the HTML content of the URL we would like to extract links from. We are only a few steps away from getting all the information we need.
Let’s see how we can extract the image links:
images = BeautifulSoup(content).find_all('img') image_links =[] for image in images: image_links.append(image['src'])
To begin with, we create a BeautifulSoup() object and pass the HTML content to it. What it does is it creates a nested representations of the HTML content.
Then, we create an empty list (image_links) that we will use to store the image links that we will extract from the HTML content of the webpage.
As the final step, what we need to do is actually discover the image links from the entire HTML content of the webapage. To do it, we use the .find_all() method and let it know that we would like to discover only the tags that are actually image links.
Once the script discovers the URLs, it will append them to the links list we have created before. In order to check what we found, simply print out the content of the final list:
for link in image_links: print(link)
And we should see each image link printed out one by one.
Downloading Images from a Web Page using Python
In this step we will use the image links we found in the above steps to download and save the images.
Let’s start with importing the required library:
import urllib.request
Next, we will iterate through the image_links list and download each image:
for link in image_links: filename = link.split("/")[-1].split("?")[0] urllib.request.urlretrieve(link, filename=filename)
Note: your string splitting for filename can be different depending on the original image link.
You should see the images being saved in the same folder as your Python file.
Complete Object-Oriented Programming Example
class Extractor(): def get_links(self, url): http = httplib2.Http() response, content = http.request(url) images = BeautifulSoup(content).find_all('img') image_links=[] for image in images: image_links.append(image['src']) return image_links def get_images(self, image_links): for link in image_links: filename = link.split("/")[-1].split("?")[0] urllib.request.urlretrieve(image_url, filename=filename)
And this is an example of getting images from a web page using the above class:
url = '' myextractor = Extractor() image_links = myextractor.get_links(url) myextractror.get_images(image_links)
Conclusion
This article introduces the basics of how to download images from a web page using Python httplib2, bs4 and urllib libraries as well as created a full process example.
Feel free to leave comments below if you have any questions or have suggestions for some edits and check out more of my Python Programming articles.
The post Download Images from a Web Page using Python appeared first on PyShark.
Want to share your content on python-bloggers? click here. | https://python-bloggers.com/2021/02/download-images-from-a-web-page-using-python/ | CC-MAIN-2022-40 | refinedweb | 786 | 60.35 |
Hi everyone.
I have been using Linux since nearly 3 years and recently, during a
reading on computer security i came up on the following question: Is my
computer and my private life really secure ?
Indeed not. my private life isn't 100 % secure and I wish I could make
it a little safer from intruders.
Considering I am running Linux, what would you do at first to make my
system safer from intruders ? I mean by intruders: ad wares, personal
infos gathered by web servers and so on... I am currently using 2
e-mails addresses (1 used for MSN, websites, forum, etc and another one
used to send and receive important mails). I consider that PGP would be
a great choice as a encryption program (mail). Mercury is absolutely
necessary when chatting on MSN. Using it allows to encrypt conversations.
If you know any way or hints to get aware from threats on Internet or
tools to encrypt my data, write me back. There are so much information
gathered about users on the WWW.
So. If you have any tutorials/links about security concerning Linux,
please post them :)
Cheers,
And... Sorry for my awful English.
--
/*
* This function is used through-out the kernel (includeinh mm and fs)
* to indicate a major problem.
*/
#include <linux/kernel.h>
volatile void panic(const char * s)
{
printk("Kernel panic: %s\n\r",s);
for(;;);
}
-=[Penguin_X]=- | http://fixunix.com/security/17688-how-secure-my-computer-print.html | CC-MAIN-2015-06 | refinedweb | 233 | 63.8 |
There are more than 82 crypto wallets available for you to use for your next smart contract or DApp. However, not all of these wallets are usable; some cannot be integrated with your smart contract, or don’t offer software development kits (SDKs). A good wallet for a smart contract supports blockchain interoperability, and is chain agnostic and has multi-session functionality.
Finding a wallet that has the above features from the pool of 82 wallets is not easy. So, I have compiled a list of the best wallets that you can use to integrate with your smart contracts and DApp, and included a review of everything you should look for while choosing the right wallet for your use case.
Contents
- What is a smart contract, and why are blockchain wallets necessary?
- Argent
- Portis
- Coinbase Wallet
- MetaMask
- Ledger
- WalletConnect
- How to choose a crypto wallet for your DApp or smart contract
What is a smart contract, and why are blockchain wallets necessary?
A smart contract is an agreement between two parties that states which actions will be executed automatically in the blockchain network when a certain condition has been satisfied. This agreement will be translated into a program typically using Solidity.
Sometimes, smart contracts have to pay or receive crypto from another party when a certain condition is met. This is where crypto wallets need to be integrated with your DApp.
A crypto wallet is software storage that keeps your cryptocurrency and tokens. This wallet also facilitates cryptocurrency transactions, and is not limited to single cryptocurrency type.
In the following sections, we will review five wallets that your smart contracts and DApp can use to carry out crypto transactions.
Argent
Argent is unarguably the best smart contract wallet, as it promotes low gas fees and gives you a free ENS that replaces a long wallet address such as
3FZbgi29cpjq2GjdwV8eyHuJJnkLtktZc5 with “Keharley.eth,” which is an ENS (Ethereum name service) address. ENS addresses are easy to remember and look good to non-geeks when compared to the generic long address.
Argent supports multi-signature technology which strengthens the wallet’s security. You also do not need a seed phrase when recovering your wallet.
Argent only specializes in facilitating smart contracts, but you can connect it to a decentralized application using WalletConnect. You can view their developer documentation here and GitHub repository here.
Portis
Portis can be easily integrated with your decentralized applications and smart contracts with only four lines of code:
import Portis from '@portis/web3'; import Web3 from 'web3'; const portis = new Portis('YOUR_DAPP_ID', 'mainnet'); const web3 = new Web3(portis.provider);
When using these wallet libraries, make sure that you do not hardcode the wallet DApp ID or any secrets associated with the wallet.
Portis uses a password and email address to authenticate users without using the private key. It can receive and send multiple cryptocurrencies, but the biggest highlight is that it accepts debit and credit cards; it’s not only focused on crypto.
Portis supports blockchain interoperability, and the following blockchains:
- Ethereum
- Bitcoin
- EOS
- SKALE
- Matic
- Ubiq
- Thundercore
Coinbase Wallet
Coinbase Wallet (formerly known as WalletLink) is used to connect with other Coinbase wallets. Only use this wallet if you already have a Coinbase account and your DApp is only making transactions with other Coinbase accounts.
You can learn how to integrate the Coinbase Wallet SDK here. Coinbase Wallet has a Node.js package you can install using the following command:
#npm install @coinbase/[email protected]
You can also use the yarn package manager like so:
#yarn add @coinbase/[email protected] yarn add @coinbase/wallet-sdk
Finally, one of the benefits of Coinbase Wallet is its chrome extension that enables you to see live crypto charts and manage NFTs.
MetaMask
MetaMask is popular even though its extension has several errors and issues. Using Metamask for your DApp transactions is convenient, because many people are familiar with it and there are numerous troubleshooting articles.
Metamask has an Ethereum Provider API you can use to request the user’s account and prompt them to sign a transaction from your DApp. You can view the full developer documentation here.
Ledger
Ledger is a hardware wallet, which makes it different from all of the previous wallets. But, it does come with software you can install on your mobile phone to interact with the physical wallet. Just like any other wallet, Ledger also allows you to make transactions and sell all your crypto assets.
Ledger offers an SDK you can use to communicate with the wallet. You can learn how to integrate their wallet here.
WalletConnect
WalletConnect is not necessarily a wallet, but rather a protocol used to connect a wallet and smart contract or DApp. It offers blockchain interoperability, supports different chains, and comes with session management. It also has a blockchain SDK that allows developers to smoothly connect their applications with WalletConnect. This SDK is comprised of a push server and QR codes. WalletConnect, you only need three lines of code and project ID to integrate it with your DApp:
import WalletConnectClient from "@walletconnect/client"; const client = await WalletConnectClient.init({ projectId: "enter your project id here", });
How to choose a crypto wallet for your DApp or smart contract
Documentation and support
Poor documentation quality and support is the main reason why developers move from one wallet to another.
Before choosing a wallet, go to the wallet’s website and check their documentation page; ensure that each and every feature you plan on using is explained clearly. Most of these wallets have been around for less than five years, so they don’t have abundant tutorials made by various contributors. Because of this, documentation is the only place you will get information on how to use the wallet’s SDK.
Also, be sure to join the wallet’s community, such as Slack workspace or Discord server. Check if the team is active in answering questions or attempting to solve problems raised by wallet users.
It’s inevitable to get errors and bugs when using a new Web3 library, so the goal is to choose a wallet that has abundant documentation and has many problem solutions online.
SDKs and maintainability
JavaScript is widely used in this space, so if you plan to use JavaScript frameworks for your DApp’s frontend, then you have many wallet options. For Android development, Kotlin has replaced Java, and wallet SDKs prefer to be integrated using Kotlin anyway. As for iOS, Swift is still the preferred language.
A wallet SDK with high quality documentation will be easy to maintain. Also, the SDK should use simple techniques and simple troubleshooting methods for developers. It is imperative to check if the wallet SDK is consistently being updated to improve security and usability.
A wallet that is easy to maintain should have the following
- Code that is easy to understand
- The ability to be repaired quickly and without struggle
- Clean code and high quality code standards
Wallet scalability
When looking for the scalability of a wallet, you should ask the following questions: How many cryptocurrencies does the wallet support? How much are the transactions charges? Does it support multi-chain or chain agnostic technology?
If the wallet has low gas fees and provides positive answers for the above questions, then it is easily scalable. It will also also allow scaling your DApp and smart contract, because the wallet will not limit you to a small handful blockchain networks.
Authentication and security factors
Secure wallets should be easy to authenticate and provide two factor authentication. Wallets such as Authereum can be used to authenticate against other websites.
It is important to check the security integrity of the wallet. A secure wallet should be compliant with the Cryptocurrency Security Standard. These security standards are a sign that the wallet is secure and can be trusted.
Also, be sure to look for any security breaches that may affect your smart contracts in the future. The wallet also needs to have a good reputation and be used by different blockchain projects that prove its trustworthiness.
Conclusion
After choosing a wallet for your smart contract, check the wallet’s performance and growth. Some wallets will close unexpectedly, and its important to make sure your DApp doesn’t suddenly lose functionality.
Hopefully this article has given you a few tips and tricks for choosing a blockchain wallet that is best for your smart contract or DApp, and introduced you to a few solid options. Feel free to leave any comments or questions. | https://blog.logrocket.com/5-blockchain-wallets-smart-contract-dapp/ | CC-MAIN-2022-40 | refinedweb | 1,415 | 52.19 |
Process.Start works great on windows, and very simply as well. But it seems a bit funkier on OSX.
I got it working to just launch an application normally. But I cannot figure out how to get it to launch an application with arguments on OSX. More specifically I am trying to launch a command line application with arguments. How do I do that??
I've already tried this:
Process p = System.Diagnostics.Process.Start(new ProcessStartInfo(info.FullName, " -fOGLPVRTC4 -q5 -pvrtciterations8 -i" + filePaths[i] + " -o" + filePaths[i]) { UseShellExecute = false });
it throws a win32 exception
I found the tip to add '{ UseShellExecute = false }' as I did there on some other mono mac forum. So I put it in there, if you leave that line out however it does nothing.
Answer by gamesbyangelina
·
Jun 24, 2013 at 01:52 AM
This is an old question but comes up prominently in Google. I want to offer an answer! I was trying to execute a command which aliased to another binary somewhere in my Mac. What I ended up doing was explicitly writing in the path to the binary. i.e. instead of:
Process p = new Process();
p.StartInfo.FileName="inkscape";
We have:
Process p = new Process();
p.StartInfo.FileName="/Applications/Inkscape.app/Contents/Resources/bin/inkscape";
I hope this helps some people as this was driving me nuts.
It works but doesn't seem to work with processes outside Applications folder, probably due to user permissions.
i've tried calling openDiff without success, it's a symbolic link to xcrun which is inside /usr/bin.
Answer by Inhuman Games
·
Feb 06, 2014 at 10:23 AM
Did you try inserting "--args " at the beginning of your arguments string? This solved the issue for me.
Thank you!
Answer by DanDixon
·
Sep 22, 2011 at 09:27 PM
This may not be helpful, but you could try this method to add arguments and set properties:
Process p = new Process();
p.StartInfo.FileName=command;
p.StartInfo.Arguments = arguments;
p.StartInfo.RedirectStandardError=true;
p.StartInfo.RedirectStandardOutput=true;
p.StartInfo.CreateNoWindow=true;
p.StartInfo.WorkingDirectory=Application.dataPath+"/..";
p.StartInfo.UseShellExecute=false;
p.Start();
I had problems getting Process to work at all until I used this method.
Note that not all of that is required. This works too:
Process p = new Process();
p.StartInfo.FileName=command;
p.Start();
And at the top of the cs file, add this:
using System.Diagnostics;
As discovered in this Unity forum thread about System.Diagnostics.Process.
I Can't get this to work on OSX, it doesn't throw an error just does nothing. Any reason that might be?
Answer by robertono
·
Jan 11, 2016 at 03:29 PM
Guys, I have been able to launch .app using Process.Start with UseShellExecute=TRUE, and not only from Applications folder. Specifficaly in inside my app: MyApp.app/Contents/Resources/ProcessStart.app
I don't know how it was before, but in Unity 5 it.
Open Terminal window for OSX System.Diagnostics.Process
0
Answers
Run Terminal Command using System.Diagnostics.Process not working on osx
2
Answers
Initialising List array for use in a custom Editor
1
Answer
Help? Problem with GUI Text button.
1
Answer
How do I replace Invoke Repeating to allow a repeat rate change
1
Answer | https://answers.unity.com/questions/161444/how-to-use-processstart-with-arguments-on-osx-in-a.html | CC-MAIN-2019-39 | refinedweb | 548 | 59.09 |
At the Forge - jQuery
to get all the tr tags with a class of even. Each call to $() might return zero, one or a number of objects matching that selector. That is, if you were to say:
$('#blahblah') // jQuery
and there isn't any such item, jQuery happily will return the set of elements matching that query—an empty set.
This might seem a bit weird at first. After all, don't you need to know in advance how many results you'll get, or even if there will be results at all? Otherwise, how can you know whether to call a function on the one returned element or to iterate over a set of elements and call the function on each of them?
Things make much more sense once you understand that many of jQuery's functions operate using what's known as implicit iteration. That is, you can say:
$('p').show();
and jQuery will grab all the paragraphs in a document and show them. If there aren't any paragraphs, nothing happens. If only one paragraph matches, it is shown (if it wasn't showing already). The idea that you don't have to use an each() loop to go through each element is a powerful one, and it repeats itself often in jQuery code.
Equally powerful is the fact that most jQuery methods return the same set they were passed. For example, if we want to show all paragraphs and then make their background color red, we could say:
$('p').show().css({'background-color': '#f00'});
Chaining methods together in this way is quite typical in jQuery, and it helps make code more readable and concise.
When I first saw the way jQuery uses $(), I realized this meant I probably wouldn't be able to use both jQuery and Prototype in the same program, because there would be a namespace collision between $() in each of the two systems. However, the authors of jQuery thought about this very issue and have made it possible to load jQuery without activating $(), making it possible to mix jQuery and Prototype in the same file:
jQuery.noConflict();
Although I'm not sure that it's a good idea to go into a project planning to mix JavaScript libraries, there are times when you might want to use a particular effect or widget, which is available in (for example) YUI, but not jQuery.
Like many other JavaScript libraries, jQuery comes with a set of visual effects that you can use on a page. We already have seen mention of show and hide, although each of these also can take an argument (slow, normal or fast, or an integer) indicating the speed at which the effect should run:
$('p').show('slow').css({'background-color': '#f00'});
Similarly, you can have elements hide and reveal themselves by sliding (slideUp and slideDown) or fading (FadeIn and FadeOut). And, of course, you always can modify one or more CSS attributes, as we saw in an earlier example, overriding their original settings.
It's easy to set an event handler in jQuery. For example, if you want to pop up an alert every time someone clicks on the button with an id attribute of mybutton, you would write:
$('#mybutton').bind('click', function() { alert("Hello, there!"); });
Because it is so common to bind an event handler to the click of a button, you can shorten it to:
$('#mybutton').click(function() { alert("Hello, there!"); });
The thing is, where do we put this event handler? We can't put it in the <head> of the document, because the <body> (in which the button presumably is contained) is not yet defined and available. We could assign it to a DOM handler, but there are issues associated with that. The jQuery method is both unobtrusive (as modern JavaScript should be), effective and cross-browser-compatible. We register an event handler for when the document is ready, and then put any event handlers we want in there:
$(document).ready(function() { $('#mybutton').click(function() { alert("Hello, there!"); }); }
If you put this in the <head> of your HTML file, jQuery will execute the function when the document is ready. This means by the time the HTML is rendered for the user, event handlers all will be in place. It is not uncommon for the invocation of $(document).ready() to contain the key JavaScript invocation code for a site, with long, complex functions placed in an external library.
Like other JavaScript libraries, jQuery also provides built-in support for AJAX (behind-the-scenes HTTP) requests. For example, we can make it such that clicking on a button sends an HTTP request to a server, grabs the returned HTML snippet and then puts that snippet in the user's browser window. To do that, we need an HTML file:
<html> <head> <link rel="stylesheet" type="text/css" media="screen" href="test.css"/> <script type="text/javascript" src="jquery.js"></script> <script type="text/javascript"> // JavaScript code here $(document).ready(function() { $('#mybutton').click(function() { $('#message-paragraph').load('blah.html'); }); }); </script> </head> <body> <h1>Test page</h1> <p id="message-paragraph">This is a paragraph</p> <p><input type="button" id="mybutton" value="Button" /></p> </body> </html>
In the head of the file, we have a standard call to $(document).ready, which assigns a handler to the click event on the button at the bottom of the page, whose id attribute is mybutton. The function, very simply, tells the paragraph (message-paragraph) to load the file blah.html from the same origin (that is, server) as the current page. The browser retrieves the file in the background, asynchronously, allowing the user to do other things while the contents of blah.html are retrieved and then stuck into the appropriate paragraph.
The above demonstrated that jQuery can retrieve HTML from an external file. But, jQuery can do more than that, retrieving not only HTML, but also XML and JSON (JavaScript Object Notation) over the network and then parsing it. We even can load a JavaScript file and execute it:
<html> <head> <link rel="stylesheet" type="text/css" media="screen" href="test.css"/> <script type="text/javascript" src="jquery.js"></script> <script type="text/javascript"> // JavaScript code here $(document).ready(function() { $('#mybutton').click(function() { $.getScript('blah.js'); }); }); </script> </head> <body> <h1>Test page</h1> <p id="message-paragraph">This is a paragraph</p> <p><input type="button" id="mybutton" value="Button" /></p> </body> </html>
Then, I create blah.js:
$('#message-paragraph').html("<h1>Boo!</h1>");
In other words, I've split the functionality from one file into two different ones. The difference is that the loaded file is treated as a program and is executed as such. So, when I click on the button, the contents of blah.js are loaded by jQuery, which then modifies the paragraph.. | https://www.linuxjournal.com/magazine/forge-jquery | CC-MAIN-2020-29 | refinedweb | 1,134 | 70.53 |
How to integrate jQuery with Tapestry 5.
Introduction
Unfortunatly it is not possible to use a current jQuery release out of the box with Tapestry 5. But only one small step is necessary to make them play together well.
It is necessary to adjust the jQuery sources, because Tapestry 5 uses the Prototype Javascript library. Both javascript libraries make use of the variable $ in the global javascript namespace. See the jQuery docs for a more detailed explanation of the problem.
Adjusting jQuery
1. Download the current jQuery release. Be sure to use the minimized version for production use
2. Open the file and append jQuery.noConflict(); on a separate line at the very end of the file
Your file should look like this:
/* * jQuery JavaScript Library v1.3.2 * * * Copyright (c) 2009 John Resig * Dual licensed under the MIT and GPL licenses. * * * Date: 2009-02-19 17:34:21 -0500 (Thu, 19 Feb 2009) * Revision: 6246 */ (function(){var l=this,g,y=l.jQuery,p=l.$,o=l.jQuery=l.$=function(E,F){return new ......... /* * Sizzle CSS Selector Engine - v0.9.3 * Copyright 2009, The Dojo Foundation * Released under the MIT, BSD, and GPL Licenses. * More information: */ (function(){var R=/((?:\((?:\([^()]+\)|[^()]+)+\)|\[(?:\[[^[\]]*\]|['"]........ /* * Must run in noConflict mode! */ jQuery.noConflict();
That's it. Now you can use this script like you use any other javascript on your Tapestry 5 pages.
Unfortunatly the $ name is still reserved for Prototype. The jQuery documentation has more information on how to overcome this limitation.
Creating a Component Library
You may want to create a component library for your adjusted jQuery source and the most useful plugins that you'll need in your projects. If you don't know how to do that, see Tapestry5HowToCreateYourOwnComponentLibrary
We have choosen to create a library that contains jQuery core, the UI plugin and the well known form plugin. Using any of these is as simple as this:
@IncludeJavaScriptLibrary({ "classpath:/x/y/z/jquery/components/jquery.min.js", "classpath:/x/y/z/jquery/components/jQuery.forms.js", } ) public class SomePageClass { }
Related Mailing List Threads
Thread integration jQuery components
Thread using jQuery in 5.0.18 | https://wiki.apache.org/tapestry/Tapestry5HowToIntegrateJQuery?action=diff | CC-MAIN-2017-30 | refinedweb | 353 | 59.09 |
Hi. I would like announce of new version of pyplusplus (0.5). What is this? The pyplusplus is a framework of components for creating C++ code generator for boost.python library Code generation with the pyplusplus framework is a very flexible and highly configurable process. You can find an introduction to pyplusplus at: 1. 2. Status: First of all, this project is under active development. Second I think from now this project could be used in production. Third: list of implemented features: * * namespace aliasing and using * writing multiple files * user code could be inserted almost any where * write code into file if there were changes * user licence is written at the top of every file * ... 4'th - examples. Now it has real world example. I created Python binding for boost.date_time library almost all functionality has been exported. Next functionality was not exported: facet, parsing and iterators. See this page for more information: 1. 2. 5'th - demo mode. In order to evaluate pyplusplus you don't have to learn new API. pyplusplus has nice and simple gui. For convenience this gui is packed as standalone executable for Windows and Linux. In order to use pyplusplus you have to have CVS version of boost.python. Ideas, comments, suggestions or help are welcomed. Roman Yakovenko | https://mail.python.org/pipermail/cplusplus-sig/2005-October/009419.html | CC-MAIN-2014-10 | refinedweb | 214 | 60.41 |
Thanks to the FileManager tutorial, I’ve figured out how to open a Finder window and I’ve made the first steps on my project. Now I’m trying to work out how to stash away an input file and an output directory for later use (it will be used to construct an ffmpeg script that will run if I want to re-encode or rewrite metadata on my audiobook files.)
Here’s what that part of my user interface will look like:
Here’s my code:
import Cocoa
class ViewController: NSViewController {
@IBOutlet weak var inputText: NSTextField! @IBOutlet weak var outputText: NSTextField! @IBOutlet weak var inputBrowseButton: NSButton! @IBOutlet weak var outputBrowseButton: NSButton! @IBAction func inputBrowseClicked(_ sender: Any) { let inputPanel = NSOpenPanel() inputPanel.canChooseFiles = true inputPanel.canChooseDirectories = false inputPanel.allowsMultipleSelection = false inputPanel.allowedFileTypes = ["mp3","m4b","aax","m4a"] let userChoice = inputPanel.runModal() switch userChoice{ case .OK : if let panelResult = inputPanel.url { //1) Loading a string from file inputString(from: panelResult) } case .cancel : print("user cancelled") default: break } } @IBAction func outputBrowseClicked(_ sender: Any) { let outputPanel = NSOpenPanel() outputPanel.canChooseFiles = false outputPanel.canChooseDirectories = true outputPanel.allowsMultipleSelection = false outputPanel.allowedFileTypes = ["mp3","m4b"] let userChoice = outputPanel.runModal() switch userChoice { case .OK: if let panelResult = outputPanel.url { //1) Saving a string to file outputString(to: panelResult) } case .cancel: print("saving cancelled") default: break } } func inputString(from loadURL: URL){ do { let inString = try String.init(contentsOf: loadURL) inputText.stringValue = inString } catch { print(error) } } func outputString(to saveURL: URL){ let outString = outputText.stringValue print(outString) // do { // try outString.write(to: saveURL, atomically: true, encoding: .utf8) // } catch { // print(error) }
}
When I click the “Browse” buttons, the NSOpenPanel function works, and it even filters for the types of files I want to work with. But when I click “open” the code thinks I’m trying to run the file instead of just stashing away the file URL. It doesn’t copy the path to the text field as it should. And pretty much nothing happens with the output Browse button except a finder window opening.
I’m so close to having this. If I can learn how to stash the file URL data away to use when constructing my ffmpeg script, I can figure out how to do it with other encoding-option data from radio buttons and drop-down boxes as well. It will involve lots of comparison. If the radio buttons match the bitrate/sampling rate/audio settings of the existing file, then probably all we’re tweaking is the metadata and the script will just copy the file without re-encoding the audio. But basically it all comes down to learning how to stash input information from the UI away at this point, right? | https://forums.raywenderlich.com/t/stashing-file-urls-away-for-later-use/84794 | CC-MAIN-2020-29 | refinedweb | 446 | 51.75 |
.
Facebook implements OAuth 2.0 as its standard authentication mechanism, but provides a convenient way for you to get an access token for development purposes, and we'll opt to take advantage of that convenience in this notebook. For details on implementing an OAuth flow with Facebook (all from within IPython Notebook), see the _AppendixB notebook from the IPython Notebook Dashboard.
For this first example, login to your Facebook account and go to to obtain and set permissions for an access token that you will need to define in the code cell defining the ACCESS_TOKEN variable below.
Be sure to explore the permissions that are available by clicking on the "Get Access Token" button that's on the page and exploring all of the tabs available. For example, you will need to set the "friends_likes" option under the "Friends Data Permissions" since this permission is used by the script below but is not a basic permission and is not enabled by default.
# Copy and paste in the value you just got from the inline frame into this variable and execute this cell. # Keep in mind that you could have just gone to # and retrieved the "User Token" value from the Access Token Tool ACCESS_TOKEN = ''
import requests # pip install requests import json base_url = '' # Get 10 likes for 10 friends fields = 'id,name,friends.limit(10).fields(likes.limit(10))' url = '%s?fields=%s&access_token=%s' % \ (base_url, fields, ACCESS_TOKEN,) # This API is HTTP-based and could be requested in the browser, # with a command line utlity like curl, or using just about # any programming language by making a request to the URL. # Click the hyperlink that appears in your notebook output # when you execute this code cell to see for yourself... print url # Interpret the response as JSON and convert back # to Python data structures content = requests.get(url).json() # Pretty-print the JSON and display it print json.dumps(content, indent=1)
Note: If you attempt to run a query for all of your friends' likes and it appears to hang, it is probably because you have a lot of friends who have a lot of likes. If this happens, you may need to add limits and offsets to the fields in the query as described in Facebook's field expansion documentation. However, the
A couple of field limit/offset examples that illustrate the possibilities follow:
fields = 'id,name,friends.limit(10).fields(likes)' # Get all likes for 10 friends
fields = 'id,name,friends.offset(10).limit(10).fields(likes)' # Get all likes for 10 more friends
fields = 'id,name,friends.fields(likes.limit(10))' # Get 10 likes for all friends
fields = 'id,name,friends.fields(likes.limit(10))' # Get 10 likes for all friends
import facebook # pip install facebook-sdk import json # A helper function to pretty-print Python objects as JSON def pp(o): print json.dumps(o, indent=1) # Create a connection to the Graph API with your access token g = facebook.GraphAPI(ACCESS_TOKEN) # Execute a few sample queries print '---------------' print 'Me' print '---------------' pp(g.get_object('me')) print print '---------------' print 'My Friends' print '---------------' pp(g.get_connections('me', 'friends')) print print '---------------' print 'Social Web' print '---------------' pp(g.request("search", {'q' : 'social web', 'type' : 'page'}))
# Get an instance of Mining the Social Web # Using the page name also works if you know it. # e.g. 'MiningTheSocialWeb' or 'CrossFit' mtsw_id = '146803958708175' pp(g.get_object(mtsw_id))
# MTSW catalog link pp(g.get_object('')) # PCI catalog link pp(g.get_object(''))
# Find Pepsi and Coke in search results pp(g.request('search', {'q' : 'pepsi', 'type' : 'page', 'limit' : 5})) pp(g.request('search', {'q' : 'coke', 'type' : 'page', 'limit' : 5})) # Use the ids to query for likes pepsi_id = '56381779049' # Could also use 'PepsiUS' coke_id = '40796308305' # Could also use 'CocaCola' # A quick way to format integers with commas every 3 digits def int_format(n): return "{:,}".format(n) print "Pepsi likes:", int_format(g.get_object(pepsi_id)['likes']) print "Coke likes:", int_format(g.get_object(coke_id)['likes'])
pp(g.get_connections(pepsi_id, 'feed')) pp(g.get_connections(pepsi_id, 'links')) pp(g.get_connections(coke_id, 'feed')) pp(g.get_connections(coke_id, 'links'))
# First, let's query for all of the likes in your social # network and store them in a slightly more convenient # data structure as a dictionary keyed on each friend's # name. We'll use a dictionary comprehension to iterate # over the friends and build up the likes in an intuitive # way, although the new "field expansion" feature could # technically do the job in one fell swoop as follows: # # g.get_object('me', fields='id,name,friends.fields(id,name,likes)') # # See Appendix C for more information on Python tips such as # dictionary comprehensions friends = g.get_connections("me", "friends")['data'] likes = { friend['name'] : g.get_connections(friend['id'], "likes")['data'] for friend in friends } print likes
# Analyze all likes from friendships for frequency # pip install prettytable from prettytable import PrettyTable from collections import Counter friends_likes = Counter([like['name'] for friend in likes for like in likes[friend] if like.get('name')]) pt = PrettyTable(field_names=['Name', 'Freq']) pt.align['Name'], pt.align['Freq'] = 'l', 'r' [ pt.add_row(fl) for fl in friends_likes.most_common(10) ] print 'Top 10 likes amongst friends' print pt
# Analyze all like categories by frequency friends_likes_categories = Counter([like['category'] for friend in likes for like in likes[friend]]) pt = PrettyTable(field_names=['Category', 'Freq']) pt.align['Category'], pt.align['Freq'] = 'l', 'r' [ pt.add_row(flc) for flc in friends_likes_categories.most_common(10) ] print "Top 10 like categories for friends" print pt
# Build a frequency distribution of number of likes by # friend with a dictionary comprehension and sort it in # descending order from operator import itemgetter num_likes_by_friend = { friend : len(likes[friend]) for friend in likes } pt = PrettyTable(field_names=['Friend', 'Num Likes']) pt.align['Friend'], pt.align['Num Likes'] = 'l', 'r' [ pt.add_row(nlbf) for nlbf in sorted(num_likes_by_friend.items(), key=itemgetter(1), reverse=True) ] print "Number of likes per friend" print pt
# Which of your likes are in common with which friends? my_likes = [ like['name'] for like in g.get_connections("me", "likes")['data'] ] pt = PrettyTable(field_names=["Name"]) pt.align = 'l' [ pt.add_row((ml,)) for ml in my_likes ] print "My likes" print pt # Use the set intersection as represented by the ampersand # operator to find common likes. common_likes = list(set(my_likes) & set(friends_likes)) pt = PrettyTable(field_names=["Name"]) pt.align = 'l' [ pt.add_row((cl,)) for cl in common_likes ] print print "My common likes with friends" print pt
# Which of your friends like things that you like? similar_friends = [ (friend, friend_like['name']) for friend, friend_likes in likes.items() for friend_like in friend_likes if friend_like.get('name') in common_likes ] # Filter out any possible duplicates that could occur ranked_friends = Counter([ friend for (friend, like) in list(set(similar_friends)) ]) pt = PrettyTable(field_names=["Friend", "Common Likes"]) pt.align["Friend"], pt.align["Common Likes"] = 'l', 'r' [ pt.add_row(rf) for rf in sorted(ranked_friends.items(), key=itemgetter(1), reverse=True) ] print "My similar friends (ranked)" print pt # Also keep in mind that you have the full range of plotting # capabilities available to you. A quick histogram that shows # how many friends. plt.hist(ranked_friends.values()) plt.xlabel('Bins (number of friends with shared likes)') plt.ylabel('Number of shared likes in each bin') # Keep in mind that you can customize the binning # as desired. See # For example... plt.figure() # Display the previous plot plt.hist(ranked_friends.values(), bins=arange(1,max(ranked_friends.values()),1)) plt.xlabel('Bins (number of friends with shared likes)') plt.ylabel('Number of shared likes in each bin') plt.figure() # Display the working plot
import networkx as nx # pip install networkx import requests # pip install requests friends = [ (friend['id'], friend['name'],) for friend in g.get_connections('me', 'friends')['data'] ] url = '' mutual_friends = {} # This loop spawns a separate request for each iteration, so # it may take a while. Optimization with a thread pool or similar # technique would be possible. for friend_id, friend_name in friends: r = requests.get(url % (friend_id, ACCESS_TOKEN,) ) response_data = json.loads(r.content)['data'] mutual_friends[friend_name] = [ data['name'] for data in response_data ] nxg = nx.Graph() [ nxg.add_edge('me', mf) for mf in mutual_friends ] [ nxg.add_edge(f1, f2) for f1 in mutual_friends for f2 in mutual_friends[f1] ] # Explore what's possible to do with the graph by # typing nxg.<tab> or executing a new cell with # the following value in it to see some pydoc on nxg print nxg
# Finding cliques is a hard problem, so this could # take a while for large graphs. # See and #. cliques = [c for c in nx.find_cliques(nxg)]] friends_in_all_max_cliques = list(reduce(lambda x, y: x.intersection(y), max_clique_sets)) print 'Num cliques:', num_cliques print 'Avg clique size:', avg_clique_size print 'Max clique size:', max_clique_size print 'Num max cliques:', num_max_cliques print print 'Friends in all max cliques:' print json.dumps(friends_in_all_max_cliques, indent=1) print print 'Max cliques:' print json.dumps(max_cliques, indent=1)
from networkx.readwrite import json_graph nld = json_graph.node_link_data(nxg) json.dump(nld, open('resources/ch02-facebook/viz/force.json','w'))
Note: You may need to implement some filtering on the NetworkX graph before writing it out to a file for display in D3, and for more than dozens of nodes, it may not be reasonable to render a meaningful visualization without some JavaScript hacking on its parameters. View the JavaScript source in force.html for some of the details.
from IPython.display import IFrame from IPython.core.display import display # IPython Notebook can serve files and display them into # inline frames. Prepend the path with the 'files' prefix. viz_file = 'files/resources/ch02-facebook/viz/force.html' display(IFrame(viz_file, '100%', '600px')) | http://nbviewer.jupyter.org/github/ptwobrussell/Mining-the-Social-Web-2nd-Edition/blob/master/ipynb/Chapter%202%20-%20Mining%20Facebook.ipynb | CC-MAIN-2017-47 | refinedweb | 1,573 | 57.37 |
Ticket #6791 (closed Feature Requests: fixed)
Support boost::array
Description
boost::hash_value appears to lack support for boost::array. Could you add it?
Isn't it possible to support all containers in a generic way?
Attachments
Change History
comment:1 Changed 4 years ago by danieljames
- Owner changed from danieljames to marshall
- Component changed from hash to array
comment:2 Changed 4 years ago by marshall
- Status changed from new to closed
- Resolution set to fixed
comment:3 Changed 4 years ago by Olaf van der Spek <olafvdspek@…>
It isn't possible to do it generically, because equality isn't always defined the same for all containers.
Does that matter? Can't you just do "return boost::hash_range(x.begin(), x.end());"?
Support needs to be added to array itself,
What about std::array?
Note: See TracTickets for help on using tickets.
It isn't possible to do it generically, because equality isn't always defined the same for all containers.
Support needs to be added to array itself, which is pretty easy, it just needs to include <boost/functional/hash_fwd.hpp> and then add something like the following to the same namespace as array (so that it will be picked up by ADL): | https://svn.boost.org/trac/boost/ticket/6791 | CC-MAIN-2016-40 | refinedweb | 204 | 62.98 |
While I was preparing another demo, I ran into an interesting problem that hadn’t occurred to me before. I was building a sample e-commerce site and thought it would be nice to always display the category list on the left hand side of every page.
The first way I could think of was always passing that data in every action method, but that seemed to be a brute force approach that lack any sort of elegance (not to mention poor application design).
So, since I’m using a master page (which most MVC projects are), the best way would be to place that data in there. However, I don’t want to hardcode it, so I would like to fetch the list of categories from the database.
Problem is…how do I pass data to the master page. The master page is not a common view and we have no specific controller action methods for it. In fact, the master page is rendered every time we return a view from an action method. Which brings us back to the original suggestion: should I pass the data in every action method?
Well, after playing around a bit, I found that the answer is “kinda”…
I absolutely refused to add a line of code to every action method passing that information in the ViewData dictionary. Not only it’s boring to keep repeating myself, it’s also against the DRY principle (Don’t Repeat Yourself).
So what to do? I’m sure there are other solutions but this one seems very elegant and in line with the best practices: we use inheritance.
Create a new class under your controllers folder and use the following code:
public class MySiteController : Controller
{
public NorthwindEFEntities nw;
public MySiteController()
{
nw = new NorthwindEFEntities();
ViewData["categories"] = from c in nw.Categories
select c;
}
}
Notice that this new class inherits from the Controller class, just like any other controller you’d create. The trick now is to change your controllers to inherit from MySiteController instead, so that the category information is injected in every action method execution.
public class HomeController : MySiteController
public ActionResult Index()
return View();
So, we can implement our action methods as usual without worrying with the category information as that is being already passed on through the ViewData dictionary.
Of course, we need to change the master page so that it renders this information. Locate the ul element named “menu” and change it to this:
<ul id="menu">
<% foreach (var c in (IEnumerable<MvcApplicationWithInheritance.Models.Category>) ViewData["categories"])
{%>
<li><%= Html.ActionLink(c.CategoryName, "Category", "Home", new { id = c.CategoryID },null) %></li>
<%} %>
</ul>
Now, if you browse your site, the menu will no longer display the links to Home and About as usual, but will instead appear as this:
If you click on any of the buttons, you will be redirected (according to the parameters we placed on the action link) to http://<yoursite>/Home/Category/<categoryID>. That allows us to implement a simple action method to display the products in that category:
public ActionResult Category(int id)
var productList = from p in nw.Products
select p;
return View(productList);
Notice that instead of creating the NorthwindEFEntities object to make the query, I can reuse the one we created in the base class.
Until next time! | http://blogs.msdn.com/b/nunos/archive/2010/02/04/quick-tips-about-asp-net-mvc-pass-data-to-your-master-page.aspx | CC-MAIN-2015-06 | refinedweb | 550 | 50.26 |
(Transferred from the wiki by Peter)
Introduction
Welcome to C++ !
I am going to teach you some basics of a very powerfull programming language called C++. C++ is language base on the famous C langugae and was created originally by Bjarne Stroustrup in 1979 .
Data types and variables
There are seven basic data types in c++ :
- char : character datatype which represents a character like 'A' or '@'.
- int : integer datatype which represents an integer number like 63.
- float: floating point datatype which represents a floating point number like 63.63 .
- double: double precision floating numbers.
- void: valueless datatype
- bool: boolean datatype which has two possible values true or false
- wchar_t wide character which uses 16 bits for representing data.
OK these are the basic datatypes but what are variables ?
A variable is a container which we declare to hold some data for us. A variable is declared like this : datatype variable_name. Here are some examples :
- int no_of_visitors;
- float temperature;
- bool is_registered;
Notice that at the end of each line of the code above there is a semicolon(
we put it at the end of every statement to tell the compiler this is a statement.we put it at the end of every statement to tell the compiler this is a statement.
We can declare several variables of the same type in one statement like this :
type first_variable , second_variable , third_variable ; .
And we can initialize the variables (give them value) when declaring them like this :
type variable_name = value;
Here is an example :
int no_of_users = 63 , no_of_guests = 100;
In C/C++, the names of variables, functions, and various other user-defined objects are called identifiers. These identifiers can vary from one to several characters. The first character must be a letter or an underscore, and subsequent characters must be either letters, digits, or underscores. Here are some correct and incorrect identifier names:
Correct Incorrect first_user 1st_user user_list user..list
Input/Output
In C++ Input/Output(I/O) is done by means of streams. This means you create a stream and link it to a device and use it. All streams behave in the same way even though they are linked to different devices. There are some predefined streams to use : cin and cout . cin which is used for inputing data is linked to the keyboard by default(it can be redirected) and it is used for inputing data . cin stands for console input.
cout on the other hand stands for concole output and is used for printing data on the monitor, since it is linked to the monitor by default.
To use these streams, we should include the file <iostream> first.We do it in this way :
#include <iostream>
This line should be placed on top of your source file. Now we can use the predefined streams like this :
std::cin >> no_of_users ;
To avoid typing std:: before each instance of cin and cout we can add a line before our main funcion like this :
using namespace std;
From now on we can use our known cin and cout freely without typing std:: .
My sites: Linux Home Networking - Linux Quick Fix Notebook
Bookmarks | http://www.linuxhomenetworking.com/forums/showthread.php/18753-C-plus-plus-tutorial | CC-MAIN-2014-42 | refinedweb | 518 | 70.73 |
From: Sven Stork (stork_at_[hidden])
Date: 2007-09-06 04:56:54
On Thursday 06 September 2007 02:29, Jeff Squyres wrote:
> Unfortunately, <iostream> is there for a specific reason. The
> MPI::SEEK_* names are problematic because they clash with the
> equivalent C constants. With the tricks that we have to play to make
> those constants [at least mostly] work in the MPI C++ namespace, we
> *must* include them. The comment in mpicxx.h explains:
>
> // We need to include the header files that define SEEK_* or use them
> // in ways that require them to be #defines so that if the user
> // includes them later, the double inclusion logic in the headers will
> // prevent trouble from occuring.
> // include so that we can smash SEEK_* properly
> #include <stdio.h>
> // include because on Linux, there is one place that assumes SEEK_* is
> // a #define (it's used in an enum).
> #include <iostream>
>
> Additionally, much of the C++ MPI bindings are implemented as inline
> functions, meaning that, yes, it does add lots of extra code to be
> compiled. Sadly, that's the price we pay for optimization (the fact
> that they're inlined allows the cost to be zero -- we used to have a
> paper on the LAM/MPI web site showing specific performance numbers to
> back up this claim, but I can't find it anymore :-\ [the OMPI C++
> bindings were derived from the LAM/MPI C++ bindings]).
>
> You have two options for speeding up C++ builds:
>
> 1. Disable OMPI's MPI C++ bindings altogether with the --disable-mpi-
> cxx configure flag. This means that <mpi.h> won't include any of
> those extra C++ header files at all.
>
> 2. If you're not using the MPI-2 C++ bindings for the IO
> functionality, you can disable the SEEK_* macros (and therefore
> <stdio.h> and <iostream>) with the --disable-mpi-cxx-seek configure
> flag.
maybe this could be a third option:
3. just add -DOMPI_SKIP_MPICXX to you compilation flags to skip the inclusion
of the mpicxx.h.
-- Sven
> See "./configure --help" for a full list of configure flags that are
> available.
>
>
>
>
> On Sep 4, 2007, at 4:22 PM, Thompson, Aidan P. wrote:
>
> > This is more a comment that a question. I think the compile-time
> > required
> > for large applications that use Open MPI is unnecessarily long. The
> > situation could be greatly improved by streamlining the number of C+
> > + header
> > files that are included. Currently, compiling LAMMPS
> > (lammps.sandia.gov)
> > takes 61 seconds to compile with a dummy MPI library and 262
> > seconds with
> > Open MPI, a 4x slowdown.
> >
> > I noticed that iostream.h is included by mpicxx.h, for no good
> > reason. To
> > measure the cost of this, I compiled the follow source file 1)
> > without any
> > include files 2) with mpi.h 3) with iostream.h and 4) with both:
> >
> > $ more foo.cpp
> > #ifdef FOO_MPI
> > #include "mpi.h"
> > #endif
> >
> > #ifdef FOO_IO
> > #include <iostream>
> > #endif
> >
> > void foo() {};
> >
> > $ time mpic++ -c foo.cpp
> > 0.04 real 0.02 user 0.02 sys
> > $ time mpic++ -DFOO_MPI -c foo.cpp
> > 0.58 real 0.47 user 0.07 sys
> > $ time mpic++ -DFOO_IO -c foo.cpp
> > 0.30 real 0.23 user 0.05 sys
> > $ time mpic++ -DFOO_IO -DFOO_MPI -c foo.cpp
> > 0.56 real 0.47 user 0.07 sys
> >
> > Including mpi.h adds about 0.5 seconds to the compile time and
> > iostream
> > accounts for about half of that. With optimization, the effect is even
> > greater. When you have hundreds of source files, that really adds up.
> >
> > How about cleaning up your include system?
> >
> > Aidan
> >
> >
> >
> >
> >
> > --
> > Aidan P. Thompson
> > 01435 Multiscale Dynamic Materials Modeling
> > Sandia National Laboratories
> > PO Box 5800, MS 1322 Phone: 505-844-9702
> > Albuquerque, NM 87185 FAX : 505-845-7442
> > mailto:athomps_at_[hidden]
> >
> >
> >
> > _______________________________________________
> > users mailing list
> > users_at_[hidden]
> >
>
>
> --
> Jeff Squyres
> Cisco Systems
>
> _______________________________________________
> users mailing list
> users_at_[hidden]
>
> | http://www.open-mpi.org/community/lists/users/2007/09/3979.php | CC-MAIN-2014-52 | refinedweb | 637 | 76.93 |
I want to show you a few more key points about inferno that are useful in building a grid. I'll show a really simple example of feeding data to all the worker processes.
Setup the grid as described in my last post.
This example is a multicore grep(1), so I want to send filenames to multiple grep workers. The are many different ways of doing this. The point here is to show more features of inferno that highlight its special qualities.
One such quality is that rcmd exports the local namespace to the remote command; we can use this to supply information to control the remote processes. The namespace is mounted on /n/client. To see this try,
% rcmd -A ${nextcpu} ls /n/client
Remember there's a lot more there than just a disk filesystem. We can create named pipes and feed data from one local to many remote processes through this pipe(3).
Make a named pipe
% mkdir /tmp/rpipe
% bind '#|' /tmp/rpipe
A simple example using the pipe (start the reading process first),
% rcmd -A ${nextcpu} cat /n/client/tmp/rpipe/data1 &
% du -a > /tmp/rpipe/data
I defined rsplit in my last post, but it only worked for .dis commands so I'm going to tweak it to work with sh braced blocks (another important inferno quality).
fn rsplit {
args := $*
rpid := ()
for i in `{ndb/regquery -n svc rstyx} {
rcmd -A $i sh -c ${quote $"args }&
rpid = $apid $rpid
}
for i in /prog/ ^ $rpid ^/wait {read < $i } >/dev/null >[2=1]
}
I'm going to use the command fs(1) to walk a file tree and print the paths. For this example I'm going to have it print all limbo source files.
fn limbofiles {
fs print {select {mode -d} {filter -d {match -ar '.*\.(b|m)$'} {walk /appl}} }
}
Now I'm going to tie the pipe, fs, and rsplit together to get a distributed grep.
fn lk {
re = $1
bind '#|' /tmp/rpipe
rsplit {
re=`{cat /n/client/env/re};
getlines {grep -i $re $line /dev/null} < /n/client/tmp/rpipe/data1
}&
sid := $apid
limbofiles > /tmp/rpipe/data
read < /prog/ ^ $sid ^ /wait >/dev/null >[2=1]
}
A few things to note about that. I set the environment variable 're' to the first argument of lk. The env(3) filesystem is also exported as part of our rcmd namespace, so I can read the value of 're' from any worker,
re=`{cat /n/client/env/re}
I use the sh-std(1) builtin 'getlines' to read one line at a time from the pipe and run grep (the /dev/null is to force grep to print filenames). The last line is there to wait for the rsplit command to finish.
You can run lk as follows,
% lk wmexport | http://code.google.com/p/inferno-lab/wiki/SimpleGridPartTwo | crawl-003 | refinedweb | 462 | 76.66 |
Modern front-end frameworks require you to download a development environment, complete with dependencies, and compile your code before even trying to view it on your browser. Is this a good thing? Is the problem that we are building more complex sites, or is it that the frameworks are complex in their own right, introducing an unnecessary level of complexity?
Web development today has evolved a lot since the ’90s; we are able to create entire experiences that are very near what any native application can do, and the development process has also changed. Gone are the days when being a front-end web developer was a matter of opening Notepad, typing a few lines of code, checking it on the browser, and uploading it to an FTP folder.
The Front-end Web Development of the Past
I must start by stating the obvious: The world isn’t like it was 10 years ago. (Shocking, I know.) The only thing that remains constant is change. Back in the day, we had very few browsers, but there were a lot of compatibility issues. Today, you don’t see things like “best viewed on Chrome 43.4.1” much, but back then, it was pretty common. We have more browsers now, but fewer compatibility issues. Why? Because of jQuery. jQuery satisfied the need to have a standard, common library that allowed you to write JavaScript code that manipulates the DOM without needing to worry about how it was going to run on each browser and on each version of each browser—a true nightmare in the 2000s.
Modern browsers can manipulate the DOM as a standard, so the need for such a library has diminished greatly in recent years. jQuery isn’t needed anymore, but we can still find a number of extremely useful plugins that depend on it. In other words, web frameworks may not be necessary, but they are still useful enough to be popular and widely used. This is a trait common to most of the popular web frameworks out there, from React, Angular, Vue, and Ember to style and formatting models like Bootstrap.
Why People Use Frameworks
In web development as in life, having a quick solution is always handy. Have you ever done a router before in JavaScript? Why go through the painful process of learning when you can npm-install a front-end framework to overcome the issue? Time is a luxury when the client wants things done yesterday or you inherit code from another developer designed for a particular framework, or if you are integrating with a team already using a given framework. Let’s face it—frameworks do exist for a reason. If there were no benefits to them, nobody would be using them.
So what are some of the benefits and unique properties of using a web development framework?
Time is money. When you are developing a project and the client doesn’t care which framework you use—indeed, probably isn’t even aware of what you use—they only care about getting results, and the faster the better. Established frameworks let you create an instant sense of progress from the beginning, which the client craves from day 1. Additionally, the faster you develop, the more money you make, since the time freed up by the framework can be redirected to taking on more projects.
It’s all about the community. When choosing a framework, this is a very important point—who is going to help you when you get stuck on an issue? You and I both know that it’s going to happen at some point. You’ll reach a spot where you need to do something that the framework wasn’t intended to do, or that the framework was never designed to give you access to, so having a community supporting you is essential. Development—especially freelance—can be hard, as you are immersed in a virtual world, and if you’re the sole front-end web developer in a team, it means you’re the only one with the experience and expertise to find a solution. But if the front-end framework you use has solid support, there is going to be someone on the other side of the world who has faced the same problem and might be able to help you.
Standards are beautiful. Have you ever noticed that, when you look into an old piece of your own code, you can navigate through it pretty easily? Or at least, more easily than a piece of code written by someone else? You think in a certain way, and you have your own way of naming things and organizing the code. That’s a standard. We all follow them, even if they’re only for ourselves. We tend to eat similar things for breakfast, wake up at a certain hour, and place our keys in the same place every day. And indeed, if we changed up our routines every day, life would be a lot harder just from the overhead of figuring out how to do stuff. Ever lost your keys because you put them in a different place than normal? Standards make life easier. When working as part of a team or a community of developers, they become absolutely indispensible.
Frameworks provide a standard from the moment you install them, guiding you to think and to code in a specific way. You don’t need to spend time creating a standard with your team; you can just follow how things are done in the framework. This makes it easier to work tgether. It’s easier to look for a function when you know that the function must be in a certain file because it is built for adding a route in an SPA, and in your framework, all routes are placed in a file with that name. People with different skill levels can work together if you have this level of standardization, because while the advanced coders know why things are done that way, even junior developers can follow the standard itself.
When Frameworks Fail
A few years ago, saying something like “I don’t use frameworks—I don’t see any real benefit from them” would bring people with torches and pitchforks to your door. But today, more and more people are asking themselves, “Why should I use a framework at all? Do I really need them? Is it that hard to code without them?”
I’m certainly one of them—I’ve never been a fan of any specific framework, and I’ve been coding without them for my entire career. If I have a choice in the matter, my choice is always, “No, thanks.” I’ve been developing in JavaScript for years and in ActionScript before that. I was coding in Flash when most people already considered it dead. (I know, I know… but I was doing lots of animations, and animation in plain HTML is hard.) So if you’re one of the many who never think about coding without frameworks, let me show you some reasons why you might be struggling.
“One size fits all” is a lie. Could you imagine writing a single piece of software that can do everything you’ve accomplished in your career? That’s one of the main problems with web development frameworks. Your project has very specific needs, which we tend to solve by adding libraries, plugins, or add-ons to extend the framework’s scope. No framework offers 100% of what you need, and no framework is 100% composed of things that you’re going to find useful.
Having too much code that you don’t use can result in load time lag for your site, which becomes more important with each additional user. Another issue is that the “one size fits all” mindset results in inefficient code. Take, for example,
$(‘sku-product').html('SKU 909090');, which is jQuery code that, in the end, we all know is going to be translated into something like
document.getElementById('sku-product').innerHTML = 'SKU 909090';.
That kind of difference on a single line might seem unimportant, but changing the content of a specific element of the page is precisely the virtue of React. Now, React goes through the process of creating a representation of the DOM and analyzing the differences in what you try to render. Wouldn’t it be easier to just target the content that you want to change from the beginning?
That tangle of weeds you’re walking through is growing thorns. Have you ever been in the situation where you’re using your framework and trying to add a library to it, just to realize that the library version you need doesn’t work well with the framework version that you’re using? Sometimes it takes more effort to make two pieces of code work together than it does to just write the code yourself. And since the frameworks and libraries that you use are often built on other frameworks and libraries that can have hidden incompatibilities that you can’t even anticipate, the problem can grow exponentially more complex, reaching a point where they’re impossible to manage if you want the project to keep growing.
Keeping up with the Joneses is a thing. Ever worked on a project in AngularJS only to find out that you need something that didn’t appear until Angular 4 was released? Did you even know that Angular 5 has been released? This is another huge issue; even if you’re sticking to a single front-end framework, when a new major release happens, things can change so much that the code you worked so hard to make won’t even run on the new version. This could result in anything from annoying little changes that need to be made on a lot of files to a complete rewrite of your code.
Keeping up with the latest builds of a framework is challenging, but on the same note, other frameworks suffer when updates stop completely and they can’t keep up with the rest of technology. In 2010, both AngularJS and Backbone were released for the first time. Today, Angular is on its fifth major version, and Backbone is completely out of the spotlight. Seven years seems like a long time. If you build websites, they’ve probably changed completely in aesthetic and function. If you’re building an app, betting on the wrong framework might put the company in a tough—and expensive—situation later, when things need to be rewritten.
When all you’ve got is a hammer, everything looks like a nail. If you’ve used web development frameworks frequently, this has probably happened to you, where a single codebase defines the shape of the code you use in the future, even if it’s only peripherally related. Let’s say you’re going to build a platform like YouTube and you want to use Framework X. There might be a point where, even if it sounds ridiculous in this day and age, you decide to use Flash for the videos because that’s what comes built in with the framework.
Frameworks have opinions, and they are strong; React, for example, forces you to use JSX in a specific way. You can see code being used in that way everywhere. Is there an alternative? Yes. But who uses it? This isn’t always a bad thing, but if you need to perform complex animations, you might only need a framework for animating and not the entirety of React. I’ve seen people do crazy things like add jQuery to a page just to append a node to an element, something that could be accomplished in vanilla JS with
document.getElementById('id_of_node').appendChild(node);.
Eval Is Evil, but
.innerHTML Is Machiavellian
I want to take the time to explore this point separately because I think this is one of the reasons more people don’t code without frameworks. When you see how most code works when trying to add something to the DOM, you’ll find a bunch of HTML injected by the
.innerHTML property. We all seem to agree that
eval is bad for running JavaScript code, but I want to put
.innerHTML in the spotlight here. When you inject HTML code as a plain string, you lose any reference you might have had to any of the nodes you created. It’s true that you might get them back by using
getElementsByClassName or assigning them an
id, but this is less than practical. When trying to change the value of one of the nodes, you’ll find yourself rendering the entire HTML back again.
This is good when you start coding. You can make lots of simple things easily without much experience. The problem happens with the complexity of modern websites, which tend to be more like apps—this means that we need to constantly change the values of our nodes, which is a high-cost operation if you’re doing it by reattaching the entire structure via
.innerHTML. React solves this problem efficiently via a shadow DOM, and Angular addresses it by using binding as an easy way to modify a value shown on a page. However, it can also be solved fairly easily by keeping track of the nodes you create and saving the ones that will be reused or updated in variables.
There are also other reasons to stay away from
.innerHTML in general.
The Biggest Myths about Coding Without Frameworks
Time is money. Yep, I’m bringing this concept back from earlier. Many people feel like if they stop using a popular web framework, we will instantly devolve to the internet of the 90s, when
<marquee> was everyone’s favorite tag, rotating GIFs on a Geocities site were hip and edgy, Alta Vista was the go-to for web searches, and hit counters were ubiquitous.
With web frameworks, your first lines of code seem to make a lot of time-saving progress, but at some point, the gains turn to losses. You spend your time reading about how to make the framework do things it isn’t built for, how to integrate libraries and make them play nice with the framework, and finding out that the code you built while following the framework’s standards isn’t going to work at all and now you need to rewrite it. When you do things without a framework, you start slower, but you make steady progress. In the end, it’s all about where you want the easy part to be. It won’t make much difference in total time.
My code will be longer than the Great Wall. Writing without a framework is like buying a movie instead of subscribing to a streaming service. You don’t get instant access to hundreds of movies that you want to watch, but you also don’t have to spend money on thousands of other movies you’d never even consider downloading from the store. You can just write what you need.
Is the middleman useful? Sure. But it’s not usually necessary. Every line of code you write has more meaning, as there’s no need for you to adapt to the requirements of a framework. It can feel like you are writing more code with pure JavaScript because the way to create the DOM elements takes lines to create an element, attach it to the DOM, and perhaps add a class for styling, as opposed to calling up a single line of code in JSX. But if you compare code using a library like jQuery or React, vanilla JS can be pretty similar in length. Sometimes it’s longer, but sometimes it’s shorter, too.
There’s no need to reinvent the wheel. The mantra of computer science professors everywhere. And it’s true—it just doesn’t have to mean frameworks specifically. Sending an Ajax request to load or save data is a requirement in almost every web app, for example, but not having a framework doesn’t mean that you need to write the code anew every time. You can create your own library or codebase, or you can extract code from others. The smaller it is, the easier it is to modify or adjust as needed, so it comes in handy when you need something specific for a project. It’s easier to modify 100-200 lines of code than navigate through the mountain of files that a third-party library or framework might contain.
It will only work for small project. This is a very common myth, but not true at all; currently, I am working on an entire system to manage all aspects of a company online in a single place, including a module that is something like Google Drive. Whether with frameworks or without them, I go through very similar steps and encounter very similar problems. The difference is negligible. However, without frameworks, my entire code is smaller and more easily manageable.
I WANT PROOF
Okay. Let’s stop talking about theory and jump to a real-world example. Some days ago, I needed to show a list of brands with logos for a store. The initial code used jQuery, but it had a problem where, when loading in Mozilla, it was showing a broken image icon for brands that didn’t have logos uploaded yet. We can’t have the store looking unfinished just because Company X hasn’t finished their end of the work yet, but the feature needed to go live.
The following code uses the jQuery equivalent of
.innerHTML:
var list_brand_names = ['amazon', 'apple', 'nokia']; var<img src='images/" + brandName + "' /></a>"; } jQuery("#brand-images").html(img_out);
Without going too deep into the pros and cons of jQuery, the problem here is that we don’t have any reference to the images we created. While there are solutions that don’t involve changing code, let’s use this opportunity to see how it can be done without any library at all:
var brands = ['amazon', 'apple', 'nokia']; var brand_images = document.getElementById("brand-images"); for (var iBrand = 0; iBrand < brands.length; iBrand++) { var link = document.createElement('a'); link.setAttribute('href', '/pages/' + brands[iBrand]); link.style.display = 'none'; brand_images.appendChild(link); var image = new Image(); image.src = "images/" + brands[iBrand] + "png"; image.onload = function(){ this.parentNode.style.display = ''; } link.appendChild(image); }
The original jQuery code was six lines long, while the vanilla JS solution took twelve. To solve the problem, hiding each image until it’s been loaded, takes twice as long to code. So let’s look at the alternative. Can it be solved in jQuery too? Check it out:
img_out += "<a href='/pages/" + brandName + "' style="display:none"><img src='images/" + brandName + "' onload="showImage(this)"/></a>"; function showImage(image){ image.parentNode.style.display = ""; }
With a couple additional lines of code, there is now only a three-line difference between the jQuery and the vanilla, but on the jQuery, you can see that the
img_out line is quickly growing very complex, to the point where you need to pause and think carefully about what you’re doing. Coding the DOM directly by using node functions to add attributes, functions, and the like might be more lengthy, but each line has a clearer, more precise meaning, making it easier to read and maintain in the future.
Let’s take a look at React:
function BrandLink(props) { var<img src={url}/></a>); } class Brands extends React.Component { constructor() { super(); this.state = {brands: ['amazon', 'apple', 'nokia']}; } render() { const links = this.state.brands.map((step, move) => { return (<BrandLink brand={step} key={step}/>); }); return (<div className="brands">{links}</div>); } } ReactDOM.render(<Brands />, document.getElementById("root"));
This version is clearly sub-optimal. The code is no shorter than it is in vanilla, and we still haven’t even gotten to the point of solving the problem and hiding the links until the image inside them has loaded.
For every example, the results are going to be different. Sometimes, jQuery will be shorter. Sometimes, React will win. There are times when vanilla JS can be shorter than both of them. In any case, the goal here wasn’t to prove that one was inherently superior to the other, but to demonstrate that there isn’t a significant difference between using vanilla JS and using a framework when it comes to code length.
Conclusion
As with just about any real-life issue, nothing is black or white. Coding without web development frameworks might be the best solution for some of your projects and a nightmare in others. Just as with every tool, the key is in learning not just how to use it, but when, and what the advantages and disadvantages to using it might be. Coding in pure JavaScript is just like with any framework—mastering it takes time before you feel comfortable using it.
But the key difference, at least for me, is that frameworks come and go, and even if a framework is popular for a long time, it can change dramatically from one version to another. Pure JavaScript will be an option for much longer, until it stops being relevant entirely and some other language emerges. Even then, there will be more concepts and strategies that you can migrate directly from one language to another than you can with a given framework to another. Time and effort being roughly equivalent where a single project is concerned, the reduction in knowledge depreciation and the lessons you can take with you to the next challenge are very important factors to consider. | https://www.toptal.com/javascript/are-big-front-end-frameworks-bad?utm_campaign=blog_post_are_big_front_end_frameworks_bad&utm_medium=email&utm_source=blog_subscribers&utm_campaign=Toptal%20Engineering%20Blog&utm_source=hs_email&utm_medium=email&utm_content=59459408&_hsenc=p2ANqtz-_4OvqG3y_B8x6Do3IinwWOn6CJdn71poq2sAtYlpJNwdaIPn1lJKmeCwsI0i76ZTdOMHVAbfeeEVnAa-YjTC_IszSEOciJqF-QO9tHmtpRDdgT0Bw&_hsmi=59459408 | CC-MAIN-2021-04 | refinedweb | 3,609 | 69.72 |
im trying to make this game. Its common and its my first project.
Game:
import random
import time
def displayIntro():
print "You are in a land full of dragons."
print "You are exploring and come upon two caves."
print "In one cave, the dragon is friendly."
print "In the other, the dragon will eat you alive"
def chooseCave():
cave = ""
while cave != "1" and cave 1= "2":
cave = raw_input("Which cave will you choose? (1 or 2)")
return cave
def checkCave(chosenCave):
print "You approach the cave slowly..."
time.sleep(3)
print "It is murky and damp."
time.sleep(3)
print "From the shadows darts a dragon, mouth open wide..."
time.sleep(3)
friendlyCave = random.randint(1,2)
if chosenCave == str(friendlyCave):
print "The dragon calms down and gives you treasure. You are flabbergasted and return home."
else:
print "It eats you in one bite"
playAgain = "yes"
while playAgain == "yes" or playAgain == "y":
displayIntro()
caveNumber = chooseCave()
checkCave(caveNumber)
playAgain = raw_input("Do you want to play again? (yes or no)?
it says "SyntaxError: multiple statements found while compiling a single statement."
What do I do? | http://forums.devshed.com/python-programming/936980-help-please-last-post.html | CC-MAIN-2016-30 | refinedweb | 183 | 78.35 |
T he Electronic Magazine of O riginal Peer-Reviewed Survey Articles ABSTRACT
- Oscar Bryan
- 2 years ago
- Views:
Transcription
1 SECOND QUARTER 2005, VOLUME 7, NO. 2 IEEE C OMMUNICATIONS SURVEYS T he Electronic Magazine of O riginal -Reviewed Survey Articles A SURVEY AND COMPARISON OF PEER-TO-PEER OVERLAY NETWORK SCHEMES ENG KEONG LUA, JON CROWCROFT, AND MARCELO PIAS, UNIVERSITY OF CAMBRIDGE RAVI SHARMA, NANYANG TECHNOLOGICAL UNIVERSITY STEVEN LIM, MICROSOFT ASIA ABSTRACT Over the Internet today, computing and communications environments are significantly more complex and chaotic than classical distributed systems, lacking any centralized organization or hierarchical control. There has been much interest in emerging -to- (P2P) network overlays because they provide a good substrate for creating large-scale data sharing, content distribution, and application-level multicast applications. These P2P overlay networks attempt to provide a long list of features, such as: selection of nearby peers, redundant storage, efficient search/location of data items, data permanence or guarantees, hierarchical naming, trust and authentication, and anonymity.. We categorize the various schemes into these two groups in the design spectrum, and discuss the application-level network performance of each group. -to-peer (P2P) overlay networks are distributed systems in nature, without any hierarchical organization or centralized control. s form self-organizing overlay networks that are overlayed on the Internet Protocol (IP) networks, offering a mix of various features such as robust wide-area routing architecture, efficient search of data items, selection of nearby peers, redundant storage, permanence, hierarchical naming, trust and authentication, anonymity, massive scalability, and fault tolerance. . We can view P2P overlay network models spanning a wide spectrum of the communication framework, which specifies a fully-distributed, cooperative network design with peers building a self-organizing system. Figure 1 shows an abstract P2P overlay architecture, illustrating the components in the overlay The Network Communications layer describes the network characteristics of desktop machines connected over the Internet or small wireless or sensor-based devices that are connected in an ad-hoc manner. The dynamic nature of peers poses challenges in the communication paradigm. The Overlay Nodes Management layer covers the management of peers, which include discovery of peers and routing algorithms for optimization. The Features Management layer deals with the security, reliability, fault resiliency, and aggregated resource availability aspects of maintaining the robustness of P2P systems. The Services Specific layer supports the underlying P2P infrastructure and the application-specific components through scheduling of parallel and computation-intensive tasks, content and file management. Meta-data describes the content stored across the P2P peers and the location information. The Application-level layer is concerned with tools, applications, and services that are implemented with specific functionalities on top of the underlying P2P overlay infrastructure. Thus, there are two classes of P2P overlay networks: Structured and Unstructured. The technical meaning of structured is that the P2P overlay network topology is tightly controlled and content is placed not at random peers but at specified locations that will make X IEEE Communications Surveys & Tutorials Second Quarter 2005
2 Applications Services management Meta-data Services messaging Services scheduling Security management Reliability and fault resiliency Routing and location lookup Resources discovery Network Resource management n Figure 1. An abstract P2P overlay network architecture. subsequent queries more efficient. Such structured P2P systems use the Distributed Hash Table (DHT) as a substrate, in which data object (or value) location information is placed deterministically, at the peers with identifiers corresponding to the data object s unique key. DHT-based systems have a property that consistently assigned uniform random NodeIDs to the set of peers into a large space of identifiers. Data objects are assigned unique identifiers called keys, chosen from the same identifier space. Keys are mapped by the overlay network protocol to a unique live peer in the overlay network. The P2P overlay networks support the scalable storage and retrieval of {key,value} pairs on the overlay network, as illustrated in Fig. 2. Given a key, a store operation (put(key,value)) lookup retrieval operation (value=get(key)) can be invoked to store and retrieve the data object corresponding to the key, which involves routing requests to the peer corresponding to the key. Each peer maintains a small routing table consisting of its neighboring peers NodeIDs and IP addresses. Lookup queries or message routing are forwarded across overlay paths to peers in a progressive manner, with the NodeIDs that are closer to the key in the identifier space. Different DHT-based systems will have different organization schemes for the data objects and its key space and routing strategies. In theory, DHT-based systems can guarantee that any data object can be located in small O(logN) overlay hops on average, where N is the number of peers in the system. The underlying network path between two peers can be significantly different from the path on the DHT-based overlay network. Therefore, the lookup latency in DHT-based P2P overlay networks can be quite high and could adversely affect the performance of the applications running over it. Plaxton et al. [1] provide an elegant algorithm that achieves nearly optimal latency on graphs that exhibit power-law expansion [2], at the same time preserving the scalable routing properties of the DHT-based system. However, this algorithm requires pair-wise probing between peers to determine latencies, and it is unlikely to scale to a large number of peers in the overlay. DHT-based systems [3 7] are an important class of P2P routing infrastructures. They support the rapid development of a wide variety of Internetscale applications ranging from distributed file and naming systems to application-layer multicast. They also enable scalable, wide-area retrieval of shared information. Tools Services Application-level layer Services-specific layer Features managment layer Overlay nodes management layer Network communications layer API interface: Put(Key, Value) In 1999 Napster [8] pioneered the idea of a peer-to-peer file sharing system supporting a centralized file search facility. It was the first system to recognize that requests for popular content need not be sent to a central server but instead could be handled by many peers that have the requested content. Such P2P file-sharing systems are self-scaling in that as more peers join the system, they add to the aggregate download capability. Napster achieved this self-scaling behavior by using a centralized search facility based on file lists provided by each peer; thus, it does not require much bandwidth for the centralized search. Such a system has the issue of a single point of failure due to the centralized search mechanism. However, a lawsuit filed by the Recording Industry Association of America (RIAA) attempted to force Napster to shut down the free-for-all file-sharing service for copyrighted digital music literally, its killer application. However, this paradigm caught the imagination of platform providers and users alike. Gnutella [9 11] is a decentralized system that distributes both the search and download capabilities, establishing an overlay network of peers. It is the first system that makes use of an unstructured P2P overlay network. An unstructured P2P system is composed of peers joining the network with some loose rules, without any prior knowledge of the topology. The network uses flooding as the mechanism to send queries across the overlay with a limited scope. When a peer receives the flood query, it sends a list of all content matching the query to the originating peer. While floodingbased techniques are effective for locating highly replicated items and are resilient to peers joining and leaving the system, they are poorly suited for locating rare items. Clearly this approach is not scalable as the load on each peer grows linearly with the total number of queries and the system size. Thus, unstructured P2P networks face one basic problem: peers readily become overloaded, and thus the system does not scale when handling a high rate of aggregate queries and sudden increases in system size. Although structured P2P networks can efficiently locate rare items since the key-based routing is scalable, they incur significantly higher overheads than unstructured P2P networks for popular content. Consequently, over the Internet today the decentralized unstructured P2P overlay networks are more Distributed structured P2P overlay application API interface: Remove(Key) API interface: Value=Get(Key) Distributed hash table Value n Figure 2. Application interface for structured DHT-based P2P overlay systems. IEEE Communications Surveys & Tutorials Second Quarter
3 X s coordinate neighbor set = {A B C D} B B A X C D E A X Z C Sample routing path from X to E lays schemes in both classes and discuss its developments. Then we attempt to use the following taxonomies to make comparisons between the various discussed structured and unstructured P2P overlay schemes: Decentralization examine whether the overlay system is distributed. Architecture describe the overlay system architecture with respect to its operation. Lookup protocol the lookup query protocol adopted by the overlay system. System parameters the required system parameters for the overlay system operation. Routing performance the lookup routing protocol performance in overlay routing. Routing state the routing state and scalability of the overlay system. s join and leave describe the behavior of the overlay system when churn and self-organization occurred. Security look into the security vulnerabilities of overlay systems. Reliability and fault resiliency examine how robust the overlay system is when subjected to faults. Finally, we conclude with some thoughts on the relative applicability of each class to some of the research problems that arise in ad-hoc, location-based, and content delivery networks, as examples. STRUCTURED P2P OVERLAY NETWORKS X s coordinate neighbor set = {A B D Z} New Z s coordinate neighbor set = {A C D X} n Figure 3. Example of 2-d space CAN before and after Z joins. commonly used. However, there are recent efforts on keybased routing (KBR) API abstractions [12] that allow more application-specific functionalities to be built over these common basic KBR API abstractions, and OpenDHT 1 (Open publicly accessible DHT service) [13] that allow the unification platform to provide developers with a basic DHT service model that runs on a set of infrastructure hosts, to deploy DHT-based overlay applications without the burden of maintaining a DHT, and with ease of use to spur the deployment of DHT-based applications. In contrast, unstructured P2P overlay systems are ad-hoc in nature, and do not present the possibilities of being unified under a common platform for application development. In later sections of this article we will describe the key features of structured P2P and unstructured P2P overlay networks and their operation functionalities. After providing a basic understanding of the various network overlay schemes in these two classes, we proceed to evaluate these various over- 1 The OpenDHT project was formerly named OpenHash. D E In this category, the overlay network assigns keys to data items and organizes its peers into a graph that maps each data key to a peer. This structured graph enables efficient discovery of data items using the given keys. However, in its simple form, this class of systems does not support complex queries and it is necessary to store a copy or a pointer to each data object (or value) at the peer responsible for the data object s key. In this section, we survey and compare the following structured P2P overlay networks: Content Addressable Network (CAN) [5], Tapestry [7], Chord [6], Pastry [4], Kademlia [14], and Viceroy [15]. CONTENT ADDRESSABLE NETWORK (CAN) The Content Addressable Network (CAN) [5] is a distributed decentralized P2P infrastructure that provides hash-table functionality on an Internet-like scale. CAN is designed to be scalable, fault-tolerant, and self-organizing. The architectural design is neighbors in the coordinate space. A CAN message includes the destination coordinates. Using the neighbor coordinates, a peer routes a message toward its destination using a simple greedy forwarding to the neighbor peer that is closest to the destination coordinates. CAN has a routing performance of O(d N 1/d ) and its routing state is of 2 d bound. As shown in Fig. 3, which we adapted from the CAN paper [5], the virtual coordinate space is used to store {key,value} pairs as follows: to store a pair {K,V}, key K is deterministically mapped onto a point P in the coordinate space using a uniform hash function. The lookup protocol 74 IEEE Communications Surveys & Tutorials Second Quarter 2005
4 retrieves an entry corresponding to key K, and any peer can apply the same deterministic hash function to map K onto point P and then retrieve the corresponding value V from the point P. If the requesting peer or its immediate neighbors do not own the point P, the request must be routed through the CAN infrastructure until it reaches the peer where P lays. A peer maintains the IP addresses of those peers that hold coordinate zones adjoining its zone. This set of immediate neighbors in the coordinate space serves as a coordinate routing table that enables efficient routing between points in this space. A new peer that joins the system must have its own portion of the coordinate space allocated. This can be achieved by splitting an existing peer s zone in half, retaining half for the peer and allocating the other half to the new peer. CAN has an associated DNS domain name that is resolved into the IP address of one or more CAN bootstrap peers (which maintain a partial list of CAN peers). For a new peer to join a CAN network, the peer looks up in the DNS of a CAN domain name to retrieve a bootstrap peer s IP address, similar to the bootstrap mechanism in [16]. The bootstrap peer supplies the IP addresses of some randomly chosen peers in the system. The new peer randomly chooses a point P and sends a JOIN request destined for point P. Each CAN peer uses the CAN routing mechanism to forward the message until it reaches the peer in which zone P lies. The current peer in zone P then splits its zone in half and assigns the other half to the new peer. For example, in a 2-dimensional space, a zone would first be split along the X dimension, then the Y, and the splitting continues. The {K,V} pairs from the half zone to be handed over are also transferred to the new peer. After obtaining its zone, the new peer learns of the IP addresses of its neighbor set from the previous peer in point P, and adds to that the previous peer itself. When a peer leaves the CAN network, an immediate takeover algorithm ensures that one of the failed peer s neighbors takes over the zone and starts a takeover timer. The peer updates its neighbor set to eliminate those peers that are no longer its neighbors. Every peer in the system then sends softstate updates to ensure that all of their neighbors will learn about the change and update their own neighbor sets. The number of neighbors a peer maintains depends only on the dimensionality of the coordinate space (i.e., 2 d) and it is independent on the total number of peers in the system. The Fig. 3 example illustrated a simple routing path from peer X to point E and a new peer Z joining the CAN network. For a d-dimensional space partitioned into n equal zones, the average routing path length is (d/4) (n 1/d ) hops and individual peers maintain a list of 2 d neighbors. Thus, the growth of peers (or zones) can be achieved without increasing per peer state while the average path length grows as O(n 1/d ). Since there are many different paths between two points in the space, when one or more of a peer s neighbors fail, this peer can still route along the next best available path. Improvement of the CAN algorithm can be accomplished by maintaining multiple, independent coordinate spaces, with each peer in the system being assigned a different zone in each coordinate space, called reality. For a CAN with r realities, a single peer is assigned r coordinate zones, one on each reality available, and this peer holds r independent neighbor sets. The contents of the hash table are replicated on every reality, thus improving data availability. For further data availability improvement, CAN could use k different hash functions to map a given key onto k points in the coordinate space. This results in the replication of a single {key,value} pair at k distinct peers in the system. A {key,value} pair is then unavailable only when all the k replicas are simultaneously unavailable. Thus, queries for a particular hash table entry could be forwarded to all k peers in parallel, thereby reducing the average query latency, and reliability and fault resiliency properties are enhanced. CAN could be used in large-scale storage management systems such as the OceanStore [17], Farsite [18], and Publius [19]. These systems require efficient insert and retrieval of content in a large distributed storage network with a scalable indexing mechanism. Another potential application for CANs is in the construction of wide-area name resolution services that decouple the naming scheme from the name resolution process. This enables an arbitrary and location-independent naming scheme. CHORD Chord [6] uses consistent hashing [20] to assign keys to its peers. Consistent hashing is designed to let peers enter and leave the network with minimal interruption. This decentralized scheme tends to balance the load on the system, since each peer receives roughly the same number of keys, and there is little movement of keys when peers join and leave the system. In a steady state, for a total of N peers in the system, each peer maintains routing state information for about O(logN) other peers. This may be efficient but performance degrades gracefully when that information is out-of-date. The consistent hash functions assign peers and data keys an m-bit identifier using SHA-1 [21] as the base hash function. A peer s identifier is chosen by hashing the peer s IP address, while a key identifier is produced by hashing the data key. The length of the identifier m must be large enough to make the probability of keys hashing to the same identifier negligible. Identifiers are ordered on an identifier circle modulo 2m. Key k is assigned to the first peer whose identifier is equal to or follows k in the identifier space. This peer is called the successor peer of key k, denoted by successor(k). If identifiers are represented as a circle of numbers from 0 to 2m 1, then successor(k) is the first peer clockwise from k. The identifier circle is called the Chord ring. To maintain consistent hashing mapping when a peer n joins the network, certain keys previously assigned to n s successor now need to be reassigned to n. When peer n leaves the Chord system, all of its assigned keys are reassigned to n s successor. Therefore, peers join and leave the system with (logn) 2 performance. No other changes of keys assignment to peers need to occur. In Fig. 4 (adapted from [6]), the Chord ring is depicted with m = 6. This particular ring has 10 peers and stores five keys. The successor of the identifier 10 is peer 14, so key 10 will be located at NodeID 14. Similarly, if a peer were to join with identifier 26, it would store the key with identifier 24 from the peer with identifier 32. Each peer in the Chord ring needs to know how to contact its current successor peer on the identifier circle. Lookup queries involve the matching of key and NodeID. For a given identifier could be passed around the circle via these successor pointers until they encounter a pair of peers that include the desired identifier; the second peer in the pair is the peer the query maps to. An example is presented in Fig. 4, whereby peer 8 performs a lookup for key invokes the find successor operation for this key, which eventually returns the successor of that key, i.e. peer 56. The query visits every peer on the circle between peer 8 and peer 56. The response is returned along the reverse of the path. As m is the number of bits in the key/nodeid space, each peer n maintains a routing table with up to m entries, called the finger table. The ith entry in the table at peer n contains the identity of the first peer s that succeeds n by at least 2 i 1 IEEE Communications Surveys & Tutorials Second Quarter
5 on the identifier circle, i.e., s = successor(n + 2 i 1 ), where 1 i m. s is the ith finger of peer n (n finger[i]). A finger table entry includes both the Chord identifier and the IP address (and port number) of the relevant peer. Figure 4 shows the finger table of peer 8, and the first finger entry for this peer points to peer 14, as the latter is the first peer that succeeds (8+20) mod 26 = 9. Similarly, the last finger of peer 8 points to peer 42, i.e., the first peer that succeeds (8 + 25) mod 26 = 40. In this way, peers store information about only a small number of other peers, and know more about peers closely following it on the identifier circle than other peers. Also, a peer s finger table does not contain enough information to directly determine the successor of an arbitrary key k. For example, peer 8 cannot determine the successor of key 34 by itself, as the successor of this key (peer 38) is not present in peer 8 s finger table. When a peer joins the system, the successor pointers of some peers need to be changed. It is important that the successor pointers are up to date at any time because the correctness of lookups is not guaranteed otherwise. The Chord protocol uses a stabilization protocol [6] running periodically in the background to update the successor pointers and the entries in the finger table. The correctness of the Chord protocol relies on the fact that each peer is aware of its successors. When peers fail, it is possible that a peer does not know K54 N56 N51 N48 K38 N56 N51 N48 N42 N42 N38 N38 N N32 N4 N N14 N21 Finger table N8 + 1 N14 N8 + 2 N14 N8 + 4 N14 N8 + 8 N21 N N32 N N42 n Figure 4. Chord ring with identifier circle consisting of ten peers and five data keys. It shows the path followed by a query originated at 8 for the lookup of key 54. Finger table entries for 8. N8 K30 N8 +1 N21 K24 N14 Lookup (K54) K10 its new successor, and that it has no chance to learn about it. To avoid this situation, peers maintain a successor list of size r, which contains the peer s first r successors. When the successor peer does not respond, the peer simply contacts the next peer on its successor list. Assuming that peer failures occur with a probability p, the probability that every peer on the successor list will fail is p r. Increasing r makes the system more robust. By tuning this parameter, any degree of robustness with good reliability and fault resiliency may be achieved. The following applications are examples of how Chord could be used. Cooperative mirroring or cooperative file system (CFS) [22], in which multiple providers of content cooperate to store and serve each others data. Spreading the total load evenly over all participant hosts lowers the total cost of the system, since each participant needs to provide capacity only for the average load, not for the peak load. There are two layers in CFS. The DHash (Distributed Hash) layer performs block fetches for the peer, distributes the blocks among the servers, and maintains cached and replicated copies. The Chord layer distributed lookup system is used to locate the servers responsible for a block. Chord-based DNS [23] provides a lookup service, with host names as keys and IP addresses (and other host information) as values. Chord could provide a DNS-like service by hashing each host name to a key [20]. Chordbased DNS would require no special servers, while ordinary DNS systems rely on a set of special root servers. DNS also requires manual management of the routing information (DNS records) that allows clients to navigate the name server hierarchy. Chord automatically maintains the correctness of the analogous routing information. DNS only works well when host names are hierarchically structured to reflect administrative boundaries. Chord imposes no naming structure. DNS is specialized to the task of finding named hosts or services, while Chord can also be used to find data object values that are not tied to particular machines. TAPESTRY Sharing similar properties with Pastry, Tapestry [7] employs decentralized randomness to achieve both load distribution and routing locality. The difference between Pastry and Tapestry is the handling of network locality and data object replication, and this difference will be more apparent when described in the section on Pasty. Tapestry s architecture uses a variant of the Plaxton et al. [1] distributed search technique, with additional mechanisms to provide availability, scalability, and adaptation in the presence of failures and attacks. Plaxton et al. propose a distributed data structure, known as the Plaxton mesh, optimized to support a network overlay for locating named data objects that are connected to one root peer. On the other hand, Tapestry uses multiple roots for each data object to avoid a single point of failure. In the Plaxton mesh, peers can take on the roles of servers (where data objects are stored), routers (forward messages), and clients (entity of requests). It uses local routing maps at each peer to incrementally route overlay messages to the destination ID digit by 76 IEEE Communications Surveys & Tutorials Second Quarter 2005
6 digit, e.g., * * * 7 * * 97 * , where * is the wildcard, similar to the longest prefix routing in the CIDR IP address allocation architecture [24]. The resolution of digits from right to left or left to right is arbitrary. A peer s local routing map has multiple levels, where each of them represents a match of the suffix with a digit position in the ID space. The nth peer that a message reaches shares a suffix of at least length n with the destination ID. To locate the next router, the (n + 1)th level map is examined to locate the entry matching the value of the next digit in the destination ID. This routing method guarantees that any existing unique peer in the system can be located within at most log B N logical hops, in a system with N peers using NodeIDs of base B. Since the peer s local routing map assumes that the preceding digits all match the current peer s suffix, the peer needs only to keep a small constant size (B) entry at each route level, yielding a routing map of fixed constant size: (entries/map) no. of maps = B log B N. The lookup and routing mechanisms of Tapestry are based on matching the suffix in NodeID as described above. Routing maps are organized into routing levels, where each level contains entries that point to a set of peers closest in distance that matches the suffix for that level. Also, each peer holds a list of pointers to peers referred to as neighbors. Tapestry stores the locations of all data object replicas to increase semantic flexibility and allow the application level to choose from a set of data object replicas based on some selection criteria, such as date. Each data object may include an optional application-specific metric in addition to a distance metric. For example, the OceanStore [17] global storage architecture finds the closest cached document replica that satisfies the closest distance metric. These queries deviate from the simple find first semantics, and Tapestry will route the message to the closest k distinct data objects. Tapestry handles the problem of a single point of failure due to a single data object s root peer by assigning multiple roots to each object. Tapestry makes use of surrogate routing to select root peers incrementally during the publishing process to insert location information into Tapestry. Surrogate routing provides a technique by which any identifier can be uniquely mapped to an existing peer in the network. A data object s root or surrogate peer is chosen as the peer that matches the data object s ID, X. This is unlikely to happen, given the sparse nature of the NodeID space. Nevertheless, Tapestry assumes peer X exists by attempting to route a message to it. A route to a non-existent identifier will encounter empty neighbor entries at various positions along the way. The goal is to select an existing link that can act as an alternative to the desired link; i.e. the one associated with a digit X. Routing terminates when a map is reached where the only non-empty routing entry belongs to the current peer. That peer is then designated as the surrogate root for the data object. While surrogate routing may take additional hops to reach a root if compared with the Plaxton algorithm, the additional number of hops is small. Thus, surrogate routing in Tapestry has minimal routing overhead relative to the static global Plaxton algorithm. Tapestry addresses the issue of fault adaptation and maintains cached content for fault recovery by relying on TCP timeouts and UDP periodic heartbeat packets to detect link, server failures during normal operations, and rerouting through its neighbors. During fault operation each entry in the neighbor map maintains two backup neighbors in addition to the closest/primary neighbor. On a testbed of 100 machines with 1000 peer simulations, the results in [7] show the good routing rates and maintenance bandwidths during instantaneous failures and continuing churn. A variety of different applications have been designed and implemented on Tapestry. Tapestry is self-organizing, fault tolerant, resilient under load, and is a fundamental component of the OceanStore system [17, 25]. OceanStore is a global-scale, highly available storage utility deployed on the PlanetLab [26] testbed. OceanStore servers use Tapestry to disseminate encoded file blocks efficiently, and clients can quickly locate and retrieve nearby file blocks by their ID, despite server and network failures. Other Tapestry applications include the Bayeux [27], an efficient self organizing application-level multicast system, and SpamWatch [28], a decentralized spam-filtering system that uses a similarity search engine implemented on Tapestry. PASTRY Pastry [4], like Tapestry, makes use of Plaxton-like prefix routing to build a self-organizing decentralized overlay network, where each peer routes client requests and interacts with local instances of one or more applications. Each peer in Pastry is assigned a 128-bit peer identifier (NodeID). The NodeID is used to give a peer s position in a circular NodeID space, which ranges from 0 to The NodeID is assigned randomly when a peer joins the system, and it is assumed to be generated such that the resulting set of NodeIDs is uniformly distributed in the 128-bit space. For a network of N peers, Pastry routes to the numerically closest peer to a given key in less than log B N steps under normal operation (where B = 2 b is a configuration parameter with typical value of b = 4). The NodeIDs and keys are considered a sequence of digits with base B. Pastry routes messages to the peer whose NodeID is numerically closest to the given key. A peer normally forwards the message to a peer whose NodeIDs share with the key a prefix that is at least one digit (or b bits) longer than the prefix that the key shares with the current peer NodeID. As shown in Fig. 5, each Pastry peer maintains a routing table, a neighborhood set, and a leaf set. A peer routing table is designed with log B N rows, where each row holds B 1 number of entries. The B 1 number of entries at row n of the routing table each refer to a peer whose NodeID shares the current peer s NodeID in the first n digits, but whose (n+1)th digit has one of the B 1 possible values other than the (n+1)th digit in the current peer s NodeID. Each entry in the routing table contains the IP address of peers whose NodeIDs have the appropriate prefix, and it is chosen according to a close proximity metric. The value of b could be chosen with a tradeoff between the size of the populated portion of the routing table [approximately (log B N) (B 1) entries] and maximum number of hops required to route between any pair of peers (log B N). The neighborhood set, M, contains the NodeIDs and IP addresses of the M peers that are closest in proximity to the local peer. The network proximity that Pastry uses is based on a scalar proximity metric such as the IP routing geographic distance. The leaf set, L, is the set of peers with L /2 numerically closest larger NodeIDs and L /2 peers with numerically smaller NodeIDs, in relation to the current peer s NodeID. Typical values for L and M are B or 2 B. Even with concurrent peer failure, eventual delivery is guaranteed with good reliability and fault resiliency, unless L /2 peers with adjacent NodeIDs fail simultaneously ( L is a configuration parameter with a typical value of 16 or 32). When a new peer (NodeID is X) joins the network, it needs to initialize its routing table and inform other peers of its presence. This new peer needs to know the address of a contact or bootstrap peer in the network. A small list of contact peers, based on a proximity metric (e.g., the RTT to each peer) to provide better performance, could be provided as a IEEE Communications Surveys & Tutorials Second Quarter
7 Routing table of a Pastry peer with NodeID 37A0x, b=4, digits are in hexadecimal, x is an arbitrary suffix 0x 1x 2x 3x 4x... Dx Ex Fx 30x 31x 32x... 37x 38x... 3Ex 3Fx 370x 371x 372x... 37Ax 37Bx... 37Ex 37Fx 37A0x 37A1x 37A2x... 37ABx 37ACx 37ADx 37AEx 37AFx Example: routing state of a Pastry peer with NodeID 37A0F1, b=4, L=16, M=32 NodeID 37A0F1 Leaf set (smaller) 37A001 37A011 37A022 37A033 37A0F1 37A044 37A055 37A066 37A077 Leaf set (larger) 37A0F2 37A0F4 37A0F6 37A0F8 Live peers in Pastry Route (B57B2D) 37A0FA 37A0FB 37A0FC 37A0FE B581F1 Neighborhood set B57B2D B573D6 B24EA3 1A223B 1B AD0 2670AB 3612AB 37890A 390AF0 3912CD 46710A AB 490CDE 279DE0 290A0B 510A0C 5213EA B573AB B5324F Routing from peer 37A0F1 with key B57B2D 11345B A 19902D C 199ABC the side of the failed peer, and request its leaf table. Let the received leaf set be L, which overlaps the current peer s leaf set L, and it contains peers with nearby NodeIDs not residing in L. The appropriate peer is chosen to insert into L, verifying that the peer is actually still alive by contacting it. The neighborhood set is not used in the routing of messages, but it is still kept fresh because this set plays an important role in exchanging information about nearby peers. Therefore, a peer contacts each member of the neighborhood set periodically to test if it is still alive. If the peer is not responding, the contacting peer asks other members for their neighborhood sets and checks for the closest proximity of each of the newly contacted peers and updates its own neighborhood set. Pastry is being used in the implementation of a scalable application-level multicast infrastructure called Scribe [29, 30]. Instead of relying on a multicast infrastructure in the network that is not widely available, the participating peers route and distribute multicast messages using only unicast network services. It supports a large number of groups with large numn Figure 5. Pastry peer's routing table, leaf set, and neighbor set. An example of routing path for a pastry peer. service in the network, and the new peer could select at random one of the peers for contact. As a result, this new peer will know initially about a closest Pastry peer A. X then asks A to route a join message with the key equal to X. Pastry routes the join message to the existing peer Z whose NodeID is numerically closest to X. Upon receiving the join request, peers A, Z and all peers encountered on the path from A to Z send their routing tables to X. Finally, X informs any peers that need to be aware of its arrival. This ensures that X initializes its routing table with appropriate information and the routing tables in all other affected peers are updated based on the information received. As peer A is topologically close to the new peer X, A s neighborhood set is used to initialize X s neighborhood set. A Pastry peer is considered to have left the overlay network when its immediate neighbors in the NodeID space can no longer communicate with the peer. To replace this failed peer in the leaf set of its neighbors, its neighbors in the NodeID space contact the live peer with the largest index on 78 IEEE Communications Surveys & Tutorials Second Quarter 2005
8 bers of members per group. Scribe is built on top of Pastry, which is used to create and manage groups and to build efficient multicast trees for dissemination of messages to each group. Scribe builds a multicast tree formed by joining Pastry routes from each group member to a rendezvous point associated with a group. Membership maintenance and message dissemination in Scribe leverages the robustness, self organization, locality, and reliability properties of Pastry. SplitStream [31] allows a cooperative multicasting environment where peers contribute resources in exchange for using the service. The key idea is to split the content into k stripes and to multicast each stripe using a separate tree. s join as many trees as there are stripes they wish to receive and they specify an upper bound on the number of stripes that they are willing to forward. The challenge is to construct this forest of multicast trees such that an interior peer in one tree is a leaf peer in all the remaining trees and the bandwidth constraints specified by the peers are satisfied. This ensures that the forwarding load can be spread across all participating peers. For example, if all peers wish to receive k stripes and they are willing to forward k stripes, SplitStream will construct a forest such that the forwarding load is evenly balanced across all peers while achieving low delay and link stress across the network. Squirrel [32] uses Pastry as its data object location service, to identify and route to peers that cache copies of a requested data object. It facilitates mutual sharing of Web data objects among client peers, and enables the peers to export their local caches to other peers in the network, thus creating a large shared virtual Web cache. Each peer then performs both Web browsing and Web caching, without the need for expensive and dedicated hardware for centralized Web caching. Squirrel faces a new challenge whereby peers in a decentralized cache incur the overhead of having to serve each other requests, and this extra load must be kept low. PAST [33, 34] is a large scale P2P persistent storage utility that is based on Pastry. The PAST system is composed of peers connected to the Internet such that each peer is capable of initiating and routing client requests to insert or retrieve files. s may also contribute storage to the system. A storage system like PAST is attractive because it exploits the multitude and diversity of peers in the Internet to achieve strong persistence and high availability. This eradicates the need for physical transport of storage media to protect lookup and archival data, and the need for explicit mirroring to ensure high availability and throughput for shared data. A global storage utility also facilitates the sharing of storage and bandwidth, thus permitting a group of peers to jointly store or publish content that would exceed the capacity or bandwidth of any individual peer. Pastiche [35] is a simple and inexpensive backup system that exploits excess disk capacity to perform P2P backup with no administrative costs. The cost and inconvenience of backup are unavoidable and often prohibitive. Small-scale solutions require significant administrative efforts. Large-scale solutions require aggregation of substantial demand to justify the capital costs of a large, centralized repository. Pastiche builds on three architectures: Pastry, which provides the scalable P2P network with self-administered routing and peer location; content-based indexing [36, 37], which provides flexible discovery of redundant data for similar files; and convergent encryption [18], which allows hosts to use the same encrypted representation for common data without sharing keys. KADEMLIA The Kademlia [14] P2P decentralized overlay network takes the basic approach of assigning each peer a NodeID in the 160-bit key space, and {key,value} pairs are stored on peers with IDs close to the key. A NodeID-based routing algorithm will be used to locate peers near a destination key. One of the key architectures of Kademlia is the use of a novel XOR metric for distance between points in the key space. XOR is symmetric and it allows peers to receive lookup queries from precisely the same distribution of peers contained in their routing tables. Kademlia can send a query to any peer within an interval, allowing it to select routes based on latency or send parallel asynchronous queries. It uses a single routing algorithm throughout the process to locate peers near a particular ID. Every message being transmitted by a peer includes its peer ID, permitting the recipient to record the sender peer s existence. Data keys are also 160-bit identifiers. To locate {key,value} pairs, Kademlia relies on the notion of distance between two identifiers. Given two 160-bit identifiers, a and b, it defines the distance between them as their bitwise exclusive OR (XOR, interpreted as d(a, b) = a b = d(b, a) for all a, b), and this is a non-euclidean metric. Thus, d(a, b) = 0, d(a, b) > 0(if a b), and for all a, b: d(a, b) = d(b, a). XOR also offers the triangle inequality property: d(a, b) + d(b, c) d(a, c), since d(a, c) = d(a, b) d(b, c) and (a + b a b) for all a, b = 0. Similar to Chord s clockwise circle metric, XOR is unidirectional. For any given point x and distance d > 0, there is exactly one point y such that d(x, y) = d. The unidirectional approach makes sure that all lookups for the same key converge along the same path, regardless of the originating peer. Hence, caching {key,value} pairs along the lookup path alleviates hot spots. The peer in the network stores a list of {IP address,udp port,nodeid} triples for peers of distance between 2 i and 2 i+1 from itself. These lists are called k-buckets. Each k-bucket is kept sorted by last time seen, i.e. least recently accessed peer at the head, most-recently accessed at the tail. The Kademlia routing protocol consists of the following steps: PING probes a peer to check if it is active. STORE instructs a peer to store a {key,value} pair for later retrieval. FIND_NODE takes a 160-bit ID, and returns {IP address,udp port,nodeid} triples for the k peers it knows that are closest to the target ID. FIND_VALUE is similar to FIND_NODE: it returns {IP address,udp port,nodeid} triples, except in the case when a peer receives a STORE for the key, in which case it just returns the stored value. Importantly, Kademlia s peer must locate the k closest peers to some given NodeID. This lookup initiator starts by picking X peers from its closest non-empty k-bucket, and then sends parallel asynchronous FIND_NODE to the X peers it has chosen. If FIND_NODE fails to return a peer that is any closer than the closest peers already seen, the initiator resends the FIND_NODE to all of the k closest peers it has not already queried. It can route for lower latency because it has the flexibility to choose any one of k peers to forward a request. To find a {key,value} pair, a peer starts by performing a FIND_VALUE lookup to find the k peers with IDs closest to the key. To join the network, a peer n must have contact with an already participating peer m. n inserts peer m into the appropriate k-bucket, and then performs a peer lookup for its own peer ID. n refreshes all k-buckets farther away than its closest neighbor, and during this refresh, peer n populates its own k-buckets and inserts itself into other peers k-buckets, if needed. VICEROY The Viceroy [15] P2P decentralized overlay network is designed to handle the discovery and location of data and resources in a dynamic butterfly fashion. Viceroy employs IEEE Communications Surveys & Tutorials Second Quarter
9 B 0 1 A vantages: it requires global knowledge to construct the overlay; an object s root peer is the single point of failure; no insertion or deletion Level 1 of peers; and no avoidance of hotspot congestion. Pastry and Tapestry schemes relied on DHT to provide the substrate for semantic-free and data-centric references, through the assignment of a semantic-free NodeID, such as a 160- bit key, and performed efficient request routing between lookup peers using an efficient and Level 2 dynamic routing infrastructure, whereby peers leave and join. Overlays that perform query routing in DHT-based systems have strong theoretical foundations, guaranteeing that a key can be found if it exists, and they do not capture the relationships between the object name and its Level 3 content. However, DHT-based systems have a few problems in terms of data object lookup latency: For each overlay hop, peers route a message to the next intermediate peer that can be located very far away with regard to physical topology of the underlying IP network. This can result in high network delay and unnecessary long-distance network traffic, from a deterministic short overlay path of O(logN), where N is the number of peers. DHT-based systems assume that all peers equally participate in hosting published data objects or their location information. This would lead to a bottleneck at low-capacity peers. Pastry and Tapestry routing algorithms are a randomized approximation of a hypercube, and routing toward an object is done by matching longer address suffixes until either the object s root peer or another peer with a nearby copy is found. Rhea et al. [41] make use of FreePastry implementation to discover that most lookups fail to complete when there is excessive churn. They claimed that short-lived peers leave the overlay with lookups that have not yet timed out. They outlined design issues pertaining to DHT-based performance under churn: lookup timeouts; reactive versus periodic recovery of peers; and the choice of nearby neighbors. Since the reactive recovery will increase traffic to congested links, they make use of periodic recovery; and for lookup they suggested an exponential weighted moving average of each neighbor s response time instead of alternative fixed timeout. They discovered that selection of nearby neighbors required global sampling, which is more effective than simply sampling neighbor s neighbors. However, Castro et al. [42] use the MSPastry implementation to show that it can cope with high churn rates by achieving shorter routing paths and a lesser maintenance overhead. Pastry exploits network locality to reduce routing delays by measuring the delay round-trip-time (RTT) to a small number of peers when building the routing tables. For each routing table entry, it chooses one of the closest peers in the network topology whose NodeID satisfies the constraints for that entry. The average IP delay of each Pastry hop increases exponentially until it reaches the average delay between two peers in the network. Chord s routing protocol is similar to Pastry s location algorithm in PAST. However, Pastry is a prefix-based routing protocol and differs in other details from Chord. Chord maps keys and peers to an identifier ring and guarantees that queries make a logarithmic number of hops and that keys are well balanced. It uses consistent hashing to minimize disruption of keys when peers leave and join the overlay network. Consistent hashing ensures that the total number of caches responsible for a particular object is limited, and when these caches change, the minimum number of object refern Figure 6. A simplified Viceroy network. For simplicity, the up link, ring, and level-ring links are not shown. consistent hashing [20] to distribute data so that it is balanced across the set of servers and resilient to servers joining and leaving the network. It utilizes the DHT to manage the distribution of data among a changing set of servers and allowing peers to contact any server in the network to locate any stored resource by name. In addition to this, Viceroy maintains an architecture that is an approximation to a butterfly network [38], as shown in Fig. 6 (adapted from the diagram in [15]), and uses links between successors and predecessors ideas that were based on Kleingberg [39] and Barriere et al. [40] on the ring (a key is mapped to its successor on the ring) for short distances. Its diameter of the overlay is better than CAN and its degree is better than Chord, Tapestry, and Pastry. When N peers are operational, one of the logn levels is selected with near equal probability. Level l peer s two edges are connected to peers at level l + 1. A down-right edge is added to a long-range contact at level l + 1 at a distance about 1/(2 l ) away, and a down-left edge is added at a close distance on the ring to the level l +1. The up edge to a nearby peer at level l 1 is included if l > 1. Then, level-ring links are added to the next and previous peers of the same level l. Routing is done by climbing using up connections to a level l 1 peer, then proceeds down the levels of the tree using the down links, and moving from level l to level l + 1. It follows either the edge to the nearby down link or the further down link, depending on distance > 1/(2 l ). This recursively continues until a peer is reached with no down links, and it is in the vicinity of the target peer. So a vicinity lookup is performed using the ring and level-ring links. For reliability and fault resiliency, when a peer leaves the overlay network, it hands over its key pairs to a successor from the ring pointers and notifies other peers to find a replacement. It is formalized and proved [15] that the routing process requires only O(logN), where N is the number of peers in the network. DISCUSSION OF STRUCTURED P2P OVERLAY NETWORK The algorithm of Plaxton was originally devised to route Web queries to nearby caches, and this influenced the design of Pastry, Tapestry, and Chord. The method of Plaxton has logarithmic expected join/leave complexity. Plaxton ensures that queries never travel further in network distance than the peer where the key is stored. However, Plaxton has several disad- 80 IEEE Communications Surveys & Tutorials Second Quarter 2005
10 ences will move to maintain load balancing. Since the Chord lookup service presents a solution where each peer maintains a logarithmic number of long-range links, it gives a logarithmic join/leave update. In Chord, the network is maintained appropriately by a background maintenance process, i.e. a periodic stabilization procedure that updates predecessor and successor pointers to cater to newly joined peers. Liben-Nowell et al. [43] ask the question of how often the stabilization procedure needs to run to determine the success of Chord s lookups and if determining the optimum involves the measurement of peers behavior. Stoica et al. [6] demonstrate the advantage of recursive lookups over iterative lookups, but future work is proposed to improve resiliency to network partitions using a small set of known peers, and to reduce the amount of messages in lookups by increasing the size of each step around the ring with a larger finger in each peer. Alima et al. [44] propose a correction-on-use mechanism in their Distributed K-ary Search (DKS), which is similar to Chord, to reduce the communication costs incurred by Chord s stabilization procedure. The mechanism makes corrections to the expired routing entries by piggybacking lookups and insertions. The work on CAN has a constant degree network for routing lookup requests. It organizes the overlay peers into a d- dimensional Cartesian coordinate space, with each peer taking ownership of a specific hyper-rectangular shape in the space. The key motivation of the CAN design is based on the argument that Plaxton-based schemes would not perform well under churn, given that peer departures and arrivals would affect a logarithmic number of peers. It maintains a routing table with its adjacent immediate neighbors. s joining the CAN cause the peer owning the region of space to split, giving half to the new peer and retaining half. s leaving the CAN will pass its NodeID, neighbors NodeID, IP addresses and its {key,value} pairs to a takeover peer. CAN has a number of tunable parameters to improve routing performance: dimensionality of the hypercube; network-aware routing by choosing the neighbor closest to the destination in CAN space; multiple peers in a zone, allowing CAN to deliver messages to anyone of the peers in the zone in an anycast manner; uniform partitioning, made possible by comparing the volume of a region with the volumes of neighboring regions when a peer joins; and landmark-based placement which causes peers, at join time, to probe a set of well known landmark hosts, estimating each of their network distances. There are open research questions on CAN s resiliency, load balancing, locality, and latency/hopcount costs. Kademlia s XOR topology-based routing resembles very much the first phase in the routing algorithms of Pastry, Tapestry, and Plaxton. For these three algorithms, there is a need for an additional algorithmic structure for discovering the target peer within the peers that share the same prefix but differ in the next b-bit digit. It was argued in [14] that Pastry and Tapestry algorithms require secondary routing tables of size O(2 b ) in addition to the main tables of size O(2 b log 2 bn), which increases the cost of bootstrapping and maintenance. Kademlia resolves in their distinctive ways through the use of XOR metrics for the distance between 160-bit NodeIDs, and each peer maintains a list of contact peers, of which longerlived peers are given preference on this list. Kademlia can easily be optimized with a base other than 2, by configuring the bucket table so that it approaches the target b bits per hop. This requiress having one bucket for each range of peers at a distance [j (2 160 (i+1)b ), (j + 1) (2 160 (i+1)b )], for each 0 < j < 2 b and 0 i < 160/b. This expects no more than (2 b 1) (log 2 bn) buckets. The Viceroy overlay network (butterfly) presents an efficient network construction proved formally in [15], and maintains constant degree networks in a dynamic environment, similar to CAN. Viceroy has logarithmic diameter, similar to Chord, Pastry, and Tapestry. Viceroy s diameter is proven to be better than CAN and its degree is better than Chord, Pastry, and Tapestry. Its routing is achieved in O(logN) hops (where N is the number of peers) and with nearly optimal congestion. s joining and leaving the system induce O(logN) hops and require only O(1) peers to change their states. Li et al. [45] suggest in their paper that limited degree may increase the risk of network partition or limitations in the use of local neighbors. However, its advantage is the constantdegree overlay properties. Kaashoek et al. [46] highlight its fault-tolerant blind spots and its complexity. Further work was done by Viceroy s authors with the proposal of a two-tier, locality-aware DHT [47] which gives lower degree properties in each lower-tier peer, and the boundeddegree P2P overlay using de Bruijn graph [48]. Since de Bruijn graphs give very short average routing distances and high resilience to peer failure, they are well suited for Structured P2P overlay networks. The P2P overlays discussed above are greedy, and for a given degree, the algorithms are suboptimal because the routing distance is longer. There are increasing improvements to de Bruijn P2P overlay proposals [46, 49 52]. The de Bruijn graph of degree k (k can be varied) could achieve an asymptotically optimum diameter (maximum hopcounts between any two peers in the graph) of log k N, where N is the total number of peers in the system. Given O(logN) neighbors in each peer, the de Bruijn graphs hop count is O(logN/loglogN). A good comparison study has been done by Loguinov et al. [50] where they use examples of Chord, CAN, and de Bruijn to study routing performance and resilience of P2P overlay networks, including graph expansion and clustering properties. They confirmed that de Bruijn graphs for a given degree k offer the best diameter and average distance between all pairs of peers (this determines the expected response time in number of hops), optimal resilience (k-peer connectivity), large bisection width (the bisection width of a graph provides tight upper bounds on the achievable capacity of the graph), and good node (peer) expansion, which guarantees little overlap between parallel paths to any destination peer. (If there is a peer failure, very few alternative paths to a destination peer are affected.) P2P DHT-based overlay systems are susceptible to security breaches from malicious peers attacks. One simple attack on a DHT-based overlay system is when the malicious peer returns wrong data objects to the lookup queries. The authenticity of the data objects can be handled by using cryptographic techniques through some cost-effective public keys and/or content hashes to securely link together different pieces of data objects. Such techniques can neither prevent undesirable data objects from polluting the search results, nor prevent denial of attacks. Malicious peers may still be able to corrupt, deny access, or respond to lookup queries of replicas of a data object, and impersonate so that replicas may be stored on illegitimate peers. Sit et al. [53] provide a very clear description of security considerations that involve the adversaries that are peers in the DHT overlay lookup system that do not follow the protocol correctly: malicious peers are able to eavesdrop the communication between other nodes; malicious peers can only receive data objects addressed to its IP address, and thus, the IP address can be a weak form of peer identity; and malicious peers can collude together, giving believable false information. They presented a taxonomy of possible attacks involving routing deficiencies due to corrupted lookup routing and updates; vulnerability to partitioning and virtualization into incorrect networks when new peers join and contact mali- IEEE Communications Surveys & Tutorials Second Quarter
11 cious peers; lookup and storage attacks; inconsistent behavior of peers; denial of service attacks preventing access by overloading the victim s network connection; and unsolicited responses to a lookup query. Defenses design principles that can be classified as defining verifiable system invariants for lookup queries, NodeID assignment, peer selection in routing, cross checking using random queries, and avoiding single points of responsibility. Castro et al. [54] relate the problems of secure routing for Structured P2P overlay networks in terms of the possibilities that a small number of peers could compromise the overlay system if peers are malicious and conspire with each other (this is also termed an Eclipse attack [55]). They presented a design and analysis of techniques for secure peer joining, routing table maintenance, and robust message forwarding in the presence of malicious peers in Structured P2P overlays. The technique can tolerate up to 25 percent of malicious peers while providing good performance when the number of compromised peers is small. However, this defense restricts the flexibility necessary to implement optimizations such as proximity neighbor selection, and only works in Structured P2P overlay networks. So, Singh et al. [55] propose a defense that prevents Eclipse attacks for both Structured and Unstructured P2P overlay networks, by bounding the degree of overlay peers, i.e., the in-degree of overlay peers is likely to be higher than the average in-degree of legitimate peers, and legitimate peers choose their neighbors from a subset of overlay peers whose in-degree is below a threshold. Having done the in-degree bounding, it is still possible for an attacker to consume the in-degree of legitimate peers and prevent other legitimate peers from referencing to them. Therefore, bounding the out-degree is necessary so that legitimate peers choose neighbors from the subset of overlay peers whose in-degree and out-degree are below some threshold. An auditing scheme is also introduced to prevent mis-stating incorrect information of the in-degree and out-degree. Another good survey on security issues in P2P overlays is from Wallach [56], which describes secured routing primitives: assigning NodeIDs, maintaining routing tables, and forwarding of messages securely. He also suggested looking at distributed auditing, the sharing of disk space resources in P2P overlay networks as a barter economy, and the mechanism to implement such an economy. The work on BarterRoam [57] sheds light on a formal computational approach that is applicable to P2P overlay systems toward exchanging resources so that higher level functionality, such as incentive-compatible economic mechanisms, can be layered at the higher layers. Formal game theoretical approaches and models [58 60] could be constructed to analyze the equilibrium of user strategies to implement incentives for cooperation. The ability to overcome free-rider problems in P2P overlay networks will definitely improve the system s reliability and its value. In a Sybil attack, described by Douceur [61], there are a large number of potentially malicious peers in the system with no central authority to certify peers identities. Thus, it becomes very difficult to trust the claimed identity. Dingledine et al. [62] propose puzzle schemes, including the use of microcash, which allow peers to build up reputations. Although this proposal provides a degree of accountability, this still allows a resourceful attacker to launch attacks. Many P2P computational models of trust and reputation systems have emerged to assess trustworthiness behavior through feedback and interaction mechanisms. The basic assumption of these computational trust and reputation models is that the peers engage in bilateral interactions and evaluations performed on a globally agreed scale. However, most such trust and reputation systems suffer from two problems, as highlighted by Despotovic et al. [63]: extensive implementation overhead and vague trust-related model semantics. The causes lie in the aggregation of the feedback about all peers in the overlay network in order to assess the trustworthiness of a single peer, and also the anti-intuitive feedback aggregation strategies resulting in outputs that are difficult to interpret. They proposed a simple probabilistic estimation technique with maximum likelihood estimation sufficient to reduce these problems when the feedback aggregation strategy is employed. Finally, since each of these basic Structured P2P DHTbased systems defines different methods in the higher-level DHT abstractions to map keys to peers and other Structured P2P application-specific systems such as cooperative storage, content distribution, and messaging, there have been recent efforts [12] to define basic common API abstractions for the common services they provide, which they called key-based routing API (KBR), at the lower tiers of the abstractions. At the higher tiers, more abstractions can be built upon this basic KBR. In addition to DHT abstraction, which provides the same functionality as the hash table in Structured P2P DHTbased systems by mapping between keys and objects, the group anycast and multicast (CAST) (which provides scalable group communication and coordination) and decentralized object location and routing (DOLR) (which provides a decentralized directory service) are also defined. However, Karp et al. [13] point out that the above mentioned bundled library model, in which the applications read the local DHT state and receive upcalls from the DHT, requires the codes for the same set of applications to be available at all DHT hosts. This prevents the sharing of a single DHT deployment by multiple applications and generates maintenance traffic from running the DHT on its underlying infrastructure. Thus, they proposed OpenDHT with ReDiR, a distributed rendezvous service model that requires only a put()/get() interface and shares a common DHT routing platform. The authors argued that this open DHT service will spur more development of DHT-based applications without the burden of deploying and maintaining a DHT. Table 1 summarizes the characteristics of Structured P2P overlay networks that have been discussed earlier. UNSTRUCTURED P2P OVERLAY NETWORKS In this category, the overlay networks organize peers in a random graph in a flat or hierarchical manner (e.g., Super-s layer) and use flooding or random walks or expanding-ring Time-To-Live (TTL) search, etc. on the graph to query content stored by overlay peers. Each peer visited will evaluate the query locally on its own content, and will support complex queries. This is inefficient because queries for content that are not widely replicated must be sent to a large fraction of peers, and there is no coupling between topology and data items location. In this section, we will survey and compare some of the more seminal Unstructured P2P overlay networks: Freenet [64], Gnutella [9], FastTrack [65]/KaZaA [66], BitTorrent [67], and Overnet/eDonkey2000 [68, 69]. FREENET Freenet is an adaptive P2P network of peers that make queries to store and retrieve data items, which are identified by location-independent keys. This is an example of loosely Structured decentralized P2P networks with the placement of files based on anonymity. Each peer maintains a dynamic routing table, which contains addresses of other peers and the data keys that they are holding. The key features of Freenet 82 IEEE Communications Surveys & Tutorials Second Quarter 2005
12 Algorithm taxonomy Decentralization Structured P2P Overlay Network Comparisons CAN Chord Tapestry Pastry Kademlia Viceroy DHT functionality on Internet-like scale Architecture Multidimensional ID coordinate space. Uni-directional and circular NodeID space. Plaxton-style global mesh network. Plaxton-style global mesh network. XOR metric for distance between points in the key space. Butterfly network with connected ring of predecessor and successor links; data managed by servers. Lookup protocol {key, value} pairs to map a point P in the coordinate space using uniform hash function. Matching key and NodeID. Matching suffix in NodeID. Matching key and prefix in NodeID. Matching key and NodeID-based routing. Routing through levels of tree until a peer is reached with no downlinks; vicinity search performed using ring and level-ring links. System parameters N-number of peers in network and d-number of dimensions. N-number of peers in network. N-number of peers in network and B-base of the chosen peer identifier. N-number of peers in network and b-number of bits (B = 2 b ) used for the base of the chosen identifier. N-number of peers in network and b-number of bits (B = 2 b ) of NodeID. N-number of peers in network. Routing performance O(d.N 1/d ) O(logN) O(log B N) O(log B N) O(log B N)+c where c = small constant O(logN) Routing state 2d logn log B N Blog B N + Blog B N Blog B N + B logn s join/leave 2d (logn) 2 log BN log B N log B N +c where c = small constant logn Security Low level. Suffers from man-in-middle and Trojan attacks. Reliability/ fault resiliency Failure of peers will not cause network-wide failure. Multiple peers responsible for each data item. On failures, application retries. Failure of peers will not cause networkwide failure. Replicate data on multiple consecutive peers. On failures, application retries. Failure of peers will not cause network-wide failure. Replicate data across multiple peers. Keep track of multiple paths to each peer. Failure of peers will not cause network-wide failure. Replicate data across multiple peers. Keep track of multiple paths to each peer. Failure of peers will not cause network-wide failure. Replicate data across multiple peers. Failure of peers will not cause network-wide failure. Load incurred by lookups routing evenly distributed among participating lookup servers. n Table 1. A comparison of various structured P2P overlay network schemes. are the ability to maintain locally a set of files in accordance with the maximum disk space allocated by the network operator, and to provide security mechanisms against malicious peers. The basic model is that requests for keys are passed along from peer to peer through a chain of proxy requests in which each peer makes a local decision about the location to send the request next, similar to Internet Protocol (IP) routing. Freenet also enables users to share unused disk space, thus allowing a logical extension to their own local storage devices. The basic architecture consists of data items being identified by binary file keys obtained by applying the 160-bit SHA-1 hash function [70]. The simplest type of file key is the Keyword-Signed Key (KSK), which is derived from a short descriptive text string chosen by the user, e.g., /music/britney.spears. The descriptive text string is used as the input to deterministically generate a public/private key pair, and the public half is then hashed to yield the data file key. The private half of the asymmetric key pair is used to sign the data file, thus providing a minimal integrity check that a retrieved data file matches its data file key. The data file is also encrypted using the descriptive string itself as a key, so as to perform an explicit lookup protocol to access the contents of their data-stores. However, nothing prevents two users from independently choosing the same descriptive string for different files. These problems are addressed by the Signed-Subspace Key (SSK), which enables personal namespaces. The public namespace key IEEE Communications Surveys & Tutorials Second Quarter
13 with object D 9 C 10 3 Object request Reply Gnutella (pronounced newtella) is a decentralized protocol for distributed search on a flat topology of peers (servents). Gnutella is widely used, and there has been a large amount of work on improving Gnutella [71 73]. Although the Gnutella protocol supports a traditional client/centralized server search paradigm, Gnutella s distinction is its peer-to-peer, decentralized model for document location and retrieval, as shown in Fig. 8. In this model, every peer is a server or client. This system is neither a centralized directory nor does it possess any precise control over the network topology or file placement. The network is formed by peers joining the network following some loose rules. The resulting topology has certain properties, but the placement of data items is not based on any knowledge of the topology, as in the Structured P2P designs. To locate a data item, a peer queries its neighbors, and the most typical query method is flooding. The lookup query protocol is flooded to all neighbors within a certain radius. Such design is extremely resilient to peers entering and leaving the system. However, the current search mechanisms are not scalable and generate unexpected loads on the network. The so-called Gnutella servents (peers) perform tasks normally associated with both clients and servers. They provide client-side interfaces through which users can issue queries and view search results, while at the same time they also accept queries from other servents, check for matches against their local data set, and respond with applicable results. These peers are responsible for managing the background traffic that spreads the information used to maintain network integrity. Due to its distributed nature, a network of servents that implement the Gnutella protocol is highly fault-tolerant, as operation of the network will not be interrupted if a subset of servents goes offline. To join the system, a new servent (peer) initially connects to one of several known hosts that are almost always availn Figure 7. A typical request sequence in Freenet. 2 4 E B and the descriptive string are hashed independently, XOR ed together, and hashed to yield the data file key. For retrieval, the user publishes the descriptive string together with the user subspace s public key. Storing data requires the private key, so that only the owner of a subspace can add files to it, and owners have the ability to manage their own namespaces. The third type of key in FreeNet is the Content-Hash Key (CHK), which is used for updating and splitting of contents. This key is derived from hashing the contents of the corresponding file, which gives every file a pseudo-unique data file key. Data files are also encrypted by a randomly generated encryption key. For retrieval, the user publishes the content-hash key itself together with the decryption key. The decryption key is never stored with the data file but is only published with the data file key, so as to provide a measure of cover for operators. The CHK can also be used for splitting data files into multiple parts in order to optimize storage and bandwidth resources. This is done by inserting each part separately under a CHK and creating an indirect file or multiple levels of indirect files to point to the individual parts. The routing algorithm for storing and retrieving data is designed to adaptively adjust routes over time and to provide efficient performance while using local knowledge, since peers only have knowledge of their immediate neighbors. Thus, the routing performance is good for popular content. Each request is given a Hops-To-Live (HTL) limit, similar to the IP Time-To-Live (TTL), which is decremented at each peer to prevent infinite chains. Each request is also assigned a pseudo-unique random identifier, so that peers can avoid loops by rejecting requests they have seen before. If this happens, the preceding peer chooses a different peer to forward to. This process continues until the request either is satisfied or has exceeded its HTL limit. The success or failure signal (message) is returned back up the chain to the sending peer. Joining the network will rely on discovering the address of one or more of the existing peers through out-ofband means, and no peer is privileged over any other peer, so no hierarchy or centralized point of failure can exist. This intuitive resilience and decentralization enhances the performance and scalability, thus giving a constant routing state while peers join and leave the overlay Failed request A 8 6 request F In addition, as described in [64], Freenet uses its datastore to increase system performance. When an object is returned (forwarded) after a successful retrieval (insertion), the peer caches the object in its datastore, and passes the object to the upstream (downstream) requester, which then creates a new entry in its routing table associating the object source with the requested key. Thus, when a new object arrives from either a new insert or a successful request, this causes the datastore to exceed the designated size and Least Recently Used (LRU) objects are ejected in order until there is space. LRU policy is also applied to the routing table entries when the table is full. Figure 7 depicts a typical sequence of request messages. The user initiates a data request at peer A, which forwards the request to peer B, and then forwards it to peer C. C is unable to contact any other peer and returns a backtracking failed request message to peer B. B tries its second choice, peer E, which forwards the request to peer F, which then delivers it to peer B. B detects the loop and returns a backtracking failure message. F is unable to contact any other peer and backtracks one step further back to peer E. E forwards the request to its second choice, peer D, which has the data. The data is returned from peer D, via peers E, B, and A. The data is cached in peers E, B, and A, therefore it creates a routing short-cut for the next similar queries. This example shows that the overlay suffers from security problems such as man-in-middle and Trojan attacks, and the failure of peers will not cause network-wide failure, because of its lack of centralized structure. This gives good reliability and fault resiliency. GNUTELLA 84 IEEE Communications Surveys & Tutorials Second Quarter 2005
14 Query Query Query Response Download Response n Figure 8. Gnutella utilizes a decentralized architecture document location and retrieval. Query able, e.g., list of peers available from com. Once connected to the network, peers send messages to interact with each other. These messages are broadcasted (i.e. sent to all peers with which the sender has open TCP connections), or simply back-propagated (i.e., sent on a specific connection on the reverse of the path taken by an initial, broadcast message). First, each message has a randomly generated identifier. Second, each peer keeps a short memory of the recently routed messages, used to prevent re-broadcasting and to implement back-propagation. Third, messages are flagged with TTL and hops passed fields. The messages that are allowed in the network are: Group Membership (PING and PONG) Messages. A peer joining the network initiates a broadcasted PING message to announce its presence. The PING message is then forwarded to its neighbors and initiates a backpropagated PONG message, which contains information about the peer, such as the IP address, number and size of the data items. Search (QUERY and QUERY RESPONSE) Messages. QUERY contains a user specified search string that each receiving peer matches against locally stored file names and it is broadcast. QUERY RESPONSE messages are backpropagated replies to QUERY messages and include information necessary to download a file. File Transfer (GET and PUSH) Messages. File downloads are performed directly between two peers using these types of messages. Therefore, to become a member of the network, a servent (peer) has to open one or many connections with other peers that are already in the network. With such a dynamic network environment, to cope with the unreliability after joining the network, a peer periodically PINGs its neighbors to discover other participating peers. s decide where to connect in the network based only on local information. Thus, the entire application-level network has servents as its peers and open TCP connections as its links, forming a dynamic, self-organizing network of independent entities. The latest versions of Gnutella uses the notion of superpeers or ultra-peers [11] (peers with better bandwidth connectivity), to help improve the routing performance of the network. However, it is still limited by the flooding mechanism used for communications across ultra-peers. Moreover, the ultra-peer approach makes a binary decision about a peer s capacity (ultra-peer or not) and to our knowledge, it has no mechanism to dynamically adapt the ultra-peer-client topologies as the system evolves. Ultra-peers perform query processing on behalf of their leaf peers. When a peer joins the network as a leaf, it selects a number of ultra-peers, and then it publishes its file list to those ultra-peers. A query for a leaf peer is sent to an ultra-peer, which floods the query to its ultra-peer neighbors up to a limited number of hops. Dynamic querying [74] is a search technique whereby queries that return fewer results are re-flooded deeper into the network. Saroiu et al. [75] examined the bandwidth, latency, availability, and file sharing patterns of the peers in Gnutella and Napster, and highlighted the existence of significant heterogeneity in both systems. Krishnamurthy et al. [76] propose a cluster-based architecture for P2P systems (CAP), which uses a network-aware clustering technique (based on a central clustering server) to group peers into clusters. Each cluster has one or more delegate peers that act as directory servers for objects stored at peers within the same cluster. Chawathe et al. [73] propose a model called Gia, by modifying Gnutella s algorithm to include flow control, dynamic topology adaptation, one-hop replication, and careful attention to peer heterogeneity. The simulation results suggest that these modifications provide three to five orders of magnitude improvement in the total capacity of the system while retaining significant robustness to failures. Thus, making a few simple changes to Gnutella s search operations would result in dramatic improvements in its scalability. FASTTRACK/KAZAA FastTrack [65] P2P is a decentralized file-sharing system that supports meta-data searching. s form a structured overlay of super-peer architectures to make search more efficient, as shown in Fig. 9. Super-peers are peers with high bandwidth, disk space, and processing power, and have volunteered to be elected to facilitate search by caching the meta-data. The ordinary peers transmit the meta-data of the data files they are sharing to the super-peers. All the queries are also forwarded to the super-peer. Then, Gnutella-type broadcast-based search is performed in a highly pruned overlay network of superpeers. The P2P system can exist without any super-peer, but this would result in worse query latency. However, this approach still consumes bandwidth so as to maintain the index at the super-peers on behalf of the peers that are connected. The super-peers still use a broadcast protocol for search, and the lookup queries are routed to peers and super-peers that have no relevant information to the query. Both KaZaA [66] and Crokster [77] are both FastTrack applications. As mentioned, KaZaA is based on the proprietary Fast- Track protocol which uses specially designated super-peers that have higher bandwidth connectivity. Pointers to each peer s data are stored on an associated super-peer, and all queries are routed to the super-peers. Although this approach seems to offer better scaling properties than Gnutella, its design has not been analyzed. There have been proposals to incorporate this approach into the Gnutella network [11]. The KaZaA peer-to-peer file sharing network client supports a similar behavior, allowing powerful peers to opt-out of network support roles that consume CPU and bandwidth. KaZaA file transfer traffic consists of unencrypted HTTP transfers; all transfers include KaZaA-specific HTTP headers (e.g., X-KaZaA-IP). These headers make it simple to distinguish between KaZaA activity and other HTTP activity. The KaZaA application has an auto-update feature, meaning a running instance of KaZaA will periodically check for updated versions of itself. If it is found, it downloads the new executable over the KaZaA network. A power-law topology, commonly found in many practical IEEE Communications Surveys & Tutorials Second Quarter
15 1 2: Object2 3: Object3 Query Object2 RESPONSE 2 GET Object2 Supernode Object upload networks such as WWW [78, 79], has the property that a small proportion of peers have a high out-degree (i.e., have many connections to other peers), while the vast majority of peers have a low out-degree (i.e., have connections to few peers). Formally, the frequency f d of peers with out-degree d exhibits a power-law relationship of the form f d d a ; a < 0. This is the Zipf property with Zipf distributions looking linear when plotted on a log-log scale. Faloutsos et al. [80] have found that Internet routing topologies follow this power-law relationship with a 2. However, Gummadi et al. [81] observes that the KaZaA measured popularity of the file-sharing workload does not follow a Zipf distribution. The popularity of the most requested objects (mostly large, immutable video and audio objects) is limited, since clients typically fetch objects at most once, unlike the Web. Thus, the popularity of KaZaA s objects tends to be short-lived, and popular objects tend to be recently born. There is also significant locality in the KaZaA workload, which means that substantial opportunity exists for caching to reduce wide-area bandwidth consumption. BITTORRENT Supernode 2 Object upload 2: Object2 BitTorrent [67] is a centralized P2P system that uses a central location to manage users downloads. This file distribution network uses tit-for-tat (peer responds with the same action that its other collaborating peer performed previously) as a method of seeking. The protocol is designed to discourage free-riders by having the peers choose other peers from which the data has been received. s with high upload speed will probably also be able to download at a high speed, thus achieving high bandwidth utilization. The download speed of a peer will be reduced if the upload speed has been limited. This will also ensure that content will be spread among peers to improve reliability. The architecture consists of a central location, which is a tracker that is connected to when you download a.torrent file, which contains information about the file, its length, name, and hashing information, and URL of a tracker, as illustrated in Fig. 10. The tracker keeps track of all the peers who have the file (both partially and completely) and lookup peers to connect with one another for downloading and uploading. The trackers use a simple protocol layered on top of Supernode 3 3: Object3 n Figure 9. FastTrack peers connect to Superpeers whereby the search is routed through the Superpeers and downloads are done from the peer peers. HTTP in which a downloader sends information about the file it is downloading and the port number. The tracker responds with a random list of contact information about the peers that are downloading the same file. Downloaders then use this information to connect to each other. A downloader that has the complete file, known as a seed, must be started and must send out at least one complete copy of the original file. BitTorrent cuts files into pieces of fixed size (256 Kbytes) so as to track the content of each peer. Each downloader peer announces to all of its peers the pieces it has, and uses SHA1 to hash all the pieces that are included in the.torrent file. When a peer finishes downloading a piece and checks that the hash matches, it announces that it has that piece to all of its peers. This is to verify data integrity. connections are symmetrical. Messages sent in both directions look the same, and data can flow in either direction. When data is being transmitted, downloader peers keep several requests (for pieces of data) queued up at once in order to achieve good TCP performance. This is known as pipelining. Requests that cannot be written out to the TCP buffer immediately are queued up in memory rather than kept in an application-level network buffer, so they can all be thrown out when a choke happens. Choking is a temporary refusal to upload; downloading can still happen and the connection does not need to be renegotiated when choking stops. Choking is done for several reasons. TCP congestion control behaves very poorly when sending over many connections at once. Additionally, choking lets each peer use a tit-for-tat-like algorithm to ensure that they achieve a consistent download rate. There are several criteria that a good choking algorithm should meet. It should cap the number of simultaneous uploads for good TCP performance. It should avoid choking and unchoking quickly, known as fibrillation. It should reciprocate service access to peers who let it download. Finally, it should try out unused connections once in a while to find out if they might be better than the currently used connections, known as optimistic unchoking. The currently deployed BitTorrent choking algorithm avoids fibrillation by only changing the peer that is choked once every ten seconds. It does reciprocation, and the number of uploads are capped by unchoking the four peers with the best download rates and greatest interest. s that have a better upload rate but are not interested get unchoked and if they become interested, the worst uploader gets choked. If a.torrent.torrent server Tracker n Figure 10. BitTorrent architecture consists of centralized Tracker and.torrent file. 86 IEEE Communications Surveys & Tutorials Second Quarter 2005
16 downloader has a complete file, it uses its upload rate rather than its download rate to decide which to unchoke. For optimistic unchoking, at any one time there is a single peer that is unchoked regardless of its upload rate. If this peer is interested, it counts as one of the four allowed downloaders. s that are optimistically unchoked rotate every 30 seconds. OVERNET/EDONKEY2000 Overnet/eDonkey [68, 69] is a hybrid two-layer P2P information storage network composed of client and server, which are used to publish and retrieve small pieces of data by creating a file-sharing network. This architecture provides features such as concurrent download of a file from multiple peers, detection of file corruption using hashing, partial sharing of files during downloading, and expressive querying methods for file search. To join the network, the peer (client) needs to know the IP address and port of another peer (server) in the network. It then bootstraps from the other peer. The clients connect to a server and register the object files that they are sharing by providing the meta-data describing the object files. After registration, the clients can either search by querying the meta-data or request a particular file through its unique network identifier, thus providing guaranteed service to locate popular objects. Servers provide the locations of object files when requested by clients, so that clients can download the files directly from the indicated locations. DISCUSSION ON UNSTRUCTURED P2P OVERLAY NETWORK The Unstructured P2P centralized overlay model was first popularized by Napster. This model requires some managed infrastructure (the directory server) and show some scalability limits. A flooding-requests model for decentralized P2P overlay systems such as Gnutella, whereby each peer keeps a userdriven neighbor table to locate data objects, are quite effective in locating popular data objects, thanks to the power-law property of user-driven characteristics. However, it can lead to excessive network bandwidth consumption, and remote or unpopular data objects may not be found due to the limit of lookup horizon typically imposed by TTL. The argument is that DHT-based systems, while more efficient at many tasks and offering strong theoretical fundamentals to guarantee a key to be found if it exists, are not well suited for mass-market file sharing. They do not capture the semantic object relationships between its name and its content or metadata. In particular, DHT-based ability to find exceedingly rare objects is not required in a mass-market file sharing environment, and their ability to efficiently implement keyword search is still not proven. In addition, they use precise placement algorithms and specific routing protocols to make searching efficient. However, these Structured P2P overlay systems have not been widely deployed, and their ability to handle unreliable peers has not been tested. Thus, in the research community, efforts are being made to improve the lookup properties of Unstructured P2P overlays to include flow control, dynamic geometric topology adaptation [82, 83], one-hop replication, peer heterogeneity, etc. Freenet, like Chord, does not assign responsibility for data to specific peers, and its lookups take the form of searches for cached copies. This prevents it from guaranteeing retrieval of existing data or from providing low bounds on retrieval costs. But Freenet provides anonymity and it introduces a novel indexing scheme whereny files are identified by content-hash keys and by secured signed-subspace keys to ensure that only one object owner can write to a file and anyone can read it. P2P overlay designs using DHTs share similar characteristics as Freenet an exact query yields an exact response. This is not surprising since Freenet uses a hash function to generate keys. Recent research in [84] shows that changing Freenet s routing table cache replacement scheme from LRU to enforcing clustering in the key space can significantly improve performance. This idea is based on the intuition from the small-world models [39] and theoretical results by Kleinberg [39]. Version 0.6 of the Gnutella protocol [9, 10] adopted the concept of ultra-peers, which are high-capacity peers that act as proxies for lower-capacity peers. One of the main enhancements is the Query Routing Protocol (QRP), which allows the leaf peers to forward an index of object name keywords to its ultra-peers [85]. This allows the ultra-peers to have their leaves receive lookup queries when they have a match, and subsequently, it reduces the lookup query traffic at the leaves. A shortcoming of QRP is that the lookup query propagation is independent of the popularity of the objects. The Dynamic Query Protocol [86] addressed this by letting the leaf peers send single queries to high-degree ultra-peers, which adjust the lookup queries TTL bounds in accordance with the number of received lookup query results. The Gnutella UDP Extension for Scalable Searches (GUESS) [87] also aimed to reduce the number of lookup queries by repeatedly querying single ultra-peers with a TTL of 1, to limit the load on each lookup query. As described earlier, Chawathe et al. [73] improve the Gnutella design in their Gia system, by incorporating an adaptation algorithm so that peers are attached to high-degree peers, and by providing a receiver-based token flow control for sending lookup queries to neighbors. Instead of flooding, they make use of a random walk search algorithm as the system keeps pointers to objects in neighboring peers. However, in [87] they proposed that Unstructured P2P overlays such as Gnutella can be built on top of Structured P2P overlays to help reduce the lookup query overhead and overlay maintenance traffic. They used the collapse point lookup query rate (defined as the per node query rate at which the successful query rate falls below 90 percent) and the average hopcounts prior to collapse. However, the comparison was done in a static network scenario with the older Gnutella and not the enhanced version of Gnutella. BitTorrent, a second-generation P2P overlay system, achieves higher levels of robustness and resource utilization based on its incentives cooperation technique for file distribution. The longest and most comprehensive measurement study of a BitTorrent P2P system [88] provides more insight by comparing a detailed measurement study of BitTorrent with other popular P2P file-sharing systems, such as FastTrack/KaZaA, Gnutella, Overnet/eDonkey, and Direct- Connect, based on five characteristics: 1 Popularity: Total number of users participating over a certain period of time. 2 Availability: System availability depending on contributed resources. 3 Download Performance: Contrast between size of data and the time required for download. 4 Content Lifetime: Time period when data is injected into the system until no peer is willing to share the data anymore. 5 Pollution Level: Fraction of corrupted content spread throughout the system. FastTrack/KaZaA has the largest file sharing community, with Overnet/eDonkey and BitTorrent gaining popularity. The popularity of the BitTorrent system is influenced by the avail- IEEE Communications Surveys & Tutorials Second Quarter
17 ability of the central components in terms of its number of downloads and technical faults in the system. Availability has a significant influence on popularity. FastTrack/KaZaA, being more advanced in architecture, has good availability because of its Super-s that allow the network to scale very well by creating indexing. Gnutella and Overnet/eDonkey provide full and partial distribution of the responsibility for shared files, respectively. The availability of content in BitTorrent is unpredictable and vulnerable to potential failures, due to its lack of decentralization. BitTorrent is well-suited for download performance due to its advanced download distribution protocol. Overnet/eDonkey takes an opposite approach by offering powerful searching capabilities and queue-based scheduling of downloads, which can have longer waiting times. The lack of archive functionality in BitTorrent results in relatively short content lifetimes. FastTrack/KaZaA, which uses directorylevel sharing policy, allows data files to be located as long as the peer holding the data file stays connected. The FastTrack/ KaZaA system does not limit the number of fake files in the overlay but it allows users to identify correct files based on hash-code verification. BitTorrent prevents fake files from floating in the system. The arising use of firewalls and NATs are growing problems for P2P systems because of the reduced download speed. One proposal [89] tries to solve the firewall problems by designing a hybrid CDN structure with a P2Pbased streaming protocol in the access network based on an empirical study of BitTorrent, which highlighted the need for additional freeloader protection and the potential negative effect of firewalls on download speeds. A fluid model for Bit- Torrent-like networks was proposed [90] to capture the behavior of the system when the arrival rate is small, and to study the steady-state network performance. The study also provided expressions for the various parameters, such as average number of sees, the average number of downloaders, and the average downloading time, and proved that Nash equilibrium exists for each peer that chooses its uploading bandwidth to be equal to the actual uploading bandwidth. It is also interesting to note that KaZaA [81] is not pure power-law networks with Zipf distribution properties; whereas the analysis in [91] shows that Gnutella overlay networks have topologies that are power-law random graphs. Gnutella overlay network bears strong characteristics of power-law networks where a small number of nodes have many peers and majority have few peers. Also, Gnutella nodes appear to group into clusters resembling small world networks. This may be attributed to the behaviors of the users of these P2P networks. Many research work on power-law networks [78, 80, 92, 93] show that networks as diverse as the Internet, organize themselves so that most peers have few links while a small number of peers have a large number of links. The interesting paper by Adamic et al. [94] studies random-walk search strategies in power-law networks, and discovers that by changing walkers to seek out high degree peers, the search performance can be optimized greatly. Several search techniques for Unstructured P2P networks are discussed in [95]: iterative deepening, directed Breadth-First Traversal Search (BFS) and local indices. Networks with power-law organizational structure, display an unexpected degree of robustness [96], that is, the ability of the peers to communicate unaffectedly even by extremely high failure rates. Although the power-law Gnutella [75] overlay network is highly resilient in the face of random failures, but it is highly prone to targeted attacks and it requires to reduce the network dependence on a small number of highly connected, easy to attack peers. Instead of using DHT as building blocks in distributed applications, SkipNet [97] is an overlay based on Skip Lists that organizes peers and data primarily by their sorted string names, as Skip Lists do, rather than by hashes of those names. In this way, SkipNet supports useful locality properties as well as the usual DHT functionality. In addition, some recent research, e.g., Loo et al. [98], proposes the design of a hybrid model by using Structured DHT techniques to locate rare object items, and Unstructured flooding to locate highly replicated contents. More interestingly, the performance studies on lightweight hierarchical P2P overlay networks [82, 83] make use of the geometric structure in Yao-Graphs [99] and Highways [100] proximity placement scheme to assign geometric co-ordinates to Super-s and s with respect to the underlying network conditions. Because of the lightweight structure of Yao-Graphs, the resulting hierarchical P2P networks have promising properties with regard to scalability and performance, while still offering the benefits of resiliency. All of the security, privacy, and trust issues discussed in the Structured P2P overlay network section applies to Unstructured P2P overlay networks. It is worthwhile to highlight the work from Bellovin [101]. The paper reports on the difficulty in limiting Napster s and Gnutella s use via firewalls and how information can be leaked through search queries in the overlay network. The work highlighted concern over Gnutella s push feature, intended to work around firewalls, which might be useful for distributed denial of service attacks. Napster s centralized architecture might be more secure toward such attacks due to a centralized trusted server. Table 2 summarizes the comparisons of Unstructured P2P networks. While by no means comprehensive, we believe that it captures the essence of the discussion and analysis done earlier. CONCLUDING REMARKS This article has presented various schemes in Structured and Unstructured P2P overlay networks that have been proposed by researchers. Which of these P2P overlay networks is best suited depends on the application and its required functionalities and performance metrics, e.g. scalability, network routing performance, location service, file sharing, content distribution, and so on. Several of these schemes are being applied to the sharing of music, replication of electronic address books, multi-player games, provisioning of mobile, location, or adhoc services, and the distribution of workloads of mirrored Web sites. Finally, we close this survey with our thoughts on some directions for the future in P2P overlay networking research. The concerns of how well the P2P overlay networks virtual topology maps on to the physical network infrastructure, which has an impact on the additional stress on the infrastructure, will undoubtedly incur costs for the service providers. It would be useful to provide a quantitative evaluation on P2P overlay applications and Internet topology matching and the scalability of P2P overlay applications by the efficient use of the underlying physical network resources. Having some sort of incentive model using economic and game theories for P2P peers to collaborate is crucial to create an economy of equilibrium. When non-cooperative users benefit from free-riding on others resources, the tragedy of the commons [102] is inevitable. Such incentives implementation in P2P overlay services would also provide a certain level of self-regulatory auditing and accounting behavior for resource sharing. Trust and reputation is also important to achieve secured and trustworthy P2P overlay communications among the peers. For example, Kademlia developed a trust and secured architecture for routing and location service, by discouraging free-riding based on honesty, and routing away from the 88 IEEE Communications Surveys & Tutorials Second Quarter 2005
18 Algorithm taxonomy Unstructured P2P Overlay Network Comparisons Freenet Gnutella FastTrack/KaZaA BitTorrent Overnet/eDonkey 2000 Decentralization Loosely DHT functionality. Topology is flat with equal peers. No explicit central server. s are connected to their Super-s. Centralized model with a Tracker keeping track of peers. Hybrid two-layer network composed of clients and servers. Architecture Keywords and descriptive text strings to identify data objects. Flat and ad-hoc network of servents (peers). Flooding request and peers download directly. Two-level hierarchical network of Super-s and peers. s request information from a central Tracker. Servers provide the locations of files to requesting clients for download directly. Lookup protocol Keys, Descriptive Text String search from peer to peer. Query flooding. Super-s. Tracker. Client-server peers. System parameters None. None. None..torrent file. None. Routing performance Guarantee to locate data using Key search until the requests exceeded the Hops-To-Live (HTL) limits. No guarantee to locate data; improvements made in adapting ultrapeer-client topologies; good performance for popular content. Some degree of guarantee to locate data, since queries are routed to the Super-s, which has better scaling; good performance for popular content. Guarantee to locate data and guarantee performance for popular content. Guarantee to locate data and guarantee performance for popular content. Routing state Constant. Constant. Constant. Constant but choking (temporary refusal to upload) may occur. Constant. s join/leave Constant. Constant. Constant. Constant. Constant with bootstrapping from other peers and connect to server to register files being shared. Security Low; suffers from man-in-middle and Trojan attacks. Low; threats: flooding, malicious content, virus spreading, attack on queries, and denial of service attacks. Low; threats: flooding, malicious or fake content, viruses, etc. Spyware monitors the activities of peers in the background. Moderate; centralized Tracker manages file transfer and allows more control, which makes it much harder to fake IP addresses, port numbers, etc. Moderate; threats similar to those in FastTrack and Bit-Torrent. Reliability/ fault resiliency No hierarchy or central point of failure exists. Degradation of the performance; peers receive multiple copies of replies from peers that have the data; requester peers can retry. The ordinary peers are reassigned to other Super-s. The Tracker keeps track of the peers and availability of the pieces of files; avoid choking by fibrillation by changing the peer that is choked once every ten seconds. Reconnecting to another server; will not receive multiple replies from peers with available data. delegating responsibility across peers, P2P overlays will improve the routing scalability. Future research would aim to reduce the stretch (ratio of overlay path to underlying network path) routing metric based on scalable and robust proximity calculations (e.g., in geometric space). This leads to improved P2P overlay operations performance globally. A mixed set of metrics which include delay, throughput, available bandwidth, and packet loss would provide a more effin Table 2. A comparison of various unstructured P2P overlay network schemes. defective or malicious peer. Proximity in P2P overlay routing is an important factor in the routing decision for P2P overlay networks which could improve global routing properties. There is ongoing research in this area based on mapping the peers into geometric coordinate-based space [100, ] and heuristic proximity routing optimizations [ ]. Taking heterogeneity of the peers and its geometric properties [82, 83] into account when IEEE Communications Surveys & Tutorials Second Quarter
19 cient global routing optimization. Cross-application of Internet P2P overlay networking models in mobile, wireless, or ad-hoc networks. Because of their similar features, such as self-organizing, peer-to-peer routing, and resilient communications, application of P2P overlay approaches would allow mobile peers to have optimized flow control, load balancing mechanism, proximityaware, and QoS routing [118]. There are also exciting challenges and issues for applying P2P overlay networks in a hybrid wired and wireless network environment. We see the future of P2P overlay networks inexorably linked to the takeup and subsequent commercial success of P2P overlay computing, personal area and ad-hoc networking, mobile location-based services, mirrored content delivery, and networked file-sharing, as examples. We may also conjecture that the prevalent problems of the Internet controlling spam, maintaining directory services, multicasting content, etc. have intuitive solutions from various P2P overlay network schemes. However, in order to move forward, the development community needs to understand the applicability of various schemes for Structured and Unstructured P2P overlay network models. This survey has been a modest attempt to address this need. ACKNOWLEDGMENT The authors would like to thank the reviewers for their valuable comments. REFERENCES [1] C. Plaxton, R. Rajaraman, and A. Richa, Accessing Nearby Copies of Replicated Objects in a Distributed Environment, Proc. 9th Annual ACM Symp. Parallel Algorithms and Architectures, [2] L. Breslau et al., Web Caching and zipf-like Distribution: Evidence and Implications, Proc. IEEE INFOCOM, [3] D. R. Karger et al., Consistent Hashing and Random Trees: Distributed Caching Protocols for Relieving Hot Spots on the World Wide Web, Proc. ACM Symp. Theory of Comp., May 1997, pp [4] A. Rowstron and P. Druschel, Pastry: Scalable, Distributed Object Location and Routing for Large-scale -to-peer Systems, Proc. Middleware, [5] S. Ratnasamy et al., A Scalable Content Addressable Network, Proc. ACM SIGCOMM, 2001, pp [6] I. Stoica, R. Morris et al., Chord: A Scalable -to- Lookup Protocol for Internet Applications, IEEE/ACM Trans. Net., vol. 11, no. 1, 2003, pp [7] B. Y. Zhao et al., Tapestry: A Resilient Global-Scale Overlay for Service Deployment, IEEE JSAC, vol. 22, no. 1, Jan. 2004, pp [8] Napster, available at [9] (2001) Gnutella development forum, the Gnutella v0.6 protocol, available at gdf/files/ [10] (2002) Gnucleus, the Gnutella Web caching system, available: at [11] (2002) Gnutella ultrapeers, available at [12] F. Dabek et al., Towards a Common API for Structured to-peer Overlays, Proc. 2nd Int l. Wksp. -to- Systems (IPTPS 2003), Berkeley, California, USA, Feb , [13] B. Karp et al., Spurring Adoption of DHTs with OpenHash, a Public DHT Service, Proc. 3rd Int l. Wksp. -to- Systems (IPTPS 2004), Berkeley, California, USA, Feb , [14] P. Maymounkov and D. Mazieres, Kademlia: A -to- Information System Based on the XOR Metric, Proc. IPTPS, Cambridge, MA, USA, Feb. 2002, pp [15] D. Malkhi, M. Naor, and D. Ratajczak, Viceroy: A Scalable and Dynamic Emulation of the Butterfly, Proc. ACM PODC 2002, Monterey, CA, USA, July 2002, pp [16] P. Francis, Yoid: Extending the Internet Multicast Architecture, unpublished, Apr index.html [17] J. Kubiatowicz et al., Oceanstore: An Architecture for Global-Scale Persistent Storage, Proc. ACM ASPLOS, Nov [18] W. J. Bolosky et al., Feasibility of a Serverless Distributed File System Deployed on an Existing Set of Desktop PCs, Proc ACM SIGMETRICS Int l. Conf. Measurement and Modeling of Comp. Sys., June 2000, pp [19] M. Waldman, A. D. Rubin, and L. F. Cranor, Publius: A Robust, Tamper-evident, Censorship-resistant, Web Publishing System, Proc. 9th USENIX Security Symp., Denver, CO, USA, [20] D. Karger et al., Consistent Hashing and Random Trees: Distributed Caching Protocols for Relieving Hot Spots on the World Wide Web, Proc. 29th Annual ACM Symp. Theory of Comp., May 1997, pp [21] National Institute of Standards and Technology (NIST), Secure hash standard, U.S. Department of Commerce, National Technical Information Service FIPS 180-1, Apr [22] F. Dabek et al., Wide-area Cooperative Storage with CFs, Proc. 18th ACM Symp. Operating Systems Principles, 2001, pp [23] R. Cox, A. Muthitacharoen, and R. Morris, Serving DNS using Chord, Proc. First Int l. Wksp. -to- Systems, Mar [24] Y. Rekhter and T. Li, An Architecture for IP Address Allocation with CIDR, IETF Internet draft RFC1518, available at [25] S. Rhea et al., Maintenance-free Global Storage in Oceanstore, IEEE Internet Comp., [26] L. Peterson et al., A Blueprint for Introducing Disruptive Technology into the Internet, SIGCOMM Comp. Commun. Rev., vol. 33, no. 1, 2003, pp [27] S. Q. Zhuang et al., Bayeux: An Architecture for Scalable and Fault-Tolerant Wide-Area Data Dissemination, Proc. 11th Int l. Wksp. Network and Op. Sys. Support for Digital Audio and Video, 2001, pp [28] F. Zhou et al., Approximate Object Location and Spam Filtering on -to- Systems, Proc. Middleware, June [29] A. Rowstron et al., SCRIBE: The Design of a Large-Scale Event Notification Infrastructure, Proc. 3rd Int l. Wksp. Networked Group Commun. (NGC2001), London, UK, Nov. 2001, pp [30] M. Castro et al., SCRIBE: A Large-Scale and Decentralized Application-Level Multicast Infrastructure, IEEE JSAC (special issue on Network Support for Multicast Communications), October [31] M. Castro et al., Splitstream: High-Bandwidth Multicast in Cooperative Environments, Proc. 19th ACM Symp. Operating Systems Principles, Oct , 2003, pp [32] S. Iyer, A. Rowstron, and P. Druschel, Squirrel: A Decentralized -to- Web Cache, Proc. 21st Symp. Principles of Distributed Computing (PODC), Monterey, California, USA, July [33] P. Druschel and A. Rowstron, PAST: A Large-Scale, Persistent -to- Storage Utility, Proc. 8th Wksp. Hot Topics in Op. Sys. (HotOS-VIII). Schloss Elmau, Germany: IEEECompSoc, May [34] A. Rowstron and P. Druschel, Storage Management and Caching in Past, a Large-Scale, Persistent -to- Storage Utility, Proc. 18th ACM Symp. Operating Systems Principles, Oct. 2001, pp [35] L. P. Cox, C. D. Murray, and B. D. Noble, Pastiche: Making Backup Cheap and Easy, SIGOPS Op. Sys. Rev., vol. 36, no. SI, 2002, pp [36] A. Muthitacharoen, B. Chen, and D. Mazières, A Low-Bandwidth Network File System, Proc. 18th ACM Symp. Op. Sys. principles, 2001, pp [37] U. Manber, Finding Similar Files in a Large File System, Proc. USENIX Winter 1994 Conf., Jan. 1994, pp [38] H. J. Siegel, Interconnection Networks for SIMD Machines, Computer, vol. 12, no. 6, 1979, pp IEEE Communications Surveys & Tutorials Second Quarter 2005
20 [39] J. Kleinberg, The Small-World Phenomenon: An Algorithm Perspective, Proc. 32nd Annual ACM Symp. Theory of Comp., 2000, pp [40] L. Barriére et al., Efficient Routing in Networks with Long Range Contacts, Proc. 15th Int l. Conf. Distributed Computing, vol. 2180, 2001, pp [41] S. Rhea et al., Handling Churn in a DHT, Proc. 2nd Int l. Wksp. -to- (IPTPS 2003), Feb [42] M. Castro, M. Costa, and A. Rowstron, Performance and Dependability of Structured -to- Overlays, Proc Int l. Conf. Dependable Sys. and Net., Palazzo dei Congressi, Florence, Italy, June 28 July [43] D. Liben-Nowell, H. Balakrishnan, and D. Karger, Analysis of the Evolution of -to- Systems, Proc. Annual ACM Symp. Principles of Distributed Comp., Monterey, California, USA, [44] L. Alima et al., Dks(n,k,f): A Family of Low Communication, Scalable and Fault-Tolerant Infrastructures for P2P Applications, Proc. 3rd IEEE/ACM Int l. Symp. Cluster Comp. and the Grid, Monterey, California, USA, 2003, pp [45] X. Li and C. Plaxton, On Name Resolution in -to- Networks, Proc. 2nd ACM Int l. Wksp. Principles of Mobile Comp., Monterey, California, USA, 2002, pp [46] F. Kaashoek and D. Karger, Koorde: A Simple Degree-Optimal Hash Table, Proc. 2nd Int l. Wksp. -to- Systems (IPTPS 03), Berkeley, CA, USA, Feb , [47] I. Abraham, D. Malkhi, and O. Dubzinski, Land: Stretch (1+epsilon) Locality Aware Networks for DHTS, Proc. ACM- SIAM Symp. Discrete Algorithms (SODA 2004), New Orleans, LA., USA, [48] N. D. de Bruijn, A Combinatorial Problem, Koninklijke Netherlands: Academe Van Wetenschappen, vol. 49, 1946, pp [49] M. Naor and U. Wieder, Novel Architectures for P2P Applications: The Continuous-Discrete Approach, Proc. 15th Annual ACM Symp. Parallel Algorithms and Architectures (SPAA 2003), San Diego, California, USA, June , pp [50] D. Loguinov, A. Kumar, and S. Ganesh, Graph-Theoretic Analysis of Structured -to- Systems: Routing Distances and Fault Resilience, Proc. ACM SIGCOMM, Karlsruhe, Germany, Aug , pp [51] M. Naor and U. Wieder, A Simple Fault Tolerant Distributed Hash Table, Proc. 2nd Int l. Wksp. -to- Systems (IPTPS 03), Berkeley, California, USA, Feb [52] P. Fraigniaud and P. Gauron, The Content-Addressable Networks D2B, Laboratoire de Recherche en Informatique, Universit e de Paris Sud, Tech. Rep. Technical Report 1349, Jan [53] E. Sit and R. Morris, Security Considerations for -to- Distributed Hash Tables, Proc. 1st Int l. Wksp. -to- Systems (IPTPS), Cambridge, MA, USA, Mar [54] M. Castro et al., Secure Routing for Structured -to- Overlay Networks, SIGOPS Oper. Syst. Rev., vol. 36, no. SI, 2002, pp [55] A. Singh et al., Defending Against Eclipse Attacks on Overlay Networks, Proc. SIGOPS European Wksp., Leuven, Belgium, Sept [56] D. S. Wallach, A Survey of -to- Security Issues, Proc. Int l. Symp. Software Security, Tokyo, Japan, November [57] E. K. Lua et al., Barterroam: A Novel Mobile and Wireless Roaming Settlement Model, Proc. QofIS, 2004, pp [58] C. Buragohain, D. Agrawal, and S. Suri, A Game-Theoretic Framework for Incentives in P2P Systems, Proc. IEEE P2P 2003, Linkoping, Sweden, Sept [59] P. Golle et al., Incentives for Sharing in -to- Networks, Lecture Notes in Computer Science, vol. 2232, pp. 75+, [60] K. Lai et al., Incentives for Cooperation in -to- Networks, Proc. Wksp. Economics of -to- Systems, Linkoping, Sweden, June [61] J. R. Douceur, The Sybil Attack, Proc. 1st Int l. Wksp. to- Systems, Mar , pp [62] R. Dingledine, M. J. Freedman, and D. Molnar, Accountability Measures for -to- Systems, -to-: Harnessing the Power of Disruptive Technologies, D. Derickson, Ed. O Reilly and Associates, Nov [63] Z. Despotovic and K. Aberer, A Probabilistic Approach to Predict s Performance in P2P Networks, Proc. 8th Int l. Wksp. Cooperative Information Agents (CIA 2004), Erfurt, Germany, Sept [64] I. Clarke et al., Freenet: A Distributed Anonymous Information Storage and Retrieval System, available at freenet.pdf, [65] Fasttrack -to- Technology Company, available at [66] Kazaa Media Desktop, available at [67] Bittorrent, available at [68] The Overnet File-sharing Network, available at http: //, [69] Overnet/edonkey2000, available at edonkey2000.com/, [70] American National Standard Institute (ANSI), Public Key Cryptography Using Irreversible Algorithms Part 2: The Secure Hash Algorithm (SHA-1), ANSI Standards, Tech. Rep. ANSI X , [71] P. Ganesan, Q. Sun, and H. Garcia-Molina, Yappers: A -to- Lookup Service over Arbitrary Topology, Proc. IEEE INFOCOM 2003, San Francisco, USA, Mar. 30 Apr. 1, [72] Q. Lv, S. Ratnasamy, and S. Shenker, Can Heterogeneity Make Gnutella Scalable? Proc. 1st Int l. Wksp. to- Systems (IPTPS), Cambridge, MA, USA, Feb [73] Y. Chawathe et al., Making Gnutella-like P2P Systems Scalable, Proc. ACM SIGCOMM, Karlsruhe, Germany, Aug [74] Gnutella Proposals for Dynamic Querying, available at [75] S. Saroiu, P. K. Gummadi, and S. D. Gribble, A Measurement Study of -to- File Sharing Systems, Proc. Multimedia Comp. and Net. (MMCN), San Jose, California, USA, Jan [76] B. Krishnamurthy, J. Wang, and Y. Xie, Early Measurement of a Cluster-Based Architecture for P2P Systems, Proc. ACM SIGCOMM Internet Measurement Wksp., San Francisco, USA, Nov [77] Grokster, available at [78] A.-L. Barabási et al., Power-Law Distribution of the World Wide Web, Science, vol. 287, [79] R. Albert, H. Jeong, and A.-L. Barabási, Diameter of the World Wide Web, Nature, vol. 401, 1999, pp [80] M. Faloutsos, P. Faloutsos, and C. Faloutsos, On Power-Law Relationships of the Internet Topology, Proc. SIGCOMM 1999, [81] K. P. Gummadi et al., Measurement, Modeling, and Analysis of a -to- File Sharing Workload, Proc. SOSP, Bolton Landing, New York, USA, Oct [82] M. Kleis, E. K. Lua, and X. Zhou, Hierarchical -to- Networks using Lightweight Superpeer Topologies, Proc. 10th IEEE Symp. Comp. and Commun. (ISCC 2005), La Manga del Mar Menor, Cartagena, Spain, June [83] M. Kleis, E. K. Lua, and X. Zhou, A Case for Lightweight Superpeer Topologies. KiVS Kurzbeiträge und Wksp., Mar. 2005, pp [84] A. Goel and R. Govindan, Using the Small-World Model to Improve Freenet Performance, Comp. Net. Journal, vol. 46, no. 4, Nov. 2004, pp [85] A. Singla and C. Rohrs, Ultrapeers: Another Step Towards Gnutella Scalability, available at group/thegdf/files/proposals/working Proposals/Ultrapeer/ [86] A. Fisk, Gnutella Ultrapeer Query Protocol v0.1, available at gdf/files/proposals/working Proposals/search/Dynamic Querying/ [87] S. Daswani and A. Fisk, Gnutella UDP Extension for Scalable Searches (GUESS) v0.1, available at fisheye/viewrep/~raw,r=1.2/limecvs/core/guess 01.html [88] J. A. Pouwelse et al., A Measurement Study of the BitTor- IEEE Communications Surveys & Tutorials Second QuarterP
Interconnection Networks. Interconnection Networks. Interconnection networks are used everywhere!
Interconnection Networks Interconnection Networks Interconnection networks are used everywhere! Supercomputers connecting the processors Routers connecting the ports can consider a router as a parallel:
Security in Structured P2P Systems
P2P Systems, Security and Overlays Presented by Vishal thanks to Dan Rubenstein Columbia University 1 Security in Structured P2P Systems Structured Systems assume all nodes behave Position themselves Survey of Peer-to-Peer File Sharing Technologies
Athens University of Economics and Business The ebusiness Centre () A Survey of Peer-to-Peer File Sharing Technologies White Paper Page 1 of 1 A Survey of Peer-to-Peer File Sharing Technologies
Definition. A Historical Example
Overlay Networks This lecture contains slides created by Ion Stoica (UC Berkeley). Slides used with permission from author. All rights remain with author. Definition Network defines addressing, routing,
Multicast vs. P2P for content distribution
Multicast vs. P2P for content distribution Abstract Many different service architectures, ranging from centralized client-server to fully distributed are available in today s world for Content Distribution
Calto: A Self Sufficient Presence System for Autonomous Networks
Calto: A Self Sufficient Presence System for Autonomous Networks Abstract In recent years much attention has been paid to spontaneously formed Ad Hoc networks. These networks can be formed without central
CS514: Intermediate Course in Computer Systems
: Intermediate Course in Computer Systems Lecture 7: Sept. 19, 2003 Load Balancing Options Sources Lots of graphics and product description courtesy F5 website () I believe F5 is market leader
GISP: Global Information Sharing Protocol a distributed index for peer-to-peer systems
GISP: Global Information Sharing Protocol a distributed index for peer-to-peer systems Daishi Kato Computer Science Department, Stanford University Visiting from NEC Corporation Abstract This paper proposes.
Global Server Load Balancing
White Paper Overview Many enterprises attempt to scale Web and network capacity by deploying additional servers and increased infrastructure at a single location, but centralized architectures,
Identity Theft Protection in Structured Overlays
Appears in Proceedings of the 1st Workshop on Secure Network Protocols (NPSec 5) Identity Theft Protection in Structured Overlays Lakshmi Ganesh and Ben Y. Zhao Computer Science Department, U. C. Santa)
Domain Name System for PeerHosting
Domain Name System for PeerHosting Arnav Aggarwal, B.Tech A dissertation submitted to the University of Dublin, in partial fulfilment of the requirements for degree of Master of Science in Computer Science Computer Networks
Introduction to Computer Networks Chen Yu Indiana University Basic Building Blocks for Computer Networks Nodes PC, server, special-purpose hardware, sensors Switches Links: Twisted pair, coaxial cable,
International Journal of Advanced Research in Computer Science and Software Engineering
Volume 2, Issue 9, September 2012 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: An Experimental
Computer Networks - CS132/EECS148 - Spring 2013 ------------------------------------------------------------------------------
Computer Networks - CS132/EECS148 - Spring 2013 Instructor: Karim El Defrawy Assignment 2 Deadline : April 25 th 9:30pm (hard and soft copies required) ------------------------------------------------------------------------------
Storage Systems Autumn 2009. Chapter 6: Distributed Hash Tables and their Applications André Brinkmann
Storage Systems Autumn 2009 Chapter 6: Distributed Hash Tables and their Applications André Brinkmann Scaling RAID architectures Using traditional RAID architecture does not scale Adding news disk implies
Computer Networking Networks
Page 1 of 8 Computer Networking Networks 9.1 Local area network A local area network (LAN) is a network that connects computers and devices in a limited geographical area such as a home, school, office
A Topology-Aware Relay Lookup Scheme for P2P VoIP System
Int. J. Communications, Network and System Sciences, 2010, 3, 119-125 doi:10.4236/ijcns.2010.32018 Published Online February 2010 (). A Topology-Aware Relay Lookup Scheme Publish/Subscribe Data Gathering Framework Integrating Wireless Sensor Networks and Mobile Phones
IT 10 066 Examensarbete 30 hp December 2010 A Publish/Subscribe Data Gathering Framework Integrating Wireless Sensor Networks and Mobile Phones He Huang Institutionen för informationsteknologi Department
Scalable Source Routing
Scalable Source Routing January 2010 Thomas Fuhrmann Department of Informatics, Self-Organizing Systems Group, Technical University Munich, Germany Routing in Networks You re there. I m here. Scalable,
Virtual PortChannels: Building Networks without Spanning Tree Protocol
. White Paper Virtual PortChannels: Building Networks without Spanning Tree Protocol What You Will Learn This document provides an in-depth look at Cisco's virtual PortChannel (vpc) technology, as
System Interconnect Architectures. Goals and Analysis. Network Properties and Routing. Terminology - 2. Terminology - 1
System Interconnect Architectures CSCI 8150 Advanced Computer Architecture Hwang, Chapter 2 Program and Network Properties 2.4 System Interconnect Architectures Direct networks for static connections Indirect
VXLAN: Scaling Data Center Capacity. White Paper
VXLAN: Scaling Data Center Capacity White Paper Virtual Extensible LAN (VXLAN) Overview This document provides an overview of how VXLAN works. It also provides criteria to help determine when and where
Interconnection Network Design
Interconnection Network Design Vida Vukašinović 1 Introduction Parallel computer networks are interesting topic, but they are also difficult to understand in an overall sense. The topological structure
An Active Packet can be classified as
Mobile Agents for Active Network Management By Rumeel Kazi and Patricia Morreale Stevens Institute of Technology Contact: rkazi,[email protected] Abstract-Traditionally, network management systems,
The Sierra Clustered Database Engine, the technology at the heart of
A New Approach: Clustrix Sierra Database Engine The Sierra Clustered Database Engine, the technology at the heart of the Clustrix solution, is a shared-nothing environment that includes the Sierra Parallel
FortiBalancer: Global Server Load Balancing WHITE PAPER
FortiBalancer: Global Server Load Balancing WHITE PAPER FORTINET FortiBalancer: Global Server Load Balancing PAGE 2 Introduction Scalability, high availability and performance are critical to the success.
Comparing Microsoft SQL Server 2005 Replication and DataXtend Remote Edition for Mobile and Distributed Applications
Comparing Microsoft SQL Server 2005 Replication and DataXtend Remote Edition for Mobile and Distributed Applications White Paper Table of Contents Overview...3 Replication Types Supported...3 Set-up &
Designing a Cloud Storage System
Designing a Cloud Storage System End to End Cloud Storage When designing a cloud storage system, there is value in decoupling the system s archival capacity (its ability to persistently store large volumes
Chapter 3. Enterprise Campus Network Design
Chapter 3 Enterprise Campus Network Design 1 Overview The network foundation hosting these technologies for an emerging enterprise should be efficient, highly available, scalable, and manageable. This | http://docplayer.net/8317983-T-he-electronic-magazine-of-o-riginal-peer-reviewed-survey-articles-abstract.html | CC-MAIN-2018-34 | refinedweb | 20,939 | 53.31 |
Mapping an input file to different RDDs
I have a text file consisting of columns of integers. Assuming that I have N number of columns, I need to have N-1 number of PairRDDs. Each PairRDD has one of the 0 to N-2 columns of my file as Key and the last column as Value. Number of columns in my file varies each time I run the program so I don't know the number of RDDs before the run.
Code below gives
task not serializable error.
val inputFile = sc.textFile(path).persist(); for (dim <- 0 to (numberOfColumns - 2)){ val temp = inputFile.map(line => { val lines = line.split(',') (lines(dim), lines(numberOfColumns - 1)) }) }
I appreciate any help for solving this issue.
2 answers
- answered 2018-01-14 11:10 anuj saxena
Move the operation you are performing outside the calling class to a serialized class or to an object:
class RDDOperation extends Serializable { def perform(inputFile: RDD[String], numberOfColumns: Int) = { for (dim <- 0 to (numberOfColumns - 2)) yield { inputFile.map(line => { val lines = line.split(',') (lines(dim), lines(numberOfColumns - 1)) }) } } }
The reason for the exception is that RDD's elements are partitioned across the nodes of the cluster. So when we use map/flatMap on an RDD all the operation happen on multiple nodes so the operation being performed inside map must be serialized. Hence moving it to a serialized class or an object will make it serialized.
And prefer returning values in scala functions hence I have used yield here which will return a collection of pairRDDs.
Also, you can refactor the perform method which doesn't depend on numberOfColumns like this:
def perform(inputFile: RDD[String]): RDD[(String, String)] = { inputFile.flatMap{ line => val lines = line.split(',').toList lines.reverse match { case lastColumn :: _ => lines.flatMap{ case column if column != lastColumn => Some((column, lastColumn)) case _ => None } case _ => List.empty[(String, String)] } } }
- answered 2018-01-14 11:10 Hoori M.
In my code, I had references to the global fields of the class. So Spark had to send the whole class instance to the executors to access those fields and my class was not serializable.
I copied all the global fields as local variables in my method so just the local variables were sent to the executors and the problem got solved. | http://codegur.com/48248964/mapping-an-input-file-to-different-rdds | CC-MAIN-2018-05 | refinedweb | 384 | 65.32 |
@geekie/navigator
We developed this navigator library at Geekie when developing our first React Native app and getting upset by the transitions of the (now deprecated) Navigator library that shipped with React Native. After evaluating some of the options, we decided that none had the APIs and transitions that would suit our needs.
Features
- Screen transitions at 60 fps (uses the
AnimatedAPI with the native driver option)
- Built-in support for Android's back button
What's missing
- No headers (we use a custom header in the app, so we didn't need it)
- Different transitions based on platform (animations are built-in and can't be changed)
- Tab based navigation
Install
This is a pure-JS library so there's no need to link/
yarn add @geekie/navigator
Usage
The navigation with this library is built on the concept of "stacks", which are a stack of screens of the app. You can present a new stack and push screens onto it when dismissing it, the navigation state returns to the last screen pushed to the previous stack. Below there's an image as an example:
The blue arrows are pushing new screens to a stack, the green arrow is presenting a new stack with a screen on it, and the red arrow is dismissing a whole stack and coming to back to the state right before presenting the new stack.
With that clear we can check what the API looks like.
Navigator
The library has a single default export which is the
Navigator component that manages the current rendered screen. It accepts the following props:
screensConfig: (required) an object like
{ "ScreenName": ScreenComponent }. The key will be used to reference the screen in the navigator commands, and the navigator will use the component defined when rendering the screen.
initialState: (required) the initial state of the navigator: an array of arrays of routes. A route is defined as an object with the keys:
screen(the name defined in
screensConfig) and
props(optional), an object that will be passed to the rendered component.
screenStyle: custom style to be applied the View wrapping the rendered screen.
resetState: a function that will be called when the navigator is rendered, with another function that can reset the whole state programatically. See an example below.
onWillFocus: a function that will be called right before a screen is focused.
The example below shows how you use those props and an example of how to implement deep linking using the
resetState function.
import Navigator from "@geekie/navigator"; import Home from "./screens/Home"; import Login from "./screens/Login"; import Profile from "./screens/Profile"; import About from "./screens/About"; import { getCurrentUser } from "./user"; const screensConfig = { Home: Home, Login: Login, Profile: Profile, About: About }; class App extends React.Component { componentDidMount() { // Deep linking support Linking.addEventListener("url", event => { if (event.url.endsWith("/about")) { this.resetState([ [{ screen: "Home" }], [{ screen: "Profile", props: { user: getCurrentUser() }, {screen: "About"} }] ]); } }); } render() { return ( <Navigator screensConfig={screensConfig} initialState={{ screen: "Home" }} onWillFocus={route => { console.log("Navigator will now focus the screen:", route); }} resetState={resetState => { this.resetState = resetState; }} /> ); } }
The components rendered will receive a
navigator prop that contains the commands to navigate through screens.
NavigatorProvider
This is useful if you're rendering screens wrapped with
withNavigator or that use
useNavigator but doesn't want to render the whole navigator, or would like to pass a mocked navigator prop to check calls.
const test = render( <NavigatorProvider navigator={mockedNavigator}> <ScreenComponent /> </NavigatorProvider> );
withNavigator(Component)
withNavigator is a higher order component that injects the
navigator prop to the wrapped
Component. This is useful if you need it in a deeply nested component or you can't pass the prop from the screen component.
useNavigator()
This is a React Hook that returns the
navigator provided with
NavigatorProvider. This is a substitute to
withNavigator().
navigator.present(route) or
navigator.present(routes)
Presents a new stack: if the argument is an array, will use that as the stack; if not an array, it will be equivalent as calling
navigator.present([route]).
The screen will be slide from the bottom.
navigator.dismiss()
Removes the whole current stack and renders the last route of the previous stack.
The screen will slide to the bottom.
navigator.push(route)
Pushes a new route (
{screen: "ScreenName", props: { }}) to the current stack.
The screen will slide from the right.
navigator.pop()
Returns to the previous route of the current stack. If there's a single route in the stack, the stack is dismissed (i.e. the same as calling
navigator.dismiss()).
The screen will slide to the right.
navigator.replace(route)
Replaces the current route but animates as
push(). This is useful if you want to navigate to a new screen, but doesn't want the next screen to navigate "back" to it.
Example:
Home -> push ->
Screen A -> replace ->
Screen B -> pop ->
navigator.reset(route)
Resets the current stack to
[route].
The animation is the same as
pop().
navigator.pushReset(route)
The same as
navigator.reset() but the animation will be the same as
push(). | https://reactnativeexample.com/a-minimal-navigator-library-for-react-native/ | CC-MAIN-2021-17 | refinedweb | 830 | 55.74 |
The task every developer fears "Hey, I sometimes get a timeout when I want to see the last gazillion days of activity in the dashboard. What gives?"
This problem started happening for a project I work on a few months ago. Through a lucky sequence of events I was able to push the problem back a couple of months ... mostly just a beefier VPS on amazon with a bit more memory.
Yesterday I finally cleaned up everything else from pivotal tracker and was left with only that one dreaded item. That ... slowness thing. Have you ever tried speeding up the execution of something when 80% of the codebase isn't yours?
Context! I need context!
Let me give you a bit of context for what's going on:
- the project lets customers have many users
- users enter events when they do stuff
- the entries can be checked in a dashboard
- there is a date-range picker to see more than just today's events
And that's where it all gets ever so slightly weird. When you choose anything other than today things start becoming increasingly slow and sometimes certain browsers decide that it's just too slow and timeout.
No, it isn't slow enough to cause the 30 second socket timeout.
No, caching wouldn't really benefit this activity either.
A simple benchmark
In light of what I've learned about unit testing being for lazy people last week, I decided to set up a unit test that is actually an integration test and most likely isn't either of those since it's some sort of benchmark.
But it goes in django's testing framework, the python module behind that is called unittest, so I'm calling it a unit test.
def test_date_range(self):t1 = datetime.now()response = self.c.post('/accounts/dashboard/',{'dateFrom': (datetime.now()-timedelta(days=356)).strftime("%m/%d/%Y")})t2 = datetime.now()self.assert_((t2-t1).seconds < 2, "Too slow")
See, simple test. All it needed was exporting the whole production database into a fixture so django loads it up every time the test is run and then cleans up after itself.
The results are abominable! To return a dashboard with 250 rows of entries, this django app takes between 5 and 8 seconds. Clearly way way too much time!
The fix?
Indexes! Yes, adding an index where appropriate should speed things up immensely. Right?
Nope.
Creating indexes didn't even put a dint in that benchmark. There was more fluctuation from the effect of memory pages on mysql's performance. And anyway ... what was I thinking really, indexes helping queries with less than 1000 rows of results? Don't be silly Swizec.
Now what?
Added a bunch of "spent time" prints in the code and realized that even though it takes less than a second to make everything ready it then takes five seconds to render the template.
Wait what! Five seconds to render the template!?
Oh right ... django's querysets are lazy. That would probably explain it. The template is where those querysets get evaluated, they ping the database and that's the explanation I'm looking for. It must be!
Except it wasn't. Due to the way the view is written all the querysets get iterated over before the template is rendered. And they all have that magical select_related so all the fields getting accessed inside the template can't possibly be triggering more db queries can it?
Well ... no.
The main module has a function added that returns the latest related object. This function, correctly, only gets called once per rendering for every model.
But the function returns a queryset.
That queryset is lazy and it gets evaluated every single time that related object is used, which turns out to be 30 times per displayed row. But can 30 queries explain the slowness in rendering? Or should it be even slower if my theory was in fact correct?
I don't know, but right now I'm working on a fix to make sure that object is solidified and is not a lazy queryset before it gets pushed to the template. If it works, great ... otherwise I have no idea what I'll try next.
user.get_profile() also gets called awfully often. Wonder how that baby works ...
Related articles
- Unit testing is for lazy people (swizec.com)
- Simple trick for testing forms full of checkboxes with django (swizec.com)
- "architect": we have big refactors to do on out architecture but first we need to unit test everything me: I'm all for finally (devsigh.com)
- Django charfield queryset filter without escaping the MYSQL wildcard (stackoverflow.com)
- Do we still need django-nonrel now that GAE (allegedly) supports Django out of the box? (stackoverflow.com)
- Retrieving only the date part of a datetime column in Django (stackoverflow.com)
- 15 Key Resources to Learn Django (css.dzone️ | https://swizec.com/blog/how-to-make-your-django-app-slow/ | CC-MAIN-2021-43 | refinedweb | 817 | 73.27 |
Building Mobile Apps With Angular or Vue.js and NativeScript {N}
Building Mobile Apps With Angular or Vue.js and NativeScript {N}
In this tutorial, we go over to quickly make mobile applications using NativeScript plus Angular and NativeScript plus Vue.js.
Join the DZone community and get the full member experience.Join For Free
More and more Java developers have been coming into contact with front-end development in recent years. For this group of developers, it is only a small step to switch to mobile app development with NativeScript. In this post, we try to clarify the possibilities of native and hybrid app development using NativeScript.
What Is NativeScript {N}?
{N} is an open source framework (under the Apache 2 license) to build native iOS and Android apps, using TypeScript and Angular. {N} is a different technology than the hybrid frameworks, such as Ionic and Phonegap. {N} is a runtime, not a web technology. Your app will not run as a mini website in a WebView and is therefore more efficient. With {N} you have direct access to all the Native APIs of your device.
TypeScript
For Java programmers, it is interesting to realize that TypeScript (TS) has many similarities to Java. TS is the best way to write {N} apps, whether combined with Angular or not. And, further, writing native code in TS for Android is quite simple, because you can take on a lot of one-to-one in the native Java code (for example:
new io.java.File() is valid in both Java and JS/TS. That is just a bit trickier in Objective-C). It is actually very fascinating to see how NativeScript manages to implement all the native constructions with pure JavaScript. From string arrays to interfaces and implementations of abstract classes.
Why NativeScript?
One of the arguments for using NativeScript {N} is the reuse of 'Skills.' That is, someone with knowledge of JavaScript/TypeScript can immediately start with {N}. {N} is written in JavaScript or TypeScript.
Reuse of code. Write Native Mobile apps for iOS and Android with a single code base and a common interface. This is not the case with Xamarin, for example, where you have to build two interfaces for iOS and Android with one common layer. It is not surprising that the slogan of {N} is: 'Write once, run everywhere.'
It is also easy to expand with the help of {N} modules (see the examples in this article) and npm modules. Actually the plug-ins from Cordova/Phonegap.
Finally, no WebViews are used, unlike with many hybrid frameworks. With JavaScript, you have direct access to the Native APIs and making NativeScript more performant.
What Is Needed for NativeScript?
In the NativeScript docs, there is a clear description of how {N} should be installed. Because this installation does not always work flawlessly, it is wise to install 'NativeScript Sidekick.'
With Sidekick, it is possible to build apps in the cloud.
If you choose to build your apps in the cloud, you can develop your apps independently of your operating system. You can, of course, even build iOS apps on a Windows machine. It is also possible to work locally with Sidekick, but for that you have to set up your own environment with iOS Xcode and the Android SDK. Sidekick still has many known issues but is already working well.
Another nice thing about {N} is Live Sync. So if you modify a rule in the code, it will be pushed directly on the phone with the cloud build and the result of the change is immediately visible.
How Does NativeScript Work With JavaScript?
The following is a simple example of JavaScript in {N}, which instantiates an Objective-C based iOS
UIAlertView control:
var myAlert = new UIAlertView(); myAlert.message = “NativeScript rocks!”; myAlert.show();
Because web developers do not want to learn iOS and Android specific APIs, {N} offers a set of {N} modules. These modules abstract the iOS and Android details in simple JavaScript APIs.
The above
UIAlertView-based code can be rewritten with the {N} 'Dialog module':
var dialogs = require ("ui / dialogs"); dialogs.alert ({message: "NativeScript rocks!"});
This
dialogs.alert() call also provides us with the
android.app.AlertDialog for your Android app.
Although this 'dialog' example is simple, the same technique can be used for building robust apps. For this, you can use the already existing, and mature, native iOS and Android UI components.
What Can Angular Add to NativeScript?
{N} can now also be written in Angular.
If you have knowledge of Angular, it is a small step to use Angular in {N}.
The big difference with Angular is that the browser-based HTML elements such as <div> and <span> are not available in {N}. You should use {N} UI components instead.
No DOM or browser is used in {N}. {N} UIs are native UIs and therefore disconnected from the DOM. Because Angular is an agnostic framework and is disconnected from the DOM, this framework can easily be integrated with {N}. The example below deals with this. AngularJS, unlike Angular, is not suitable for {N} because this framework is linked to the DOM.
By using Angular in {N}, you have the ability to share code between your existing web application and your Native apps. Let's look at an example.
Reuse of Code Between Web and Mobile Apps
As an example, let's show a 'Grocery List' in a web application.
import {Component} from ‘@angular/core'; @Component({ selector: ‘grocery - list ' templateUrl: ‘grocerylist.template.html ' }) export class GroceryListComponent { groceries: string[]; constructor() { this.groceries = [‘Cheese, ‘Eggs’, 'Tomatoes', 'Tea'] } }
Listing 1
In Listing 1, an Angular Component is defined, which fills an array of 'groceries,' via the constructor.
TypeScript is used. This is a 'typed superset' of JavaScript and is the standard for writing Angular applications.
How Does This Cmponent Work?
Via the selector associated with the Component, you can ask Angular to instantiate and render this component, where it finds a <grocery-list> tag in the HTML (see Listing 2).
…. <grocery-list></grocery-list> ….
Listing 2
In order to render the 'Grocery list,' a template has to be defined with the name: grocerylist.template.html (see Listing 3). This template is also defined under the
templateUrl of the Component.
<p>Groceries: </p> <ul> <li *ng-for = “let grocery of groceries"> {{grocery}} </li> </ul>
Listing 3
The list of 'groceries' is iterated, using Angular's
ng-for directive.
The question now is: what else do we have to program to get the above code working for your NativeScript iOS and Android apps?
The comprehensive answer is:
The component in Listing 1 remains the same. This TypeScript code can therefore be reused between the web application and mobile apps.
The only thing you need to change is the template for this component!
You must use {N} UI components instead of HTML elements in your template (see Listing 4). What is striking is that you can still use Angular's
ng-for directive for iterating through the list.
And converting a template to {N} UI components is a simple exercise, because there is a Native UI component for every HTML element. This example also shows that Angular is disconnected from the DOM, and can therefore deal with {N} UI components.
<StackLayout * <Label [text] = “grocery"></Label> </StackLayout>
Listing 4
HTTP Module
Another example for code reuse is the following.
{N} provides support for web APIs, such as
XMLHttpRequest and
fetch().
You can use the HTTP module within {N}. This module makes it possible to send web requests and receive web responses.
var http = require(“http”); import HTTP module http.getJSON(‘’) .then(function (result)) { … });
Listing 5
In the diagram below, you see on the left that {N} translates the XMLHttpRequests Web API into native code for iOS and Android. Also shown, on the right side of this diagram, is the implementation of the XMLHttpRequest that is used for the web application. This is a different implementation than the NativeScript variant.
The problem here is that we can not reuse the XMLHttpRequest implementations between the web application and mobile app. There is a ''missing link'' here, namely a generic HTTP module for the web application and the mobile app.
This can be solved with Angular. Angular adds an extra abstraction layer for handling HTTP requests/responses. This is possible with the HTTP module from Angular.
We can reuse this module between the web application and the mobile app.
For example, the code to be reused will look like this in Angular:
import {Http} from ‘@angular/http’ export class GroceryListService { constructor(private http: Http) {} getGroceries() { return this.http.get(``) .map((res: Response) => res.json()); } }
Listing 6
More Tools and Features
Outside the above mentioned examples, we also have other features, such as the use of plug-ins, scaffolding, and the {N} Playground. Furthermore, you can now build a {N} app with the Vue.js framework.
Plug-ins
The interesting thing about {N} is that there are many plug-ins available in the marketplace.
You do not have to build everything yourself. Think of Facebook, fingerprint/face authentication, 'text to speech,' ads with Admob, or in-app purchases. If the plug-in is not yet available, it can always be built by you.
NativeScript Playground
{N} Playground is a tool that makes it easy to build an app using a web interface within a few minutes. See.
In addition to the website, two apps must be installed on a device:
- NativeScript Playground
- NativeScript Preview
After scanning the QR code of the website using the playground app, the app will be installed on the device. With Live Sync, all changes will be immediately visible in the app. You can also drag and drop Components (Buttons, Sliders, Charts) in your template. Super cool.
Try it out. With Playground you can easily share your {N} code and issues with other developers. It is a means to quickly get started with {N} and experience its benefits. You can also build an app with NativeScript and Vue.js.
NativeScript and Vue.js
Like Angular, Vue.js is also component based. It focusses on the view layer. Vue.js integrates easily with Angular or other front-end frameworks. And, finally, is very small and compact (21 Kb).
I'll show you a simple example of how a component looks in Vue.js and NativeScript.
First, a Vue instance is created with
new Vue(). This instance contains the data or model you want to render. In this case, we want to render the message: 'Hello Vue!'
And the second step is to define a template to show the message.
const Vue = require("nativescript-vue"); new Vue({ data() { return { message: 'Hello Vue!', } }, template: ` <Page class="page"> <Label : </Page> `, }).$start();
This component will render the following on your mobile device:
The <Page> element is the top-level/main UI element of every NativeScript+Vue.js app. All other user interface elements are nested within.
All the code in one file!
You can experiment with Vue.js and NativeScript in the NativeScript Playground.
Build Both Web and Mobile Apps From a Single Project
Reusing code is a very important challenge for development in general.
At this moment, you can generate a web application with the Angular CLI. And for {N} you use the NativeScript CLI. Before Angular 6.1, this was a problem because the CLIs give us no possibility to create one project for the web application and the native mobile application. Of course, it is possible to maintain two separate projects and to copy-paste the shared files between the two projects. This can also be solved by using the following 'seed project.' For example:.
But with @angular/cli 6.1.0 it is now possible to build both web and mobile apps from a single project. To realize the code-sharing dream, the Angular and NativeScript teams teamed up to create nativescript-schematics. This is a schematic collection or repository for generating components in NativeScript+Angular apps using the Angular CLI. To generate a new NativeScript+Angular project, you can use
ng newwith @nativescript/schemes specified as the schema collection.
Note, you may need to install @nativescript/schematics first:
npm install --global @nativescript/schematics
For starting a new web and mobile code sharing project:
ng new --collection=@nativescript/schematics my-shared-app --shared
Conclusion
In summary, {N} is a powerful framework for building native cross-platform mobile applications with Angular, Vue.js, TypeScript, or JavaScript. You have direct access to the Native platform APIs.
With NativeScript Playground you can build an app within a few minutes. And with NativeScript SideKick you can build robust apps in the cloud.
{N} works differently from hybrid frameworks such as Phonegap, but it is extremely suitable for web developers. It is more performant and easy to learn.
Opinions expressed by DZone contributors are their own.
{{ parent.title || parent.header.title}}
{{ parent.tldr }}
{{ parent.linkDescription }}{{ parent.urlSource.name }} | https://dzone.com/articles/building-mobile-apps-with-angular-and-nativescript | CC-MAIN-2019-51 | refinedweb | 2,150 | 57.87 |
Welcome to part 2 of our 3 part series on user authentication with Titanium Mobile. Titanium is an open source cross compiler that allows you to write iPhone and Android (soon to be blackberry too!) applications using Javascript. No Objective-C required! We will be using PHP as the server side code and my database will be MySQL. For this example, I am using MAMP to develop locally. I strongly recommend that you go through the first part of this series before continuing if you haven't already. However, you can alternatively download the source from part 1, create the database table, and setup the PHP database connections on your own before skipping to this tutorial if you would like.
Synopsis
In part 1, we setup the database for our app and added a user. We then made our login interface by creating a tab group, a tab, and a window. We then gave our login button some actions. Our PHP file queried our database and, upon successful login, it returned our name and email. If login authentication failed, we returned a string simply stating invalid username and/or password. In part 2, we will create a new tab on the main screen that allows a user to create a new account and then login.
Step 1: Creating the Account Window and Tab
Open up app.js and create the account window and tab underneath our login tab script. Also note that I removed the tabBarHidden property on the login window that we did in part 1. Removing that property allows us to see the tabs on the bottom of the phone. We have also added the accountTab to the tabGroup.
Titanium.UI.setBackgroundColor('#fff'); var tabGroup = Titanium.UI.createTabGroup();();
The URL property on the account var tells the compiler to use account.js as our window. If you skip this part, Titanium will throw an ugly red error in the emulator. Upon a successful compile, your screen should look like this:
Traditionally, you would see the tab bar on the bottom have icons. Well, with Titanium, that is super easy too! Simply use the icon property in each tab. For example:
var accountTab = Titanium.UI.createTab({ title:'New Account', icon:'images/account_icon.png', window:account });
Step 2: Create account.js
Create a new file and name it account.js and save it in your Resources/main_windows folder. This is the same place we saved our login.js file in part 1.
var win = Titanium.UI.currentWindow; /* * Interface */ var scrollView = Titanium.UI.createScrollView({ contentWidth:'auto', contentHeight:'auto', top:0, showVerticalScrollIndicator:true, showHorizontalScrollIndicator:false }); win.add(scrollView); var username = Titanium.UI.createTextField({ color:'#336699', top:10, left:10, width:300, height:40, hintText:'Username', keyboardType:Titanium.UI.KEYBOARD_DEFAULT, returnKeyType:Titanium.UI.RETURNKEY_DEFAULT, borderStyle:Titanium.UI.INPUT_BORDERSTYLE_ROUNDED }); scrollView.add(username); var password1 = }); scrollView.add(password1); var password2 = Titanium.UI.createTextField({ color:'#336699', top:110, left:10, width:300, height:40, hintText:'Password Again', passwordMask:true, keyboardType:Titanium.UI.KEYBOARD_DEFAULT, returnKeyType:Titanium.UI.RETURNKEY_DEFAULT, borderStyle:Titanium.UI.INPUT_BORDERSTYLE_ROUNDED }); scrollView.add(password2); var names = Titanium.UI.createTextField({ color:'#336699', top:160, left:10, width:300, height:40, hintText:'Name', keyboardType:Titanium.UI.KEYBOARD_DEFAULT, returnKeyType:Titanium.UI.RETURNKEY_DEFAULT, borderStyle:Titanium.UI.INPUT_BORDERSTYLE_ROUNDED }); scrollView.add(names); var email = Titanium.UI.createTextField({ color:'#336699', top:210, left:10, width:300, height:40, hintText:'email', keyboardType:Titanium.UI.KEYBOARD_DEFAULT, returnKeyType:Titanium.UI.RETURNKEY_DEFAULT, borderStyle:Titanium.UI.INPUT_BORDERSTYLE_ROUNDED }); scrollView.add(email); var createBtn = Titanium.UI.createButton({ title:'Create Account', top:260, width:130, height:35, borderRadius:1, font:{fontFamily:'Arial',fontWeight:'bold',fontSize:14} }); scrollView.add(createBtn);
Okay, this mean looking block of code is really super easy to understand, yet it does so much for us. Just by looking at our variable names, it is pretty easy to decipher what is going on here. We created 5 fields:
- Username
- Password1
- Password2
- Name
We also made our 'create account' button.
You will also notice the var at the top called scrollView. Adding the objects to a scroll view allows the view to be scrollable so when the keyboard slides up, it isn't overlapping our text fields.
Go ahead and compile. Upon a successful compile, your screen should look like this after switching to the account tab. The create account button doesn't do anything yet, but play around with selecting text fields and see how the scroll view works.
Step 3: Adding the Click Event to Our Button
We now need to create an event listener on our button so when they click 'create account', it sends the data off as well as some validation.
var testresults; function checkemail(emailAddress) { var str = emailAddress; var filter = /^([A-Za-z0-9_\-\.])+\@([A-Za-z0-9_\-\.])+\.([A-Za-z]{2,4})$/; if (filter.test(str)) { testresults = true; } else { testresults = false; } return (testresults); }; var createReq = Titanium.Network.createHTTPClient(); createReq.onload = function() { if (this.responseText == "Insert failed" || this.responseText == "That username or email already exists") { createBtn.enabled = true; createBtn.opacity = 1; alert(this.responseText); } else { var alertDialog = Titanium.UI.createAlertDialog({ title: 'Alert', message: this.responseText, buttonNames: ['OK'] }); alertDialog.show(); alertDialog.addEventListener('click',function(e) { win.tabGroup.setActiveTab(0); }); } }; { alert("Everything looks good so send the data"); } } } else { alert("All fields are required"); } });
Starting from the top, the checkEmail() method is a simple function that uses Regular Expression to check if the email the user inputs is the correct format. We created a new HTTPClient that will be used to send our data to our PHP file.
In the click event, we first check if any fields are empty. If they are, alert them saying "All fields are required." Our next check is to see if the two password fields are the same. If they aren't, alert them saying "Your passwords do not match." Our final check is to check if the email address is valid. If it isn't, alert them saying "Please enter a valid email."
Once the form is validated, it will alert "Everything looks good so send the data." Go ahead and compile and test submitting the form with no values, non-matching passwords and an invalid email. Upon submitting a valid form, you will see the alert below:
Step 4: Actually Send Some Data
Go ahead and delete the "Everything looks good so send the data" line. We need to replace that with the open() and send() methods. { createBtn.enabled = false; createBtn.opacity = 0.3; createReq.open("POST",""); var params = { username: username.value, password: Ti.Utils.md5HexDigest(password1.value), names: names.value, email: email.value }; createReq.send(params); } } } else { alert("All fields are required"); } });
So, in replacing that line, we disable our 'create account' button and set the opacity to 30%. We then take the HTTPClient we made and call the open() method on it. It is pointing at a PHP file that we will make in the next step. We then create a params object to contain all the form data. Notice I am running an MD5 encryption on the password field. Final step is to call the send() method and pass it our params object.
Step 5: Creating our Register PHP File
This file will be the PHP file our app talks to when hitting the 'create account' button. The name must reflect the name in our createReq.open() method in the previous step. I've named mine post_register.php. Replace my mysql_connect and mysql_select_db settings with your connection settings.
<?php $con = mysql_connect('localhost','root','root'); if (!$con) { echo "Failed to make connection."; exit; } $db = mysql_select_db('test'); if (!$db) { echo "Failed to select db."; exit; } $username = $_POST['username']; $password = $_POST['password']; $names = $_POST['names']; $email = $_POST['email']; $sql = "SELECT username,email FROM users WHERE username = '" . $username . "' OR email = '" . $email . "'"; $query = mysql_query($sql); if (mysql_num_rows($query) > 0) { echo "That username or email already exists"; } else { $insert = "INSERT INTO users (username,password,name,email) VALUES ('" . $username . "','" . $password . "','" . $names . "','" . $email . "')"; $query = mysql_query($insert); if ($query) { echo "Thanks for registering. You may now login."; } else { echo "Insert failed"; } } ?>
So here we connect to our database and select the database named 'test' (that name will change depending on the name of your database obviously). You can see our $_POST variables reflect the names we set in the params object in our last step. That part is crucial. We then see if the username or email address they entered already exists. If it doesn't, insert the the data into the database. Okay, don't compile just yet! We will next step, I promise.
Step 6: Receiving Data in account.js
Okay back to account.js. Let do some data handling for when our PHP returns something. Place this code under var createReq and above our click event:
createReq.onload = function() { if (this.responseText == "Insert failed" || this.responseText == "That username or email already exists") { win.remove(overlay); alert(this.responseText); } else { var alertDialog = Titanium.UI.createAlertDialog({ title: 'Alert', message: this.responseText, buttonNames: ['OK'] }); alertDialog.show(); alertDialog.addEventListener('click',function(e) { win.tabGroup.setActiveTab(0); }); } };
So when PHP returns something, if 'this.responseText' is equal to "Insert failed" OR "That username or email already exists," remove the overlay window (so they can re-enter information and submit) and alert them with 'this.responseText'.
Upon successful registration, alert them with "Thanks for registering. You may now login" (defined in our post_register.php file). We also add an event listener to the OK button so when clicking it, it automatically takes us to the login screen.
If the alert coming back is some garbled message about mysql_connect and/or access denied, then you need to check your mysql connection settings in the PHP.
Conclusion
In part 2 of this series, we added in tabbed windows that you can switch between. We then made a new form where a user can input data into and submit it. Upon submitting we did some form validation and then had our PHP return a message based on if the data was in use and, if not, we successfully inserted it. I hope you enjoyed reading this mini series tutorial as much as I enjoyed writing it!
Envato Tuts+ tutorials are translated into other languages by our community members—you can be involved too!Translate this post
| https://code.tutsplus.com/tutorials/titanium-user-authentication-part-2--mobile-3742 | CC-MAIN-2020-29 | refinedweb | 1,697 | 51.85 |
React 17 is out and it has been 2.5 years since the release of React 16, React 16.x included many new changes like Hooks,Context etc.,But the new React 17 has no new features but it's a right move they have done this time before going to that let's see some minor changes which has happened
- Changes to Event Delegation : React will no longer attach event handlers at the document level. Instead, it will attach them to the root DOM container into which your React tree is rendered
Why this is important ?
- It will make it easier to use React with other JS frameworks.
- It is safer to use a React tree managed by one version inside a tree managed by a different React version.
- No React import : If you were using Create React App boiler plate(cra) or npm to download react and you would be importing React in every jsx or js code ,now it is no longer required
import React from "react";
onScroll Bubbling Event : In React previous version there was a lesser known bug , the parent element used to capture the scroll event of the children this was causing an issue in using the scroll event listener, this has been fixed now
No Event Pooling : For those who don't know what Event Pooling is: The event handlers that we have in any react-app are actually passed instances of SyntheticEvent(A Wrapper for native browser events so that they have consistent properties across different browsers).
Whenever an event is triggered, it takes an instance from the pool and populates its properties and reuses it. When the event handler has finished running, all properties will be nullified and the synthetic event instance is released back into the pool.
This was build to actually increase the performance but It didn't improve the performance in the modern browsers and also it used to confuse the developers so they decided to remove it.
Effect Cleanup Timing :The useEffect hook in React 16 runs asynchronously but the cleanup which we used , like this used to run synchronously this introduced an issue , Example: If you called an Api or An Animation and before the action is complete if the component gets unmounted
Now it is no longer an issue as it runs asynchronously even if the component is unmounted , the cleanup will happen , resulting in better performance
useEffect(() => { event.subscribe() return function cleanup() { event.unsubscribe() } })
You might be confused are these not features ?,these are more of changes internally and bug fixes good to know ,but this release is a right move, now is the right time to migrate your old react projects from class to functions based on hooks or implement context , this will give enough time for both business and developers to catch up to the fast pace in which react was moving.
If you find any new feature , please put it in the comments
I am a Full Stack JS Developer,This is my first article in Dev, any suggestions or constructive feedback on the article are welcome
Posted on by:
Goutham JM
#Full Stack JS Developer
Hello @gouthamjm , thank you for the interesting article. | https://dev.to/gouthamjm/react-17-why-it-s-so-important-il6 | CC-MAIN-2020-40 | refinedweb | 533 | 55.51 |
This is a video of a home with Chinese Drywall. I took a video of the house to save investors 000 + by knowing what to look for in a home prior to purcha…
Video Rating: 4 / 5
This is a video of a home with Chinese Drywall. I took a video of the house to save investors 000 + by knowing what to look for in a home prior to purcha…
Video Rating: 4 / 5
what is that black stuff on the wires. was the house previously been
through a fire.
OMG! I am living in a rental home in Brandon Florida. Been smelling this
sulfur “dead” smell for 2 months. Before we moved in, the house had MAJOR
repair work inside (including a new roof). Took a light switch off and it
was all black. Told property management. They said get out. They didn’t
even want to come inspect the situation! I have 7 days to find another
place and move me and my 2 children.
Never knew that these types of properties existed, I will be careful before
buying anything built during the boom.
Dude, fasten your seat belt.
To learn more about the stories of the families dealing with this toxic
import disaster visit our videos at Victims of Chinese Drywall published by
chinesedrywallva. Families have had their health, homes and financial well
being decimated by this toxic import and we have been begging local, state
and government officials to do something to help protect future families
from becoming second generation victims of this toxic import! Nothing has
been done for 3 years! Thank you for speaking out!
Where do you conduct your courses on buying houses I am ready to learn from
someone like you
Ok, so dry wall screws are supposed to be black. There was no discoloration
on the dry wall screws.
This particular home was located in Lehigh Acres Florida. Unfortunately it
happens in many more states. Other training videos have over 37,000 views.
You can visit: Wealth.me to check out other videos
Just black all black. Thats the way it is lol
Great information thanks for sharing this.
Wow! Thank you for posting!!! Saw a lot of things in that house!
PS where is this? Thank you!
This home is in Lee County Florida
There are many good thing the Chinese make well, many they don’t. Dry wall
is one of them. Hinges and light fixture as well. You pay for what you get.
great video | http://www.imsdbe.com/drywall-stamped-china-chinese-drywall-what-to-look-for-before-investing-in-properties/ | CC-MAIN-2018-05 | refinedweb | 417 | 84.17 |
Export rosbag content issues (frames drop or can't open with ROS API)baptiste-mnh Jun 22, 2018 8:55 AM
Hi!
Since de 2.12.0 release of the SDK, their is a tool (rs-convert) which can export the PNG images and the PLYs.
But I got this issues : problems in rs_convert · Issue #1919 · IntelRealSense/librealsense · GitHub
"Even with small bag files (~150MB), it drops frames. I was unable to extract all frames from a bag with 427 color and 430 depth frames. The number of frames dropped varies in each execution instance."
So is there any way to get all the frames or may be another way (with ROS API) to get all of my images in my rosbag?
I also tried some exemple based on ROS API but I often got this error when it try to open my rosbag file I got this error :
[FATAL] [1529679163.359016032]: Character [ ] at element [32] is not valid in Graph Resource Name [/device_0/sensor_0/option/Enable Auto Exposure/value]. Valid characters are a-z, A-Z, 0-9, / and _.
Thanks in advance for your help!
1. Re: Export rosbag content issues (frames drop or can't open with ROS API)3DScanProduction Jun 23, 2018 1:42 PM (in response to baptiste-mnh)
No answer ? It looks to me like a bugg in the sdk. And it is a very blocking issue.
2. Re: Export rosbag content issues (frames drop or can't open with ROS API)Jun 25, 2018 8:25 AM (in response to 3DScanProduction)1 of 1 people found this helpful
Hello bapriste-mnh,
Thank you for your interest in the Intel RealSense Cameras.
We will report this issue to the RealSense SDK team.
We apologize for the inconvenience this may have caused.
Thank you in advance,
Eliza
3. Re: Export rosbag content issues (frames drop or can't open with ROS API)Jun 28, 2018 9:31 AM (in response to Intel Corporation)
Hello baptiste-mnh,
The RealSense team is aware of this bug and is working on it.
Unfortunately, we don't have a timeline yet for when it will be fixed. Please give us more details (how you created the ROS bag, which ROS example you are running, etc.) on how you got the second error when you tried to open the ROS bag so that we can reproduce your issue.
Thank you,
Eliza
4. Re: Export rosbag content issues (frames drop or can't open with ROS API)baptiste-mnh Jul 10, 2018 3:08 AM (in response to Intel Corporation)
Hi, I simply use the realsense viewer to create my rosbag and rs-convert to extract the content.
I also tried to export with this code :
import argparse import pyrealsense2 as rs import numpy as np import cv2 import os LOAD_BAG = True SAVE_DEPTH = False SAVE_RGB = False SAVE_IRL = True SAVE_IRR = False def main(): if not os.path.exists(args.directory): os.mkdir(args.directory) if SAVE_DEPTH and not os.path.exists(args.directory+"/depth"): os.mkdir(args.directory+"/depth") if SAVE_RGB and not os.path.exists(args.directory+"/image"): os.mkdir(args.directory+"/image") if SAVE_IRL and not os.path.exists(args.directory+"/irl"): os.mkdir(args.directory+"/irl") if SAVE_IRR and not os.path.exists(args.directory+"/irr"): os.mkdir(args.directory+"/irr") try: config = rs.config() pipeline = rs.pipeline() if LOAD_BAG: rs.config.enable_device_from_file(config, args.input, False) if SAVE_DEPTH: config.enable_stream(rs.stream.depth, 1280, 720, rs.format.z16, 30) if SAVE_RGB: config.enable_stream(rs.stream.color, 1280, 720, rs.format.rgb8, 30) if SAVE_IRL: config.enable_stream(rs.stream.infrared, 1, 1280, 720, rs.format.y8, 30) if SAVE_IRR: config.enable_stream(rs.stream.infrared, 2, 1280, 720, rs.format.y8, 30) pipeline.start(config) i = 0 while True: print("Saving frame:", i) frames = pipeline.wait_for_frames() if SAVE_DEPTH: depth_frame = frames.get_depth_frame() depth_image = np.asanyarray(depth_frame.get_data()) depth_image[np.where((depth_image > 3000))] = 0 cv2.imwrite(args.directory + "/depth/" + str(i).zfill(6) + ".png", depth_image) if SAVE_RGB: color_frame = frames.get_color_frame() color_image = np.asanyarray(color_frame.get_data()) cv2.imwrite(args.directory + "/image/" + str(i).zfill(6) + ".png", color_image) if SAVE_IRL: irl_frame = frames.get_infrared_frame(1) irl_image = np.asanyarray(irl_frame.get_data()) cv2.imwrite(args.directory + "/irl/" + str(i).zfill(6) + ".png", irl_image) if SAVE_IRR: irr_frame = frames.get_infrared_frame(2) irr_image = np.asanyarray(irr_frame.get_data()) cv2.imwrite(args.directory + "/irr/" + str(i).zfill(6) + ".png", irr_image) i += 1 except RuntimeError: print("No more frames arrived, reached end of BAG file!") finally: pass if __name__ == "__main__": parser = argparse.ArgumentParser() parser.add_argument("-d", "--directory", type=str, help="Path to save the images") parser.add_argument("-i", "--input", type=str, help="Bag file to read") args = parser.parse_args() main()
5. Re: Export rosbag content issues (frames drop or can't open with ROS API)Jul 13, 2018 1:25 PM (in response to baptiste-mnh)
Hello @baptiste-mnh,
I see that you have responded in this thread:. It is not clear whether you have resolved your issue. There is an open bug with the RealSense team where the rs-convert tool drops frames from ros bags.
Regards,
Jesus G.
Intel Customer Support
6. Re: Export rosbag content issues (frames drop or can't open with ROS API)Leponzo Jul 13, 2018 1:47 PM (in response to baptiste-mnh)
MATLAB worked flawlessly for me: Re: Converting .bag video to .raw frames | https://communities.intel.com/thread/126676 | CC-MAIN-2018-39 | refinedweb | 889 | 60.72 |
array manipulation - Java Beginners
[] nums, int val) {
} Hi friend,
Read in detail with array example at: manipulation We'll say that a value is "everywhere
array example - Java Beginners
array example hi!!!!! can you help me with this question... dependents to Employee that is an array of Dependent objects, and instantiate a five-element array
* while this isn't the best practice, there isn't a much better
Java read binary file
Java read binary file I want Java read binary file example code that is easy to read and learn.
Thanks
Hi,
Please see the code at Reading binary file into byte array in Java.
Thanks
Hi,
There is many
Convert InputStream to Byte
Convert InputStream to
Byte
In this example we are going to convert input
stream to byte. Here we are going to read input stream
and converting
ARRAY SIZE!!! - Java Beginners
ARRAY SIZE!!! Hi,
My Question is to:
"Read integers from the keyboard until zero is read, storing them in input order in an array A. Then copy them to another array B doubling each integer.Then print B."
Which seems
How do I initialize a byte array in Java?
How do I initialize a byte array in Java? How do I initialize a byte array in Java
How to read a byte.
Description
In the given Example, you will see how to use read method of CheckedInputStrem class.
It reads single byte of data from the input stream and
updates the checksum from that byte data which it read. When data transmit
What is the byte range? - Java Beginners
What is the byte range? Hi,
Please tell me range in byte.
Thanks
The range is: 128 to 127
Thanks
how to create a zip by using byte array
how to create a zip by using byte array hi,
How to convert byte array to zip by using java program.can u plz provide it......
Thanks,
krishna
Convert a string representation of a hex dump to a byte array using Java?
Convert a string representation of a hex dump to a byte array using Java? Convert a string representation of a hex dump to a byte array using Java
Java Array - Java Beginners
Java Array Can someone help me to convert this to java?
I have an array called Projects{"School", "House", "Bar"}
I want all the possible combinations but only once each pair, for example ["House", "Bar"] is the SAME sorting - Java Beginners
Array sorting Hello All.
I need to sort one array based on the arrangement of another array.
I fetch two arrays from somewhere and they are related.
For example,
String type[] = {"X","T","3","R","9"};
String
ARRAY SIZE. - Java Beginners
ARRAY SIZE. Thanks, well that i know that we can use ArrayList... the elements in array A. Then doubled array A by multiplying it by 2 and then storing it in array B. But gives me error.
import java.io.*;
import
java image converting to byte codes - Java Beginners
java image converting to byte codes i want to convert an image to human unreadable format which is uploaded by client to server.How can i do
The byte Keyword
;
The byte Java Keyword defines the 8-bit integer primitive type.
The keyword byte in Java is a primitive type that designates with eight bit signed integer
in java primitive type. In java keyword byte will be stored as an integer
array - Java Beginners
array Write a program that read 15 integers between 0 and 6 and a method to search, displays and returns the number that has the highest occurrence. Prompt the user if the number is not within this range.
Hi
read xml using java
read xml using java <p>to read multiple attributes... consumption shall not exceed 230 Byte The Runtime of the software shall... consumption shall not exceed 120 Byte The ROM consumption shall not exceed 16000
Program to read a dynamic file content - Java Beginners
Program to read a dynamic file content Hi,
In a single....
For example: Folder name "Sample". Inside that "010209.txt... the database. Im using MySql Database and a standalone java program.
I
sorting an array of string with duplicate values - Java Beginners
sorting an array of string Example to sort array string
Read the File
the data is read in form
of a byte through the read( ) method. The read...
Read the File
As we have read about the BufferedInputStream
class
Java Read File Line by Line - Java Tutorial
Java example for Reading file into byte array...
Java example for Reading file into byte array... Java Read File Line by Line - Example code of
reading the text file
convert zip file to byte array
convert zip file to byte array Hi,How to convert zip file to byte array,can you please provide program??
Thanks,
Ramanuja
read a file
read a file read a file byte by byte
import java.io.File... {
FileInputStream fin = new FileInputStream(file);
byte fileContent[] = new byte[(int) file.length()];
fin.read(fileContent
Read from a window - Java Beginners
Read from a window HI,
I need to open a website and read the content from the site using Java script.
Please suggest. Thanks
array - Java Beginners
array WAP to perform a merge sort operation. Hi Friend,
Please visit the following link:
Hope that it will be helpful for you.
Thanks
java - Java Beginners
();
// Serialize to a byte array
ByteArrayOutputStream baos = new ByteArrayOutputStream...://
Thanks... in Filesystem that Object is to be Deserialize
would u give and example source code
Array - Java Beginners
Array how to declare array of hindi characters in java
Array in Java
] = 80;
int[4] = 90;
Example of array in Java:
import java.util.*;
public....
Different types of array used in Java are One-dimensional, Two-dimensional and multi..., or an array of Objects. For example:
int[][] i;
int[][] i = new int[3][4];
int
array sort - Java Beginners
array sort hi all,
can anybody tell me how to sort an array... array[], int len){
for (int i = 1; i < len; i++){
int j = i;
int tmp = array[i];
while ((j > 0) && (array[j-1] > tmp
read image
read image java code to read an image in the form of an array.
... image = ImageIO.read(input);
int array[]=getImagePixels(image);
for(int i=0;i<array.length;i++){
System.out.println(array[i
RandomAccessFile & FileInputStream - Java Beginners
://
Thanks...("abc.txt");
byte[] b = new byte[(int)f.length()];
RandomAccessFile raf = new
Example Code - Java Beginners
Example Code I want simple Scanner Class Example in Java and WrapperClass Example.
What is the Purpose of Wrapper Class and Scanner Class .
when i compile the Scanner Class Example the error occur :
Can not Resolve symbol a zip file.
zip_entry_read()
In this example the function zip_entry_read() will help... file and direct the zipread command to read it through while loop. The zip_entry_read command will read one byte at a time. when it reads all the contents
array - Java Beginners
array Accept a two dimensional array from the user check if this array is symetric display a message yes,if it is symetric otherwise display it ,is not symetric data again - Java Beginners
Read data again OK my problem some like my first question.First i...","root","suprax");
//Read File Line By Line
while ((strLine = br.readLine...");
//Read File Line By Line
while ((strLine = br.readLine()) != null
://
Hope...array in javascript how to initialize array in javascript and can we increase the size of array later on?
is Array class in javascript ?
also
example explanation - Java Beginners
example explanation can i have some explanation regarding...();
}
}
}
---------------------------------------------
Read for more information.
Read file in java
Read file in java Hi,
How to Read file in java?
Thanks
Hi,
Read complete tutorial with example at Read file in java.
Thanks
How to read loops in Java
How to read loops in Java Example that explains how to read loop in Java...) {
int array[]={1,2,1,1,3,4,4,3,6,8,0,6,0,3};
int num;
int count;
for(int i = 0... = 0; j < array.length; j++){
if(j >= i){
if(array[i
java - Java Beginners
Java array add What is array and how can i add an element into an array in Java? IT would be nice if you can give an example.Thanks
Java I/O Examples
Byte Streams Example
In this section we will discuss how to read one byte... the stream, byte stream and array of byte stream.
Classes...
provided by java. io package. This is a class that allows you to read
java array - Java Beginners
java array 1.) Consider the method headings:
void funcOne(int[] alpha, int size)
int funcSum(int x,int y)
void funcTwo(int[] alpha, int[] beta...];
int num;
Write Java statements that do the following:
a. Call
array 1 - Java Beginners
array 1 WAP to input values in 2 arrays and merge them to array M...;
for (int[] array : arr) {
arrSize += array.length;
}
int[] result = new int[arrSize];
int j = 0;
for (int[] array : arr
array - Java Beginners
array how to determine both the diagonals of a 2-d array?
How... the size of 2D array :");
int i=input.nextInt();
int d[][]=new int[i][i];
int j,k;
System.out.println("Enter the values of 2D array of "+i+" * "+i
Array - Java Beginners
using one-dimensional array that accept five input values from the keyboard...(System.in));
int array[] = {2, 5, -2, 6, -3, 8, 10, -7, -9, 4,15};
Arrays.sort(array);
int index = Arrays.binarySearch(array, 2 | http://www.roseindia.net/tutorialhelp/comment/80744 | CC-MAIN-2014-10 | refinedweb | 1,588 | 64.1 |
IMPORTANT Generic methods with generic params cannot be used in Source XML as in-rule methods or rule actions. Use plain .Net class(es) if your source object declares generic methods.
If applied to a source object, references an external qualified public method as an in-rule method. To qualify, the method must return a value type and be parameterless or have only parameters of source object type or any value type supported by Code Effects.
[AttributeUsage(AttributeTargets.Class | AttributeTargets.Interface,
AllowMultiple = true, Inherited = true)]
public class ExternalMethodAttribute : System.Attribute,
CodeEffects.Rule.Attributes.IExternalAttribute,
CodeEffects.Rule.Attributes.ISortable
Assembly Type: System.String
Gets or sets the full name of the assembly that declares the method class, which can be obtained by executing Assembly.GetAssembly( className ).FullName. If the Type property is not set, this property is required.
Method Type: System.String
Gets or sets the name of the in-rule method declared outside of the source object. This is a required property.
SortOrder
Type: System.Int16
Gets or sets the sort order in which the in-rule method is placed in fields menu on Rule Area. This property is optional. Default value is 0.
Type Type: System.Type
Gets or sets the type of the class that declares the external in-rule method. If the Assembly and TypeName properties are not set, this property is required.
TypeName Type: System.String
Gets or sets the full name of the type of the class that declares the external in-rule method, which can be obtained by executing typeof( className ).FullName. If the Type property is not set, this property is required.
This attribute is optional. Its purpose is to provide references to in-rule methods declared outside of the source object. An exception will be thrown if this attribute points to a non-qualified method. If the external method declares multiple overloads, decorate each overload that you would like to use as an in-rule method with the MethodAttribute and set its DisplayName property to a unique display name.
using System;
using CodeEffects.Rule.Attributes;
namespace TestLibrary
{
// The source object references an
// external method declared in the Helper class
[ExternalMethod(typeof(Helper), "GetFullName")]
public class Person
{
[Field(DisplayName = "First name", Max = 30)]
public string FirstName { get; set; }
[Field(DisplayName = "Last name", Max = 30)]
public string LastName { get; set; }
}
public class Helper
{
[Method("Full name", "Joins first and last names")]
public string GetFullName(Person person)
{
return string.Format(
"{0}, {1}", person.LastName, person.FirstName);
}
}
} | https://codeeffects.com/Doc/Business-Rule-External-Method-Attribute | CC-MAIN-2021-31 | refinedweb | 411 | 57.87 |
liberty, equality, and boxed primitive types
I want to understand how equality works in Scala. It's a complicated topic that's been going on for ten years.
Major concerns are:
-
null
- Unboxed number types
- Boxed number types
- Reference types
- Collections (especially of
F[+A])
Understanding equality means knowing how these combinations are compared.
Scala Language Specification
The language spec provides some hints, although it does not have the full information. Chapter 12 contains the definition of
Any as follows:
package scala /** The universal root class */ abstract class Any { /** Defined equality; abstract here */ def equals(that: Any): Boolean /** Semantic equality between values */ final def == (that: Any): Boolean = if (null eq this) null eq that else this equals that .... }
First thing to note is that both
equals and
== method are provided by
Any, encompassing both the value types and reference types. This is often called universal equality. In Scala 2.x, this allows comparison of two completely unrelated types such as
scala> 1 == "1" ^ warning: comparing values of types Int and String using `==` will always yield false res0: Boolean = false scala> Option(1) == Option("1") res1: Boolean = false
Given that
== is final, you might expect that the operator is strictly a symbolic alias of
equals method. However, later in the [numeric value types] section it says: .... }
This gives a glimpse at the fact that
== is not just a symbolic alias of
equals since Scala can overload the operators.
2010 'spec for == and ##'
The best reference of Scala 2.x behavior I found was 'spec for == and ##' draft Paul Phillips sent to Martin Odersky and scala-internals list on April 13, 2010, then reposted to == and equals in 2010.
Paul wrote:
Here is a kind of off the top of my head attempt to spec out equality and hash codes. I don't really speak spec-ese but this is written in the pidgin spec-ese within my grasp. Does this look approximately correct? (Anyone else feel free to chime in on that point.) What if anything would you like me to do with it?
- 4) If the static type of the left hand side allows for the possibility that it is a boxed or unboxed primitive numeric type (any of
Byte,
Short,
Int,
Long,
Float,
Double, or
Char) then: go to step 5.
If the static type definitively excludes those types, then: the result is
x.equals(y).
- 5) If the static types of both operands are primitive types, then: the result is that of the primitive comparison, exactly as performed in java.
If the static types are identical final types (for instance, both are
java.lang.Longs) then the result is
x.equals(y).
In all other cases, both operands are boxed if necessary and a method in
BoxesRunTimeis called. (The method will be semantically equivalent to
BoxesRunTime.equals, but a different method may be chosen to avoid repeating the above tests.)
BoxesRuntime.equals
All of the preceding logic is preserved, and then it proceeds as follows, where 'x' remains the left hand side operand and 'y' the right.
- 1) Runtime instance checks will be done to determine the types of the operands, with the following resolutions. (Resolutions represent the semantics, not necessarily the implementation.)
- 1a) If both sides of the comparison are boxed primitives, then they are unboxed and the primitive comparison is performed as in java.
- 1b) If 'x' is a class implementing the
scala.math.ScalaNumbertrait, then the result is
x.equals(y).
- 1c) If 'x' is a boxed primitive and 'y' is a class implementing the
scala.math.ScalaNumbertrait, then the result is
y.equals(x).
- 1d) Otherwise, the result is
x.equals(y).
....
The rest of the draft includes details about hash codes.
The notable points are:
- Reflexivity is specified.
- Unboxed primitive type equality is delegated to Java's primitive comparison.
- It tries to emulate unboxed primitive comparison even when boxed primitive types are given.
Java Language Specification
Java Language Specification 15.21.1 defines Numerical Equality Operators == and !=. JLS says that the numeric types are widened to
double if either lhs or rhs is a
double, otherwise
float,
long, or
int. Then it says it will follow IEEE 754 standard, including the fact that any comparison to
NaN returns
false.
Here's an example of
int getting converted into
float:
jshell> 1 == 1.0F $1 ==> true
In Java, numerical equality applies only to the unboxed primitive types.
jshell> java.lang.Integer.valueOf(1).equals(java.lang.Float.valueOf(1.0f)) $2 ==> false
cooperative equality
Scala emulates Java's widening even with boxed primitive types:
scala> java.lang.Integer.valueOf(1) == java.lang.Float.valueOf(1.0f) val res0: Boolean = true
I am not sure who coined the term, but this behavior is called cooperative equality.
In Java, whenever two values are
equal,
hashCode is required to return the same integer. Since we can't change
hashCode for the boxed primitives
## method was created. Here's from Paul's draft again:
The unification of primitives and boxed types in scala necessitates measures to preserve the equality contract: equal objects must have equal hash codes. To accomplish this a new method is introduced on
Any:
def ##: Int
This method should be called in preference to
hashCodeby all scala software which consumes hashCodes.
Here's a demonstration of
hashCode vs
##:
scala> 1.hashCode res1: Int = 1 scala> 1.## res2: Int = 1 scala> java.lang.Float.valueOf(1.0F).hashCode res3: Int = 1065353216 scala> java.lang.Float.valueOf(1.0F).## res4: Int = 1 scala> 1.0F.## res5: Int = 1
The conversion to boxed primitive types happens transparently in Scala when a numeric type is upcasted to
Any.
scala> (1: Any) res6: Any = 1 scala> (1: Any).getClass res7: Class[_] = class java.lang.Integer scala> (1: Any) == (1.0F: Any) res8: Boolean = true
This allows
Int and
Float to unify in Scala collections:
scala> Set(1, 1.0f, "foo") val res9: Set[Any] = Set(1, foo)
However it won't work for Java collection:
scala> import scala.jdk.CollectionConverters._ import scala.jdk.CollectionConverters._ scala> new java.util.HashSet(List(1, 1.0f, "foo").asJava) res10: java.util.HashSet[Any] = [1.0, 1, foo]
narrowness of Float
The details of
Float type is described in the Wikipedia entry IEEE 754 single-precision binary floating-point format: binary32. The 32 bits in
float breaks down as follows:
- 1 bit for sign
- 8 bits for exponents ranging from -126 to 127 (all zeros and all ones reserved)
- 23 bits represents 24 bits of signicand ranging from 1 to 16777215
The resulting floating-point number is
sign * 2^(exponent) * signicand.
Note that
int stores 32 bits of integers (or 31 bit for positives) but the float can express 23 bits accurately. This could lead to a rounding error by 1 for any integer above 0xFFFFFF (16777215).
jshell> 16777216 == 16777217F $2 ==> true
As the
int becomes larger, you can get more ridiculous results:
jshell> 2147483584F == 2147483647 $3 ==> true
This will break the
## contract for Scala as well.
scala> 16777217 == 16777217F res7: Boolean = true scala> 16777217.## == 16777217F.## res8: Boolean = false
In my opinion, we should treat Float type as 24 bit integer, and Double as 53 bit integer when it comes to widening. I've reported this as Weak conformance to Float and Double are incorrect #10773. There's also an open PR by Guillaume Deprecate numeric conversions that lose precision #7405.
NaN
Since this comes up in the discussion of equality, I should note that the comparison with
java.lang.Double.NaN would always return
false. An easy way to cause NaN is dividing
0.0 by
0. The most surprising thing about NaN comparison is that the NaN itself does not
== NaN:
scala> 0.0 / 0 res9: Double = NaN scala> 1.0 == (0.0 / 0) res10: Boolean = false scala> (0.0 / 0) == (0.0 / 0) res11: Boolean = false
In other words, Java or Scala's
== is not reflexive when NaN is involved.
Eq typeclass
Around 2010 was also the time when some of the Scala users started to adopt
=== operators introduced by Scalaz library. This bought in the concept of typeclass-based equality used in Haskell.
trait Equal[A] { self => def equal(a1: A, a2: A): Boolean }
This was later copied by libraries like ScalaTest and Cats.
scala> 1 === 1 res4: Boolean = true scala> 1 === "foo" <console>:37: error: type mismatch; found : String("foo") required: Int 1 === "foo" ^
I personally think this is a significant improvement over the universal equality since it's fairly common to miss the comparison of wrong types during refactoring. But the invariance also creates a fundamental issue with the way Scala 2.x defines data types through subclass inheritance. For example
Some(1) and
None would need to be upcasted to
Option[Int].
2011 'Rethinking equality'
Martin Ordersky was well aware of
===. In May 2011 he sent a proposal titled 'Rethinking equality': invariant
areEqual
@inline def areEqual[T](x: T, y: T)(implicit eq: Equals[T]) = eq.eql(x, y).
==would use
areEqualif it typechecks, otherwise it falls back to universal equality.
- Step 2:
x == yuses
areEqualeither lhs
A1or rhs
A2has
Equalsinstance.
- Step 3:
x == ybecomes equivalent to
areEqual.
This 2011 proposal went dormant quickly, but it's noteworthy since Martin did eventually change equality for Scala 3.x (Dotty).
2016 'Multiversal equality for Scala'
In May of 2016 Martin proposed Multiversal equality for Scala with dotty#1246.
Here's the definition of Eql:
/** A marker trait indicating that values of type `L` can be compared to values of type `R`. */ @implicitNotFound("Values of types ${L} and ${R} cannot be compared with == or !=") sealed trait Eql[-L, -R] object Eql { /** A universal `Eql` instance. */ object derived extends Eql[Any, Any] /** A fall-back instance to compare values of any types. * Even though this method is not declared as given, the compiler will * synthesize implicit arguments as solutions to `Eql[T, U]` queries if * the rules of multiversal equality require it. */ def eqlAny[L, R]: Eql[L, R] = derived // Instances of `Eql` for common Java types implicit def eqlNumber : Eql[Number, Number] = derived implicit def eqlString : Eql[String, String] = derived // The next three definitions can go into the companion objects of classes // Seq, Set, and Proxy. For now they are here in order not to have to touch the // source code of these classes implicit def eqlSeq[T, U](implicit eq: Eql[T, U]): Eql[GenSeq[T], GenSeq[U]] = derived implicit def eqlSet[T, U](implicit eq: Eql[T, U]): Eql[Set[T], Set[U]] = derived // true asymmetry, modeling the (somewhat problematic) nature of equals on Proxies implicit def eqlProxy : Eql[Proxy, AnyRef] = derived }
As noted in the comment as well as the Dotty documentation for Multiversal Equality:
Even though
eqlAnyis not declared a given instance, the compiler will still construct an
eqlAnyinstance as answer to an implicit search for the type
Eql[L, R], unless
Lor
Rhave
Eqlgiven instances defined on them, or the language feature
strictEqualityis enabled.
scala> class Box[A](a: A) // defined class Box scala> new Box(1) == new Box("1") val res1: Boolean = false scala> { | import scala.language.strictEquality | new Box(1) == new Box("1") | } 3 | new Box(1) == new Box("1") | ^^^^^^^^^^^^^^^^^^^^^^^^^^ | Values of types Box[Int] and Box[String] cannot be compared with == or !=
The documentation for Multiversal Equality also shows how
Eql instances can be derived automatically!
scala> class Box[A](a: A) derives Eql // defined class Box
reevaluation of cooperative equality
In September 2017, about a year after the multiversal equality proposal, Martin posted 'Can we get rid of cooperative equality?' on Contributors forum. In there he says that cooperative equality makes the data structure in Scala slower than the equivalent data structures in Java, and that the equality in Scala is complicated. The responses to the thread made by Jason Zaugg, Sébastien Doeraene and other major contributors to Scala makes this a interesting read into understanding the tradeoffs of cooperative equality.
Jason Zaugg wrote:
My intuition is the ability to use primitives as type arguments sends a signal that
Some[Long](x) == Some[Int](y)is morally equivalent to
x == y. In Java, you’d have to explicitly use the box type as the type argument.
Sébastien Doeraene wrote:
If
1 == 1Lis true, then I strongly believe that
(1: Any) == (1L: Any). However, nothing says that
1 == 1Lneeds to be true! We can instead make it false, or, even better, a compile error.
I thought Oliver Ruebenacker wrote the best summary:and
Double. Since these days almost every platform is 64 bit,
Longand
Doubleare natively efficient.
Requiring explicit conversion to
Double when comparing
Int and
Double sounds like a good tradeoff to me too, and it sounds consistent with the direction we are already taking with the multiversal equality.
warning for 1L equal 1
During 2.13.0 RC3, Kenji Yoshida reported #11551 showing that
Set was broken under
++ operation. He also sent a fix #8117, which was a one liner change from:
- if (originalHash == element0UnimprovedHash && element0.equals(element)) { + if (originalHash == element0UnimprovedHash && element0 == element) {
On Twitter he also suggested that we should warn about calling
equals or
hashCode on non-AnyRef. I've sent a PR #8120 so that the following would cause a warning:
[info] Running (fork) scala.tools.nsc.MainGenericRunner -usejavacp Welcome to Scala 2.13.0-pre-db58db9 (OpenJDK 64-Bit Server VM, Java 1.8.0_232). Type in expressions for evaluation. Or try :help. scala> def check[A](a1: A, a2: A): Boolean = a1.equals(a2) ^ warning: comparing values of types A and A using `equals` is unsafe due to cooperative equality; use `==` instead check: [A](a1: A, a2: A)Boolean
A few days ago I posted a poll:
Using Scala 2.13.1 or Dotty 0.21.0-RC1 what is the result of the following expression?
((1L == 1) == (1L equals 1)) -> (2147483584F == 2147483647)
- (false, false)
- (false, true)
- (true, false)
- (true, true)
Thanks to all 56 people who participated the poll. The results were
(false, false) 16.1% (false, true) 30.4% (true, false) 35.7% (true, true) 17.9%
My intent of the poll was to survey the awareness of cooperative equality among the Twitter users. Before that I want to mention James Hoffmann's video about high humidity coffee stroage. To test if high humidity affects the flavor of coffee, he conducts triangle test where samples X, Y, and Y are given blindly, and the first test it trying to pick out X, and then determine which one is better. If I simply posted a poll with true / false, just from the fact alone people would know there's something suspicious. It's not perfect, but I though adding one extra bit would make the poll better.
The first part is about cooperative equality, whereas the second part is about narrowness of 24 bit signicand. By summing the total we can say that 46.5% got the cooperative equality part right, and 48.3% got the Float rounding right. A random chance would get 50% so it's hard to know what percentage of people knows for sure.
summary
Here is some summary.
- Scala 2.x uses universal equality which allows comparison of
Intand
String. Dotty introduces "multiversal"
strictEqualitythat works similar to
Eqtypeclass.
- Currently both Scala 2.x and Dotty use Java's
==to compare unboxed primitive types. This mixes comparison of
Intwith
Floatand
Doubleetc.
Floatis narrower than
Int, and
Doubleis narrower than
Long.
- Because Scala transparently boxes
Intinto
java.lang.Integeras
(1: Any), it implements cooperative equality for
==and
##, but not for
equalsand
hashCode, which emulates widening for boxed primitive types. Many people are unaware of this behavior, and this could lead to surprising bugs if people believed that
equalsis same as
==.
- We might be able to remove cooperative equality if we are willing to make unboxed primitive comparison of different types
1L == 1an error. | http://eed3si9n.com/liberty-equality-and-boxed-primitive-types | CC-MAIN-2020-34 | refinedweb | 2,636 | 55.95 |
On Mon, 28 Jul 2008, Mike Travis wrote:> > Sorry, I didn't know that was the protocol. And yes, the clever idea of> compacting the memory is a good one (wish I would have thought of it... ;-)> But, and it's a big but, if you really have 4096 cpus present (not NR_CPUS,> but nr_cpu_ids), then 2MB is pretty much chump change.Umm. Yes, it's chump change, but if you compile a kernel to be generic, and you actually only have a few CPU's, it's no longer chump change.> But I'll redo the patch again.Here's a trivial setup, that is even tested. It's _small_ too. /* cpu_bit_bitmap[0] is empty - so we can back into it */ #define MASK_DECLARE_1(x) [x+1][0] = 1ul << (x) #define MASK_DECLARE_2(x) MASK_DECLARE_1(x), MASK_DECLARE_1(x+1) #define MASK_DECLARE_4(x) MASK_DECLARE_2(x), MASK_DECLARE_2(x+2) #define MASK_DECLARE_8(x) MASK_DECLARE_4(x), MASK_DECLARE_4(x+4) static const unsigned long cpu_bit_bitmap[BITS_PER_LONG+1][BITS_TO_LONGS(NR_CPUS)] = { MASK_DECLARE_8(0), MASK_DECLARE_8(8), MASK_DECLARE_8(16), MASK_DECLARE_8(24), #if BITS_PER_LONG > 32 MASK_DECLARE_8(32), MASK_DECLARE_8(40), MASK_DECLARE_8(48), MASK_DECLARE_8(56), #endif }; static inline const cpumask_t *get_cpu_mask(unsigned int nr) { const unsigned long *p = cpu_bit_bitmap[1 + nr % BITS_PER_LONG]; p -= nr / BITS_PER_LONG; return (const cpumask_t *)p; }that should be all you need to do.Honesty in advertizing: my "testing" was some trivial user-space harness, maybe I had some bug in it. But at least it's not _horribly_ wrong.And yes, this has the added optimization from Viro of overlapping the cpumask_t's internally too, rather than making them twice the size. So with 4096 CPU's, this should result 32.5kB of static const data. Linus | http://lkml.org/lkml/2008/7/28/320 | CC-MAIN-2015-06 | refinedweb | 278 | 70.33 |
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20051017 Firefox/1.6a1 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20051017 Firefox/1.6a1 Each "output", "input", and "select1" element inside a repeat will generate an error in the JavaScript Console for each top node in the repeat nodeset. This happens even if the data of these elements are shown according to the @bind / @ref in the output. The result is the same when using relative references, absolute references, or binds. Reproducible: Always Steps to Reproduce: 1. Add a repeat to your XForms, with @nodeset 2. Add an "output", "input", or "select1" element inside the repeat 3. Bind these to instance data using @ref or @bind Actual Results: Results are shown, but "size(repeat nodeset) * count(elements with @ref or @bind in the repeat)" errors show up for the repeat. Expected Results: Results are shown without any errors. Typical error message: "Error: XForms Error (10): Error parsing XPath expression: jobs:Description". Moving the "offending" elements outside the repeat still results in output (if the reference is absolute), but no errors.
Here's the relevant part of my XForms (chopped off unnecessary parts): <xf:repeat <xf:input <xf:label>Slot</xf:label> </xf:input> <xf:select1 <xf:label>Job</xf:label> <xf:item> <xf:label>INterconnection check before starting</xf:label> <xf:value>INterconnection check before starting</xf:value> </xf:item> <xf:item> <xf:label>IWP01.020 Main superconduct. cables soldering</xf:label> <xf:value>IWP01.020 Main superconduct. cables soldering</xf:value> </xf:item> <xf:alert>The job does not exist!</xf:alert> </xf:select1> <xf:output <xf:label>Executed by</xf:label> </xf:output> </xf:repeat>
I see the same, and yes only for controls inside a repeat using namespaces in their expressions, like ref="be:x".
Created attachment 200181 [details] Testcase
The failure starts here:
Created attachment 202246 [details] [diff] [review] Patch Problem was, that we (I) were cloning children into the contextcontainer, before inserting the cc into the repeat... I vaguely remember having done that for a reason, but that's lost on me. This patch inserts the cc into the repeat before inserting children. That makes the namespace lookup succeed. It also removes some dead code from nsXFormsUtils.
Comment on attachment 202246 [details] [diff] [review] Patch >+ // Insert context node >+ nsCOMPtr<nsIDOMNode> domNode; >+ rv = mHTMLElement->AppendChild(riElement, getter_AddRefs(domNode)); >+ NS_ENSURE_SUCCESS(rv, rv); >+ nit: please comment why you are adding this node before populating it with children since it seems contrary to what we do in other places. And in case someone thinks about moving it back in the future for some other reason, they'll at least know a problem that it could cause.
Created attachment 202964 [details] [diff] [review] Added comment
Checked into trunk
Thanks a lot! Keep up the great work! Too bad I have to wait for the next nightly build to test it ;-)
(In reply to comment #9) > Thanks a lot! Keep up the great work! > > Too bad I have to wait for the next nightly build to test it ;-) Well, that's all up to you: ;-)
checked into MOZILLA_1_8_BRANCH via bug 323691. Leaving open for now until it gets into 1.8.0
verfied fixed on MOZILLA_1_8_BRANCH | https://bugzilla.mozilla.org/show_bug.cgi?id=312848 | CC-MAIN-2017-26 | refinedweb | 553 | 58.48 |
outside air also attach themselves to people and animals, making clothing, shoes, bags, and pets convenient vehicles for carrying mold indoors (3).
- Shannon Davis
- 1 years ago
- Views:
Transcription
1 TECHNICAL RESOURCE SHEET frequently asked questions about mold The presence of unwanted or excessive moisture in buildings can lead to structural performance problems, as well as concerns about possible health risks. One area of concern is mold and mildew growth. Homeowners, builders, and contractors can benefit from having accurate, fact-based information on mold, mildew, and wood decay fungi. Unlike decay-producing fungi, mold and mildew alone do not cause significant loss in the strength of wood products. Q: What is mold? A: Mold and/or mildew are microscopic fungi that are present virtually everywhere, indoors and outdoors. They grow on a variety of organic materials, including wood products (1)(2). Q: how do molds enter an indoor environment and how do they grow? A: Molds may be present in outdoor or indoor air. Mold spores from the outside may enter a house through open doorways and windows, or though heating, ventilation, and airconditioning systems. Spores in the outside air also attach themselves to people and animals, making clothing, shoes, bags, and pets convenient vehicles for carrying mold indoors (3). Mold needs oxygen, water, nutrients, and a temperature between 40 degrees and 100 degrees Fahrenheit to grow (4). Mold spores can grow when they come into contact with a food source where there is excessive moisture, such as where there are leaks in roofs, pipes, walls, or plant pots, or where there has been flooding. Additionally, an average relative humidity of 80% or more over a month's duration will provide sufficient moisture for mold growth (5). Many organic materials (e.g., wet cellulose materials, including paper and paper products, cardboard, ceiling tiles, wood, and wood products) provide suitable nutrients to support mold growth. Inorganic materials such as dust, paints, wallpaper, insulation materials, drywall, carpet, fabric, and upholstery can also support mold growth (3). Q: Is mold occurring because buildings are now "air tight"? A: This rationale for moisture/mold problems is often presented by the media as the main driving force for the current increase in moisture/ mold claims. However, mold needs moisture to grow and moisture can come from multiple sources in a house, many of which have nothing to do with the amount of fresh air exchange a building experiences. A leaky foundation wall or a chronic plumbing leak releasing water onto drywall are two examples. There can be cases where the humidity level in a house would be lower with more fresh air, potentially alleviating some moisture/mold problems, but to characterize this as the root cause of mold problems is incorrect (6). Weyerhaeuser.com
2 Q: Why is mold a concern? A: Mold and/or mildew fungi do not cause decay; however, the environment that fosters mold growth will also support decay-producing organisms (10). Additionally, mold produces spores, which often become airborne and may create allergic reactions in some people (4). Q: a house has a water leak. What is the potential for growing mold in the leak area? A: Materials that are exposed to a constant leak or have been soaked and not dried thoroughly can support mold growth. Some molds can form a new colony in one or two days on damp materials. Molds do not require light and can continue growing indefinitely without light (7). Q: What about moisture in the crawl space? A: There is no definitive answer regarding moisture problems in the crawl space. A symposium on the subject held by the American Society of Heating, Refrigeration, and Air-Conditioning Engineers produced the following recommendations (8) : There should be proper drainage, clearance, and access. Crawl spaces should have ground covers for moisture control. These should be installed to limit evaporation from the soil. Heated crawl spaces should not be vented with outdoor air. Unheated crawl spaces may be vented, but there is no overriding need to do so for reasons of moisture control if an effective ground cover is present. (Note: check with local building codes to verify whether this practice is permitted in your jurisdiction.) Q: With wood, at what moisture content does mold become a concern? A: A moisture content greater than 19% is sufficient to support mold growth. This moisture content is also sufficient to support decayproducing organisms (9), which, unlike mold, can cause permanent loss of structural strength (10). Wood can achieve this moisture content when exposed to direct wetting from any source of moisture, or from extended exposure to an appropriate combination of high humidity and temperature. Q: Do structural adhesives cause mold to grow more readily? A: A recent study on fungal susceptibility of pine and aspen oriented strand board (OSB) found that the amount of mold growth on both OSB types was approximately equal to the mold growth on solid aspen (11). These results indicate that the adhesive in OSB has little or no effect on mold growth. Additionally, molds grow best on sources with freely available nutrients (12), and since the cured resins used in structural adhesives are a poor source of freely available nutrients, these adhesives are generally not associated with mold growth. Q: If someone thinks they have mold in their house, should they test for it? A: According to the EPA, sampling is unnecessary in most cases where visible mold growth is present. Since no EPA or other federal limits have been set for mold or mold spores, sampling cannot be used to check a building's compliance with federal mold standards (1). Testing for mold is also difficult because mold is everywhere, and therefore, testing will not prove that a house is free of mold (6). The U.S. Centers for Disease Control and Prevention (CDC) advises that it is not practical to test for mold growth in a house because large mold infestations can usually be seen or smelled (3). When testing is done, it is usually to compare levels of mold spores inside the house with levels outside the house (7). A thorough inspection of a house, to check for signs of moisture problems or active mold growth, is likely to be more effective than testing as a way to size up potential problems (6). It is usually not necessary to identify the species of mold growing in a residence, and CDC does not recommend routine sampling for molds. Current evidence indicates that allergies are the health problems most often associated with molds, and because the susceptibility of individuals varies greatly depending on the amount or type of mold, sampling and culturing would not be reliable in determining the health risk. CDC recommends that if the occupants are susceptible to mold and mold is seen or smelled indoors, there is a potential health risk and the homeowner should arrange for mold removal (3). Q: If a qualified environmental lab took samples of the mold inside a home and returned the results, can CDC or anyone else interpret these results? A: Standards for judging acceptable, tolerable, or normal quantities of mold have not been established. If the home-owner decides to pay for environmental sampling for molds, CDC recommends that the following items be addressed before the work starts: 2
3 Who will establish the criteria for interpreting the test results? What are their qualifications? What will be done, or what recommendations will be made based on the sampling results? Keep in mind that the results of samples taken in a unique situation cannot be interpreted without physical inspection of the contaminated area or without considering the building's characteristics and the factors that led to the present condition (3). Q: how do you determine whether a mold is "toxic"? A: Unfortunately, it is impossible for home-owners to distinguish between so-called "toxic" and "nontoxic" molds because they all look like black or gray sooty patches. Press coverage about lawsuits and health studies involving mold has focused on one type of mold called Stachybotrys chartarum, which is often referred to as "the toxic mold." However, there is no particular reason why this mold should be singled out, as all molds should be treated with caution (4). Q: What is Stachybotrys chartarum (Stachybotrys atra) and what should people do if they determine it is present in their buildings or homes? A: Stachybotrys chartarum (also known as Stachybotrys atra) is a greenish-black mold. It can grow on material with a high cellulose and low nitrogen content, such as fiberboard, gypsum board, paper, dust, and lint. However, it is not necessary to determine what type of mold may be present all molds should be treated the same with respect to potential health risks and removal (3), including (per the CDC) Stachybotrys chartarum. Mold growing in homes and buildings, whether it is Stachybotrys chartarum or other molds, indicates that there is a problem with water or moisture that needs to be addressed. Q: how common is mold, including Stachybotrys chartarum, in buildings? A: Molds are very common in buildings and homes and will grow anywhere there is moisture. The most common indoor molds are Cladosporium, Penicillium, Aspergillus, and Alternaria. The scientific community does not have accurate information about how often Stachybotrys chartarum is found in buildings and homes. It is less common than other mold species; however, it is not considered rare (3). Q: Will mold affect the structural integrity of the wood in my home? A: Surface mold on wood from shortterm wetting does not structurally damage the wood (6). Mold and mildew are a different type of fungi than those that cause wood to rot. Unlike decay-producing fungi, mold and mildew alone do not cause significant loss in the strength of wood products. Nonetheless, mold and mildew on wood indicate high moisture, and prolonged periods of high moisture may also support the growth of decay-producing fungi. This is one of the reasons why it is important to prevent the growth of mold and mildew (13). Q: how can mold growth be prevented? A: Controlling moisture is the most important part of mold-growth prevention. Mold will not grow if moisture is not present. A properly constructed building is designed to keep the inside of it dry, which prevents mold growth (1)(7)(10). As part of routine maintenance, buildings should be inspected for evidence of water damage and visible mold, and any conditions that could cause mold (such as water leaks, condensation, infiltration, or flooding) should be corrected (3). Specific Recommendations from CDC: Keep the overall humidity level in the house below 50%. This 50% threshold is often recommended to minimize localized conditions of relative humidity greater than the commonly referenced threshold of 80% that supports active mold growth. Use an air conditioner or a dehumidifier during humid months. Be sure the home has adequate ventilation, including exhaust fans in the kitchen and bathrooms. Use paints that contain mold inhibitors. Clean bathrooms with moldkilling products. Do not carpet bathrooms. Remove and replace flooded carpets. Q: how do you get rid of existing mold? A: It is impossible to get rid of all mold and mold spores indoors; some mold spores will always be found floating through the air and settling with house dust. However, mold spores will not grow unless moisture is present. Indoor mold growth can and should be prevented or controlled by controlling moisture indoors. If there is mold growth in a home, the mold must be cleaned up and the moisture problem fixed. If the mold is cleaned up, but the moisture problem remains, then the mold will probably return (1). 3
4 Q: Hhow do I determine if a product has been damaged beyond cleaning? A: In an initial assessment, the key question is: Does this appear to be a "mold only" condition, or does it appear to have become a potential decay situation? Mold and mildew cause no structural damage to wood other than unsightly discoloration (10). Mold growing on solid lumber or other structural wood products is most likely a surface contamination issue, not a structural issue, so it can be cleaned, dried, and used. However, if there is any evidence of decay, the product should be replaced (14). Conversely, porous materials that are moldy, such as ceiling tiles, sheetrock, and carpet should be discarded (13). If the wood is badly decayed, this will be quite visible. Two common visual results of decay are a bleached and stringy appearance to the wood, or a darkened surface with cubical cracking. If fungal growth is visible on the surface, the wood has probably suffered strength loss, but do not rely on visual cues alone. Wood can appear stained and be sound, or can appear normal yet still have already suffered significant strength loss from decay. Use the pick test to determine whether or not the wood is sound: Insert the point of a knife at a shallow angle to the surface and attempt to lever up a thin splinter. If the wood splinters, it is sound. If instead it breaks just above the blade Figure 1: Pick Test on Sound Wood Splintering Break Graphics reproduced with the permission of FPInnovations. (15) like a carrot snapping in half, it is decayed (15). See Figure 1 below. Q: The homeowner has identified mold growth and fixed the moisture problem. Now, how do they clean up the mold? A: A detergent and water solution (1) (recommended by the EPA) or a bleach-to-water solution (1 cup bleach per 1 gallon water as recommended by the CDC) (3) is suggested to clean mold and kill fungi. The detergent/water solution is often recommended because it has less potential than the bleach/water solution to be used in an unsafe manner. This cleaning process will not prevent future growth only an environmental change (i.e., eliminating the moisture) can prevent future mold growth. The CDC recommends that large mold infestations should be addressed by a professional who has experience with cleaning mold in buildings and homes (3). Q: Does cleaning have any negative effect on the structural performance of engineered wood products? A: Mold cleaning procedures recommended by the EPA or CDC (i.e., scrubbing with a detergent and water solution or scrubbing with a dilute bleach solution) will not degrade the structural characteristics of engineered wood products. If a remediator wishes to use another type of cleaning solution, he Pick Test on Decayed Wood Brash Break 4 or she should be asked to provide documentation that the solution is not detrimental to wood-based products. If the information is not available, request a copy of the MSDS (Material Safety Data Sheet) for the product and contact Weyerhaeuser engineering for further advice. Q: "Toxic" mold has attracted a lot of media attention. Should a homeowner be concerned about possible health risks? A: The CDC provides the following answer to this question: "The hazards presented by molds that may contain mycotoxins should be considered the same as other common molds that can grow in [a]. (3) A common-sense approach should be used for any mold contamination existing inside buildings and homes. The common health concerns from molds include hay-fever-like allergic symptoms. Certain individuals with chronic respiratory disease (such as chronic obstructive pulmonary disorder or asthma) may experience difficulty breathing when exposed to mold. Individuals with immune suppression may be at increased risk for infection from molds. If the homeowner or any family members have these conditions, a qualified medical clinician should be consulted for diagnosis and treatment. For the most part, one should take routine measures to prevent mold growth in the home. (3)
5 Links to More Information NAHB Research Center's ToolBase Services provides two documents that deal with mold: Helping Your Buyers Understand Mold During the Building Process and Mold in Residential Buildings. Canadian Wood Council and FPInnovations have collaborated on a website addressing the durability of wood, with a section specifically addressing mold. U.S. Centers for Disease Control and Prevention has developed a Q&A covering mold related topics. U.S. Environmental Protection Agency covers issues, including cleanup, health concerns, and air quality, in A Brief Guide to Mold, Moisture, and Your Home. APA - The Engineered Wood Association's informational flyer briefly covers cleanup and prevention of mold and moisture problems. Their Q&A section also has one entry dealing with mold. USDA Forest Service, Forest Products Laboratory has a few documents related to moisture in homes. The most notable are Recognize, Remove, and Remediate Mold and Mildew and Mold and Mildew on Wood: Causes and Treatment. American Wood Council (AWC) offers facts about how and when mold might start to grow in or around your home. Kansas State University. Although published in 1995, this document contains excellent information relating to the identification of mold in the home, preventative measures, and a highly detailed section on cleanup. New York City Health Department publishes Facts About Mold, a fact sheet with questions and answers about health concerns, healthy homes, and cleanup information. American Phytopathological Society publishes Stachybotrys chartarum: The Toxic Indoor Mold. Building Science Corporation, a private consulting firm that specializes in preventing and resolving problems related to building design, construction, and operation, has written a few documents relating to mold. Topics covered include testing, basic mold Q&A, cleanup, and designing for moisture. References 1. U.S. Environmental Protection Agency A Brief Guide to Mold, Moisture, and Your Home. 2. California Dept. of Health Services Mold in My Home: What Do I Do? 3. CDC National Center for Environmental Health Questions and Answers on Stachybotrys Chartarum and Other Molds. 4. NAHB Research Center's ToolBase Services Mold in Residential Buildings. 5. International Energy Agency Annex 14: Condensation and Energy, Vol. 2: Guidelines and Practice. 6. NAHB Research Center's ToolBase Services Helping Your Buyers Understand Mold During the Building Process. 7. Bode M, Munson D Controlling Mold Growth in the Home. Kansas State University. 8. Rose W, TenWolde A Recommended Practices for Controlling Moisture in Crawl Spaces. Issues in Crawl Space Design and Construction-A Symposium Summary. ASHRAE Technical Bulletin 88192, ASHRAE. 9. Kirby S, Wiggins C Moisture Control and Prevention Guide. NC State University. 10. Clausen C Recognize, Remove, and Remediate Mold and Mildew. Proceedings of 2nd Annual Conference on Durability and Disaster Mitigation in Wood-Frame Housing. 11. Laks P et al Fungal Susceptibility of Interior Commercial Building Panels. Forest Products Journal Vol. 52(5); Forintek Canada Corp. and University of British Columbia School of Occupational and Environmental Hygiene Discolorations on Wood Products: Causes and Implications. 13. APA - The Engineered Wood Association Mold and Mildew. 14. Yost N, et al What You Need to Know About Mold. Building Science Corporation. 15. Canadian Wood Council and FPInnovations. Website on the durability of wood products. CONTACT US October 2011 Reorder 1504 This document supersedes all previous versions. If this is more than one year old, contact your dealer or Weyerhaeuser rep., Weyerhaeuser, and Trus Joist are registered trademarks of Weyerhaeuser NR Weyerhaeuser NR Company. All rights reserved.
Facts Regarding Mold on Wood Structural Building Components
Facts Regarding Mold on Wood Structural Building Components Issues involving mold on building materials, whether during construction or in completed and occupied structures, have gained considerable
Got Mold? Frequently Asked Questions About Mold
Office of Environmental Health & Safety Got Mold? Frequently Asked Questions About Mold What are molds? With more than 100,000 species in the world, it is no wonder molds can be found everywhere. Neither Vermont
University of Vermont Department of Physical Plant Burlington, Vermont WATER INTRUSION GUIDELINES In accordance with: IIRC S500 IIRC S520 EPA Document 402-K-01-001 REVISED AND DISTRIBUTED BY: THE UNIVERSITY
The Facts About Mold. Operation Outreach
The Facts About Mold Operation Outreach Mold/IAQ Resources from AIHA More AIHA Operation Outreach Brochures: Do I Work In a Sick Building? Guidelines for Selecting an Indoor Air Quality Consultant Is Air
Mold. Guidelines for New Jersey Residents. Understanding Mold Investigations & Remediation
Mold Guidelines for New Jersey Residents Understanding Mold Investigations & Remediation What Services Should I Ask For? What Are Important Inspection Procedures? Is Mold Sampling Helpful? What Information
Mold and mold spores
Mold and mold spores Mold and mold spores are everywhere around us and have always been a part of our environment. The air we breath is a virtual jungle of fungal spores. We routine encounter mold spores:
Health Effects of Fungi (Mold) in Residential Construction 1990-2009
Health Effects of Fungi (Mold) in Residential Construction 1990-2009 Paul Ellringer PE, CIH Air Tamarack Inc. 651-696-0267 What Health Problems Have Been Related to Fungi? Odor Concerns
Facts About Moulds. Should I be concerned about mould in my building?
Introduction This information has been developed by Saskatchewan Health and the Occupational Health and Safety Division of Saskatchewan Labour and is intended to apply to private residences, public buildings
Mold Remediation in Occupied Homes
building science.com 2008 Building Science Press All rights of reproduction in any form reserved. Mold Remediation in Occupied Homes Research Report - 0210 January-2002 Joseph Lstiburek, Terry Brennan
Diamond Environmental Services, Inc. s Most Frequently Asked Questions (FAQs)
Diamond Environmental Services, Inc. s Most Frequently Asked Questions (FAQs) The following are our most frequently asked questions (FAQs) concerning asbestos and mold. Asbestos What is asbestos? Asbestos
Healthy Homes Training MOLD AND MILDEW
Healthy Homes is a program of the Southern New Jersey Perinatal Cooperative with funding by the New Jersey Department of Health and Senior Services snjpc.org 2012 Why Parents Should Worry 6 million children
Safety Policy Manual Policy No. 112
Policy: Mold Prevention, Assessment and Remediation Program Page 1 of 9 APPLICATION NYU Langone Medical Center (NYULMC) POLICY SUMMARY NYULMC is committed to protecting employees, patients, and visitors
PolyMaster Foam Insulation and Resistance to Mold PM Mold Statement RetroFoam is naturally resistant to mold growth, and does not contain cellulose or other fiber which will sustain mold growth. RetroFo
Moisture Management. Infection Prevention and Corporate Safety. Contents. Posttest... 12
Moisture Management Infection Prevention and Corporate Safety This self-directed learning module contains information you are expected to know to protect yourself, our patients, and our guests. Target
Managing Water Infiltration into Buildings. Water Damage Check List
Managing Water Infiltration into Buildings A Systematized Approach for Remediating Water Problems in Buildings due to Floods, Roof Leaks, Potable Water Leaks, Sewage Backup, Steam Leaks and Groundwater in My Home: What Do I Do?
Mold in My Home: What Do I Do? This packet is meant to provide basic information to people who have experienced water damage to their home and resulting mold concerns. It describes health concerns related
NYU Safety Policy Manual
NYU Safety Policy Manual Page 1 of 6 Subject: Mold Prevention, Assessment, and Remediation Program Policy No. 167 ISSUE DATE REPLACES ORIGINATOR APPLICATION NYU Washington Square Campus PURPOSE The purpose
Guidelines for Cleaning Staff on Managing Mould Growth in State Buildings
Guidelines for Cleaning Staff on Managing Mould Growth in State Buildings Prepared by the State Claims Agency 2 Index 1. Background 2. What are moulds? 3. What are the possible health effects? 4. How do.
Water Incursion Standard Operating Procedure
Water Incursion Standard Operating Procedure Purpose: To provide a standardized procedure in the event of water incursion into any Penn State University facility. References: A Brief Guide to Mold, Moisture
There s Mold In My House! What Do I Do?
City of West Sacramento Community Development Department 1110 West Capitol Avenue, West Sacramento CA 95691 PLACE STAMP HERE This document is best used in conjunction with the CDPH handout, Mold in My
PROPERTY INSPECTION OR ASSESSMENT OF DAMAGES
PROPERTY INSPECTION OR ASSESSMENT OF DAMAGES ADDRESS: 123 MAIN ST MIAMI, FL CLIENT: BUYER ID No: 2014193 DATE: 5/13/2014 INSPECTION OR ASSESSMENT BY: GAIA CONSTRUCTION INC. CGC 1516136 FLORIDA HI-2792
Flood Response and Mold Prevention Program
Part 2: Drying Out Your Home
Part 2: Drying Out Your Home Now you're ready to begin drying out your home and establishing your plan for rebuilding. The information contained in this section will help you to dry out and decontaminate
Regulations & Guidelines
Regulations & Guidelines Short History of Mold Standards Leviticus 13: 47-50, 14: 39-47 Initial assessment by priest; follow-up visit in 7 days If mold spreads to walls, tear out contaminated stones REMEDIATION KEY STEPS
MOLD REMEDIATION KEY STEPS The EPA has developed the following guidelines for mold remediation managers. These guidelines are generally helpful, but we believe an expert in the industry should be consulted
Nick s Inspection Services
Nick s Inspection Services 909 Shorthorn Grain Valley Mo. 64029 Phone: 816-355-0368 Cell: 816-225-5783 Email: [email protected] Building Inspection Report and Protocol Inspection Date: 08-20-14
Mold Growth In Buildings
Mold Growth In Buildings Austin Sumner, MD, MPH State Epidemiologist Environmental Health 10/21/2009 Today s Presentation Introduction to mold Role of Town Health Officers Building compliance Health effects
STANDARD OPERATING PROCEDURES FOR MOLD REMEDIATION
West Virginia University Environmental Health and Safety STANDARD OPERATING PROCEDURES FOR MOLD REMEDIATION Origination Date August 2013. Contents Purpose:... 2 Job Scope... 2 Definitions... 2 Roles/Responsibilities:...,
Mold and Human Health
North Carolina Department of Health and Human Services Division of Public Health Occupational and Environmental Epidemiology Mold and Human Health Mold is a term used to describe a type of fungus that
CENTER FOR ENVIRONMENTAL HEALTH Emergency Response/Indoor Air Quality Program
CENTER FOR ENVIRONMENTAL HEALTH Emergency Response/Indoor Air Quality Program Guidance Concerning Remediation and Prevention of Mold Growth and Water Damage in Public Schools/Buildings to Maintain Air
ASTHMA REGIONAL COUNCIL
ASTHMA REGIONAL COUNCIL WHAT S THAT SMELL? Simple Steps to Tackle School Air Problems ARC is a coalition of governmental and community agencies dedicated to addressing the environmental contributors to
BUREAU OF ENVIRONMENTAL HEALTH Emergency Response/Indoor Air Quality Program
BUREAU OF ENVIRONMENTAL HEALTH Emergency Response/Indoor Air Quality Program Use of Moisture Measuring Devices in Evaluating Water Damage in Buildings July 2007 The evaluation of mold colonization of building
MOULD AND CONDENSATION IN YOUR HOME
MOULD AND CONDENSATION IN YOUR HOME Indoor condensation can cause damage to fabrics, discolour paint and wallpaper but, more importantly, it promotes conditions suitable for the growth of mould. When water
Mold Inspection Report (Initial Assessment)
Morlin Home Services, LLC 4435 Nanticoke Court Sw Lilburn, GA. 30047 Phone: (770) 564-1505 Fax: (770) 564-1575 Cell: (770) 344-7416 [email protected] Mold Inspection
Molds
NORMI Professional Guidance for DIY Projects
NORMI Professional Guidance for DIY Projects In general, a well-trained and experienced indoor mold remediation professional (CMR) should be consulted when performing remediation. You may find a NORMI
Steps for Cleaning Mold
Michigan Department of Community Health Steps for Cleaning Mold Before getting started, get to know MOLD: Mend There are many places you can find mold in your home. However, mold always needs a damp or
E-mail: [email protected] Moisture Challenges & Solutions: Proactive Construction Strategies 62 nd Annual Convention of the International Builders Show Orange
Moulds. Workplace Guidelines for Recognition, Assessment, and Control. Occupational Health and Safety Council of Ontario
Moulds Workplace Guidelines for Recognition, Assessment, and Control Occupational Health and Safety Council of Ontario 1 Moulds More and more workplaces are involved in investigating or removing mould | http://docplayer.net/998410-Outside-air-also-attach-themselves-to-people-and-animals-making-clothing-shoes-bags-and-pets-convenient-vehicles-for-carrying-mold-indoors-3.html | CC-MAIN-2016-50 | refinedweb | 4,552 | 52.49 |
Faster playback start means more people watching your video. That's a known fact. In this article I'll explore techniques you can use to accelerate your media playback by actively preloading resources depending on your use case.
Credits: copyright Blender Foundation | .
TL;DR
Video preload attribute
If the video source is a unique file hosted on a web server, you may want to
use the video
preload attribute to provide a hint to the browser as to how
much information or content to preload. This means Media Source Extensions
(MSE) is not compatible with
preload.
Resource fetching will start only when the initial HTML document has been
completely loaded and parsed (e.g. the
DOMContentLoaded event has fired)
while the very different
window.onload event will be fired when resource
has actually been fetched.
Setting the
preload attribute to
metadata indicates that the user is not
expected to need the video, but that fetching its metadata (dimensions, track
list, duration, and so on) is desirable. Note that starting in Chrome 64, the
default value for
preload is
metadata. (It was
auto previously).
<video id="video" preload="metadata" src="file.mp4" controls></video> <script> video.addEventListener('loadedmetadata', function() { if (video.buffered.length === 0) return; var bufferedSeconds = video.buffered.end(0) - video.buffered.start(0); console.log(bufferedSeconds + ' seconds of video are ready to play!'); }); </script>
Setting the
preload attribute to
auto indicates that the browser may cache
enough data that complete playback is possible without requiring a stop for
further buffering.
<video id="video" preload="auto" src="file.mp4" controls></video> <script> video.addEventListener('loadedmetadata', function() { if (video.buffered.length === 0) return; var bufferedSeconds = video.buffered.end(0) - video.buffered.start(0); console.log(bufferedSeconds + ' seconds of video are ready to play!'); }); </script>
There are some caveats though. As this is just a hint, the browser may completely
ignore the
preload attribute. At the time of writing, here are some rules
applied in Chrome:
- When Data Saver is enabled, Chrome forces the
preloadvalue to
none.
- In Android 4.3, Chrome forces the
preloadvalue to
nonedue to an Android bug.
- On a cellular connection (2G, 3G, and 4G), Chrome forces the
preloadvalue to
metadata.
Tips
If your website contains many video resources on the same domain, I would
recommend you set the
preload value to
metadata or define the
poster
attribute and set
preload to
none. That way, you would avoid hitting
the maximum number of HTTP connections to the same domain (6 according to the
HTTP 1.1 spec) which can hang loading of resources. Note that this may also
improve page speed if videos aren't part of your core user experience.
Link preload
As covered in other articles, link preload is a declarative fetch that
allows you to force the browser to make a request for a resource without
blocking the
window.onload event and while the page is downloading. Resources
loaded via
<link rel="preload"> are stored locally in the browser, and are
effectively inert until they're explicitly referenced in the DOM, JavaScript,
or CSS.
Preload is different from prefetch in that it focuses on current navigation and fetches resources with priority based on their type (script, style, font, video, audio, etc.). It should be used to warm up the browser cache for current sessions.
Preload full video
Here's how to preload a full video on your website so that when your JavaScript asks to fetch video content, it is read from cache as the resource may have already been cached by the browser. If the preload request hasn't finished yet, a regular network fetch will happen.
<link rel="preload" as="video" href=""> <video id="video" controls></video> <script> // Later on, after some condition has been met, set video source to the // preloaded video URL. video.src = ''; video.play().then(_ => { // If preloaded video URL was already cached, playback started immediately. }); </script>
Because the preloaded resource is going to be consumed by a video element in
the example, the
as preload link value is
video. If it were an audio
element, it would be
as="audio".
Preload the first segment
The example below shows how to preload the first segment of a video with
<link
rel="preload"> and use it with Media Source Extensions. If you're not familiar
with the MSE Javascript API, please read MSE basics.
For the sake of simplicity, let's assume the entire video has been split into smaller files like "file_1.webm", "file_2.webm", "file_3.webm", etc.
<link rel="preload" as="fetch" href=""> "'); // If video is preloaded already, fetch will return immediately a response // from the browser cache (memory cache). Otherwise, it will perform a // regular network fetch. fetch('') .then(response => response.arrayBuffer()) .then(data => { // Append the data into the new sourceBuffer. sourceBuffer.appendBuffer(data); // TODO: Fetch file_2.webm when user starts playing video. }) .catch(error => { // TODO: Show "Video is not available" message to user. }); } </script>
Support
Link preload is not supported in every browser yet. You may want to detect its availability with the snippets below to adjust your performance metrics.
function preloadFullVideoSupported() { const link = document.createElement('link'); link.as = 'video'; return (link.as === 'video'); } function preloadFirstSegmentSupported() { const link = document.createElement('link'); link.as = 'fetch'; return (link.as === 'fetch'); }
Manual buffering
Before we dive into the Cache API and service workers, let's see how to manually buffer a video with MSE. The example below assumes that your web server supports HTTP Range requests but this would be pretty similar with file segments. Note that some middleware libraries such as Google's Shaka Player, JW Player, and Video.js are built to handle this for you.
"'); // Fetch beginning of the video by setting the Range HTTP request header. fetch('file.webm', { headers: { range: 'bytes=0-567139' } }) .then(response => response.arrayBuffer()) .then(data => { sourceBuffer.appendBuffer(data); sourceBuffer.addEventListener('updateend', updateEnd, { once: true }); }); } function updateEnd() { // Video is now ready to play! var bufferedSeconds = video.buffered.end(0) - video.buffered.start(0); console.log(bufferedSeconds + ' seconds of video are ready to play!'); // Fetch the next segment of video when user starts playing the video. video.addEventListener('playing', fetchNextSegment, { once: true }); } function fetchNextSegment() { fetch('file.webm', { headers: { range: 'bytes=567140-1196488' } }) .then(response => response.arrayBuffer()) .then(data => { const sourceBuffer = mediaSource.sourceBuffers[0]; sourceBuffer.appendBuffer(data); // TODO: Fetch further segment and append it. }); } </script>
Considerations
As you're now in control of the entire media buffering experience, I suggest you consider the device's battery level, the "Data-Saver Mode" user preference and network information when thinking about preloading.
Battery awareness
Please take into account the battery level of users' devices before thinking about preloading a video. This will preserve battery life when the power level is low.
Disable preload or at least preload a lower resolution video when the device is running out of battery.
if ('getBattery' in navigator) { navigator.getBattery() .then(battery => { // If battery is charging or battery level is high enough if (battery.charging || battery.level > 0.15) { // TODO: Preload the first segment of a video. } }); }
Detect "Data-Saver"
Use the
Save-Data client hint request header to deliver fast and light
applications to users who have opted-in to "data savings" mode in their
browser. By identifying this request header, your application can customize and
deliver an optimized user experience to cost- and performance-constrained
users.
Learn more by reading our complete Delivering Fast and Light Applications with Save-Data article.
Smart loading based on network information
You may want to check
navigator.connection.type prior to preloading. When
it's set to
cellular, you could prevent preloading and advise users that
their mobile network operator might be charging for the bandwidth, and only start
automatic playback of previously cached content.
if ('connection' in navigator) { if (navigator.connection.type == 'cellular') { // TODO: Prompt user before preloading video } else { // TODO: Preload the first segment of a video. } }
Checkout the Network Information sample to learn how to react to network changes as well.
Pre-cache multiple first segments
Now what if I want to speculatively pre-load some media content without
knowing which piece of media the user will eventually pick. If the user is on a
web page that contains 10 videos, we probably have enough memory to fetch one
segment file from each but we should definitely not create 10 hidden video
elements and 10
MediaSource objects and start feeding that data.
The two part example below shows you how to pre-cache multiple first segments of video using the powerful and easy-to-use Cache API. Note that something similar can be achieved with IndexedDB as well. We're not using service workers yet as the Cache API is also accessible from the Window object.
Fetch and cache
const videoFileUrls = [ 'bat_video_file_1.webm', 'cow_video_file_1.webm', 'dog_video_file_1.webm', 'fox_video_file_1.webm', ]; // Let's create a video pre-cache and store all first segments of videos inside. window.caches.open('video-pre-cache') .then(cache => Promise.all(videoFileUrls.map(videoFileUrl => fetchAndCache(videoFileUrl, cache)))); function fetchAndCache(videoFileUrl, cache) { // Check first if video is in the cache. return cache.match(videoFileUrl) .then(cacheResponse => { // Let's return cached response if video is already in the cache. if (cacheResponse) { return cacheResponse; } // Otherwise, fetch the video from the network. return fetch(videoFileUrl) .then(networkResponse => { // Add the response to the cache and return network response in parallel. cache.put(videoFileUrl, networkResponse.clone()); return networkResponse; }); }); }
Note that if I were to use HTTP Range requests, I would have to manually recreate
a
Response object as the Cache API doesn't support Range responses yet. Be
mindful that calling
networkResponse.arrayBuffer() fetches the whole content
of the response at once into renderer memory, which is why you may want to use
small ranges.
For reference, I've modified part of the example above to save HTTP Range requests to the video pre-cache.
... return fetch(videoFileUrl, { headers: { range: 'bytes=0-567139' } }) .then(networkResponse => networkResponse.arrayBuffer()) .then(data => { const response = new Response(data); // Add the response to the cache and return network response in parallel. cache.put(videoFileUrl, response.clone()); return response; });
Play video
When a user clicks a play button, we'll fetch the first segment of video available in the Cache API so that playback starts immediately if available. Otherwise, we'll simply fetch it from the network. Keep in mind that browsers and users may decide to clear the Cache.
As seen before, we use MSE to feed that first segment of video to the video element.
function onPlayButtonClick(videoFileUrl) { video.load(); // Used to be able to play video later. window.caches.open('video-pre-cache') .then(cache => fetchAndCache(videoFileUrl, cache)) // Defined above. .then(response => response.arrayBuffer()) .then(data => {"'); sourceBuffer.appendBuffer(data); video.play().then(_ => { // TODO: Fetch the rest of the video when user starts playing video. }); } }); }
Create Range responses with a service worker
Now what if you have fetched an entire video file and saved it in the Cache API. When the browser sends an HTTP Range request, you certainly don't want to bring the entire video into renderer memory as the Cache API doesn't support Range responses yet.
So let me show how to intercept these requests and return a customized Range response from a service worker.
addEventListener('fetch', event => { event.respondWith(loadFromCacheOrFetch(event.request)); }); function loadFromCacheOrFetch(request) { // Search through all available caches for this request. return caches.match(request) .then(response => { // Fetch from network if it's not already in the cache. if (!response) { return fetch(request); // Note that we may want to add the response to the cache and return // network response in parallel as well. } // Browser sends a HTTP Range request. Let's provide one reconstructed // manually from the cache. if (request.headers.has('range')) { return response.blob() .then(data => { // Get start position from Range request header. const pos = Number(/^bytes\=(\d+)\-/g.exec(request.headers.get('range'))[1]); const options = { status: 206, statusText: 'Partial Content', headers: response.headers } const slicedResponse = new Response(data.slice(pos), options); slicedResponse.setHeaders('Content-Range': 'bytes ' + pos + '-' + (data.size - 1) + '/' + data.size); slicedResponse.setHeaders('X-From-Cache': 'true'); return slicedResponse; }); } return response; } }
It is important to note that I used
response.blob() to recreate this sliced
response as this simply gives me a handle to the file (in Chrome) while
response.arrayBuffer() brings the entire file into renderer memory.
My custom
X-From-Cache HTTP header can be used to know whether this request
came from the cache or from the network. It can be used by a player such as
ShakaPlayer to ignore the response time as an indicator of network speed.
Have a look at the official Sample Media App and in particular its ranged-response.js file for a complete solution for how to handle Range requests. | https://developers.google.cn/web/fundamentals/media/fast-playback-with-video-preload?hl=pl | CC-MAIN-2018-39 | refinedweb | 2,116 | 57.67 |
♦
;
♦
VOL. I.
CALDWELL, IDAHO TERRITORY, SATURDAY, JANUARY 20, 1884.
NO. 7.
The Caldwell Tribune
Is Published Every Saturday at
Caldwell, Idaho Territory.
BY
W. J, CUDDY.
OFFICE, 509 MARKET AVENUE.
sUMSClUITION:
One Year....
Six Months...
Three Months.
$ 0.00
1.50
1.00
)
Single Copy, Ten Cents.
(öfAdvertistug rates given on applica
tion.
* I
Tenders his professional sendees to the citi
zens of Caldwell and Boise valley.
Office at Cox & Martin's drug store.
OFFICE HOUR* from 9 a. m. till 4 p. m.
BURTON&BROWN,
Real Estate and Law Office.
j
!
j
j
Attornoy fit XjRW !
I
i
i
I
j
~ I
Apply at Danielson s.
AND
NOTARY PUBLIC,
CALDWELL,
Office next door to Town Co. 's Office.
IDAHO.
i
LDANFORTH, M. 0.,
Physician and Surgeon,
Has permanently located in the town of
ildwell, and will attend promptly to all
calls, day or night, In his profession. I also
have a good assortment of drugs and patent
medicines at Danielson's store.
F. S. EASTON,
Physician and Surgeon,
CALDWELL, IDAHO.
Diseases of women and children a special
ty. Obstetrical and office cases cash. Office
at tbe Haskell House; also leave orders at
the drugstore of Cox A Martin.
Barber Shop
J
GÜS. WOHLGEMUTH, Prop.
:
First-class tonsorial work by tbe liest ar
lists In Idaho
A. A. KUMMET,.
H. .T. GOETZ MAN.
ROMMEL 4 60EIZMAH.
CONTRACTORS AND BDILDKRS.
Fine Job Work a Specialty. Keep on
Hand a Full Stock of Lumber,
Sash, Doors and Mould
ings.
/
CALDWELL,
IDAHO.
SEWING
MACHINES!
-FOR —
Sewing Machines, Parts, Oil,
Needles, Etc.,
Oiill on or writ«' to
C. ELLSWORTH,
IDAHO.
BOISE CITY,
Branch Office at Welser City, Hon. T. M.
Jeffreys A Co., Agents.
John M. Lamii,
Boise City, I. T.
Chas. II. Ukki>,
Caldwell. I. T.
/
REED & LAMB
5
REAL ESTATF,
CoDvevancii and Collection Die
Front Avenue, next door to Town Com
pany's Office,
Caldwell, I. T.
Real Estate Transfers made on reasonable
terms. All kinds of Conveyances carefully
and correctly drawn.
J
Special Attention Given to Collections,
I
NOTARY rUDLIC IS OFFICE.
NEWS OF THE WEEK
GENERAL.
The Chinese fear that the French are
entertaining designs upon Canton, and the
people'are very uneasy. Hai Phong reports
say that Admiral Courbet would advance
upon Bacnlnh without awaiting reinforce
ments.
The French government, it is report
ed,"has been assured that England has urged
China to accept the accomplished facts and
arrange terms of peace with France.
It is stated that the Egyptian govern
ment has given orders to evacuate Khar
toum. The guns will be spiked and the
powder destroyed. It is believed that all
efforts will now be concentrated in the de
fense of Massa wall and Suakim.
The Paris Figaro asserts that direct
negotiatiations between China and France
will he resumed on a basis on the new fron
tier of Tonquin and the amnesty of the
Black Flags. China will also guarantee the
free navigation of the Song Noi river to Lao
Kay.
• The whiskey men of Louisville have
sent out a circular asking constituents to im
press their representatives in congress on
the necessity of pressing the hill which pro
vides for the extension of the bonded period
of two years on the present stock of
whiskies.
A vacancy will soon occur in the
grade of brigadier-generals by the retire
ment of Wesley Merritt, superintendent of
the West Point Academy. Col. D. S. Stan
ley of the Twenty-second infantry, is prora
Inently mentioned as likely to receive
promotion.
The Newark canal and j; Passaic and
Hackensack rivers are being dragged for
the body of Chas. Delmonlco, who has
been missing for more than a week.
The United States treasury gives no
tice that the principal and accrued interest
on $10,000,0(10 three per cent, bonds, issued
under the act approved July 12th, 1882,
will be paid ou the 15th of March, 1884.
The interest will cease on that day.
UK 0 f u u kcs, in Pennsylvania, is now in
The trial of James Nutt for the kill
progress at Pittsburg, having commenced
on the 14th.
The rush to the Guyotoa mines in
Arizona is unabated. Water is scarce and
selling at a dollar a barrel. Many new dis
coveries are reported, and the excitement
ts so great that the reports are not consid
ered reliable.
The powder magazine of the Corn
wall ore hills exploded at Lebanon, Pa.*
and a workman named Posey was blown to
atoms.
Reports from twenty-seven leading
clearing houses of tbe United States for tbe
week ended on the 12th, gives the total
clearances at $19,446,277.57, being an in
caease of 4.6 per cent, as compared with the
same period last year.
I seven construction trains will be put in
j operation over the Canada Southern division
' of the Michigan Central railroad for the pur
pose of double-tracking the line during the
ensuing year.
I S. A. Bridges died at Allentown, Pa.,
j on the 13tli, of dropsy, tiged 82. Ho was a
! member of congress from thcTenth district,
I during the years 1848 to 1855. anil 1876 to
1878.
Henry B. Payne was electeil United
States senator from Ohio without opposi
tion from the republicans, both houses vot
ing blank.
Application will be made to ( Canadian
parliament for an act to Incorporate a com
pany to construct a tunnel under the St.
Clair river, for railway purposes, from
•Sarltia to Port Huron.
The Baptist church at Port Norris,
N. J., burned. Two hundred children
were attending Sunday school at the time,
but were removed unharmed.
j : It is rtnnored on 'change in New
1 York that several large grain houses are In
a shaky condition, one of the principal ones
only putting up half margins.
At the republican caucus of tbe Iowa
legislature Senator Allison was renominated
for United States senator by acclamation.
Every republican member in the legislature
was present, and the nomination was made
amid great cnthusiasnl by a rising vote, and
i every one of the ninety-one members rose
, and voted In the affirmative. Prolonged and
repeated cheers greeted the result.
The statue to the late Senator Mor
ton was unveiled at Indianapolis on the 15th.
Senator Allison was renominated by
acclamation by the Iowa republicans.
The bill for an immediate appropria
; lion for Mississippi river Improvement gave
J rise to a heated discussion In the house.
j 4Fourteen people were injured on the
j Texas Pacific, near Weatherford Texas,
! caused by a broken rail.
! A consolidation will in all probability
be formally effected by the Hannibal and
Council Bluffs roads with tbe Burlington, ut
a meeting (or that purpose early in Febru
ary.
The Indiana republican state commit
tee has Issued a call fora contention to nom- 1
inatc candidates for state offices June 19th.
The state convention to elect delegatcs-at
large to the republican national convention |
will be held here June 17th. i
I
j
'
j
In the case of a man injured while )
traveling on a railroad on Sundav, the court
at „ostou ia * tructcd the jury that tbo plaiu . ,
.. . \ 3 , , V, , 1
tiff could not recover unless he should be
traveling on a mission of necessity orcharity*
Invitations have been sont out to all
i libera; members of commons, requesting
j their attendance at the opening of parlia
( ment, on the 5th of February.
A young man in jail at Anderson,
I Ind.. confined on the charge of bastardy,
nearly succeeded in killing his prosecutor,
.iThe plea of emotional insanity has
been entered by the defense in the Nutt
trial.
No action will he taken on the hills
relating to the Hennepin canal project until
printed reports of the engineer 'are re
ceived
A joint republican caucus of senators
and representatives was held to appoint
congressional campaign committee. More
than 125 members of congress were present.
Senator Edmunds presided. Senator Mil
ler, of Pennsylvania, was chosen secretary.
In taking the chair Senator Edmunds said
the outlook for the republican party for
1884 was at this early period of the cam
paign better titan at any time for the past
fifteen years.
The message of Governor Hale, of
Wyoming, compliments the people on their
prosperity, and predicts a great future.
He reviews the mineral and agricultural re
sources and advises a radical change in the
veterinary laws, in orderte prevent the In
troduction of contagious diseases among
horses.
received a
Pittsburg glass workers
proposition from Toledo capitalists to go to
that place, build a warehouse, and take a
large Interestin the concern.
The Bartholdi pedestal committee
announce the Travelers' Insurance com
pany, of Hartford, has subscribed $1,700
to the fund. A special engraving of the
statue complete will be prepared for the
American press.
CRIME.
A plot to burn the Forster grammar
school, at Somerville, Mass., was prevented
by the janitor extinguishing the fiâmes.
Several hundred children were in the build
ing at the time. The miscreant is unknown.
George Layhon and Lawrence May,
arrested for the murder of August Deltz,
at Rockway, N. J., August 11th. were found
guilty of manslaughter with recommenda
tions (or mercy,
INear the Colorado river, on the 12th,
the San Angelo stage, south-bound, was
halted by four mounted men, who robbed
tbe mail sacks and passengers, and then de
layed the coach until the arrival of the
north-bound stage, which was also plun
dered of a large quantity of registered mat
ter south-bound. It is believed that the
road agents made a heavy haul.
On the 13th, iu Alexandria, Ky., Miss
Weaver, affianced of Ed. Beicri went to
church with Nicholas Blehl. Beier became
so enraged that he went to the house of Miss
Weaver and demanded his presents. Be
ing refused, he put a pistol to Miss Weaver's
head and snapped it twice without shooting.
He then went to the back door and shot
himself twice, one of the halls penetrating
the heart.
John Flemming and Fred F. Loring,
of Chicago, convicted before Judge Blod
gett, of the United States district court, of
carrying on an extensive grain swindle under
the firm name of Fleming and Merriam, were
sentenced to twelve months in the county
jail and pay a fine $500. A writ of error was
granted in the case, however, by Judge
Drummond, of the circuit court, and the
prisoners released on $1,000 bail.
The case of Frank James, for the
Blue Cut train robbery was called In the
criminal court at Kansas City on the 14th,
and continued till Febuaryll, on account
of illness of the defendant. The ease of
Charles Ford on tbe same charge, was also
continued to February 11, It appears that
Ford Is at St. Louis, too 111 to attend.
John Elfers, who killed Bon. Hag
garty, because ho would not pay him four
bits, was hanged at San Franclcco On the
15th. '
John Kippe, a grain buyer at New
Alhla, Iowa, hung himself In his ware
house. He was aged 30 year.3. Whisky
and a love affair are supposed to have been
the cause.
While resisting arrest, W. A. Alex
ander, a cowboy and noted desperado, was
shot and killed on the reservation at Reno,
Colorado, by a detachment of soldiers. One
soldier was killed and two wounded.
Tommy G. Walker, aged 14, was
arrested at Boston for setting fire to a school
building. He had a mania for setting fires.
The dead body of Amelia Disen, 17
years old, was discovered on the open
prairie near the northwestern outskirts of
Chicago. The deceased was employed as a
domestic, and met her death while return
ing home after nightfall from her place of
work. There were evidences that she had
met with violence, and tbe belief Is enter
tained that she was choked and left insensi
ble, and died from the effect of her rough
usage, or that combined with the exposure
of the cold night. The locality through
which the girl passed is infested with a
rough class.
WASHINGTON.
The senate committee on foreign
relations, at a meeting on the 12lh, took
up a bill prepared by the Pacific coast del
égalions, Introduced in the senate by Senator
Miller (Cal.,) amending the Chinese Immi
gration act of the last congress. The dis
«»ssion was long and the hill was finallj
referred to a sub-committee consisting of
., M
debs Jgavc warrrot to the opinion that
measure for the purpose of correcting
the defects of the present law. and the pro-(
j bibitionof the importation
ers will be reported by the committee.
A call for $10,000,000 three per cent,
bonds was issued on the 12th.
An influential committee of local
lawyers waited on the president to urge the
re-appointment of District Attorney Cork
hill.
of ( hlnese labor
Sonor Romero, the Mexican minister,
paid on the 14th the eighth installment of
the indemnity due January 31. 1884. from
Mexico to the United States.
Follett, who will have charge of the
pension bill when it reaches the house, is
strongly in favor of abolishing entirely the
pension agency business.
Senator Beck has introduced . in the
senate a hill Identical with WtIUs' house hill
extending for two years the bonded period
on distilled spirits. Also a hill to provide
an act empowering the secretary of the
treasury to use the surplus In the treasury
for the ledempllon of United States bonds,
hut not to be construed to authorize him to
pay a premlum therefor.
In response to the house resolution
the secretary of the treasury has addressed
a letter to that body staling that the em
ployes from Indiana in his department t£
celved an assessment circular from the In
diana republican state central committee,
but he was unable to discover the person
who distributed them.
The live stock dealers, through Rep
resentative Hatch, submitted to the house a
petition asking for legislation to stamp out
pleuro-pueumonla by slaughtering all In
fected cattle and that government inspec
tion he made of all export meats at the ex
pense of the importer.
At a meeting of the house committee
on public lands consideration was given the
arguments of Pryor in favor of the Texas
Pacific land grant to the Southern Pacific.
The suh-commlttec in charge of the forfeit
ure of land grants to railroads directed the
report of a hill declaring the land grant for
feited .
The committee of ways and means
has decided not to act for the present on
Townshcnd's hill for the restriction of Im
portation of goods from governments which
prohibit the imports of American goods
It Is thought that the mere introduction
of the raeasuremlght have the desired ef
fect.
At a mooting of the senate committee
on public lands, Senator Van Wyck's bill
(or the relief of settlers on the public do
main in Nebraska and Kansas was ordered
reported favorably. It provides (or the pay
ment of $2.60 per acre to persons who took
up lands under the homestead or pre
emption laws within the limits of the North
ern Kansas land grant. With this sum, the
claimant Is expected to extinguish the title
of tbe company. Two hundred thousand
dollars was appropriated.
Robert Murray, nominated as sur
geon-general of the navy, has been con
j
a majority of the Spanish republicans I am |
favorable to free trade as a means of im
■ , -, .., , , ,, !
provl ig our relations with England, I ranee ;
and the United.States.
firmed.
The house committee on elections
has decided that the scat neither belongs to
Chalmers nor Manning on prima facie evi
dence.
The bill prepared by the cattlemen
for the extirpation of tbe lung plague was
submitted for consideration to the foil ow
ing members of the house commltt ee on
agriculture; Hatch, Debrell, Wlnans, Cul
len, Wilson and Ochiltree.
Senator Edmunds was present at the
meeting of the senate committee on post
offices and post roads, and gave his views
upon the points involved in the considera
tion of the postal telegraph bill, and enter
tained no doubt of the constitutional right
of the government to build telegraph lines,
but strongly opposed the purchase of exist
ing lines. The committee authorized the
chairman to fix an early day for persons
who represented the telegraph Interests to
be heard.
OTho senate session rejected the Mex
ican treaty. A motion was made to recon
sider, pending which the senate adjourned
until Monday. Beyond these facts the re
ports In regard to the matter are contradic
tory.
Senate confirmations : Elias Skinner,
Hanson, lewa; Commodore
postmaster,
Robert W. Slinfeldt, Alexandere. RUind,
and Thos. Pattison. rear admirals .
I The house committee on private laud
claims has unanimously agreed to report
favorably the bill for the relief of Myra
CIrrk Gaines. It provides for the issue of
patents for 38,457 acres of land on account
of grants made by Spain to John Lynd and
Thos. W. Quhart, provided that no mineral
lands are Included.
FOREIGN.
FRANCE.
Prime Minister Ferry received a
telegram from Tricon, French diplomatic
representative in Annam, in which he says:
"The king and members of the council ex
ercising regency formally received me to
day in a ceremony without precedent. It
was conducted with oriental pomp. After
salutations were exchanged the kingjrcqucst
cd me to approach, as he desired me to con
vey to the French government the assurance
of his respect and devotion. He expressed
the hope that the severity of the treaty stipu
lation between the two countries would be
mitigated. I assured the king of our sym
pathy and good will."
ENGLAND.
Castelar, in an interview, said :
'Like
i
lional friends. The English-speaking race
on both sides of the Atlantic have no better
friend than myself, even though I some
times have dissented from their foreign
policy. ' '
CHINA AND FRANCK.
A letter from Canton, dated Decem
her 5th, says that Chinais determined to
fight, and war can only be avoided by Franco
hacking clear down. France made an awful
muddle of the whole affair by not acting
with force at the first. Chinese troops arc
pouring in from the north and being raised
at Canton. The feeling Is that the Chinese
must rise up as one man and crush the pride
of the French, which they pronounce the
most trouolesnme nation that ever existed.
CONGRESSIONAL.
Senate —Monday, January 14.
senate proceeded to the election of president
f tro tem. and elected Senator Anthony, who,
n a few fitting words and with much feel
ing, declined the honor, owing to 111 health.
The question then arose whether the d( clj
nation of Anthony retained Edmunds as
president pro tempore w ithout further ac
tion, and after a debate It was decided to
avoid all doubt by a new election.
The following resolution was offered by Mr.
Sherman and agreed to: "Resolved, That
the secretary of the sennte inform the Vres
Ident of the United States and the bouse of
representatives that the senate has chosen
Hon. George F. Edmunds, senator from
Vermont, president pro tem. of the sen
ate." Senator Hill spoke on his postal tel
egraph hill. At the conclusion of Hill's
speech, messages from the president were
read transmitting the communications of the
secretary of the Interior submitting esti
mates from certain freedmett for lands of
the Oklahoma district, for the relief of the
mission Indians of California, and estimates
for $3,000 for (he survey of lands purchased
from the Creek Indians'for the Semlnoles.
-The
House.—B ills wore introduced : By
Mr. Belford, for the public welfare by se
curing reasonable rates of transportation on
railroads, aided by the Issue of United
States bonds. By Mr. Pusey, authorizing a
bridge across the Missouri river between
Council Bluffs and Omaha. By Mr. Wlnans,
to regulate transportation rates on railroads.
By Mr. Morgan, to abolish postage on
newspapers. By Mr. Flncrty, to reorgan
ize .he infantry regiments of the United
States; also to regulate promotion and in
crease the efficiency of the army. Mr.
Dunham, for the establishment of a de
partment of commerce; also, to authorize
he secretary of the treasury to
Issue 2'« per cent, fortv-years bonds.
By Mr. Ferrell, to protect American labor
from the effect of the Importation of foreign
labor in the contract system. By Mr.
Throckmorton, to appoint a delegate to
house of representatives from the Indian
territory. By Mr. Shaw, to repeal the civil
service act.
Senate— Tuesday, January 15 —Van
Wyck introduced a bill providing that rates
for the Union and Central Pacific roads be
reduced one-half the average rales existing
in 1882 and 1883, without regard to classifi
cation. Referred. A petition from the
citizens of Kansas was presented (or a con
stitutional amendment on woman suffrage;
also a petition from the citizens of Minneso
ta, praying that colonies of families be
allowed to lay out villages
public lands, in order Jo estab
lish co-operative Industrial societies.
Mr. Anthony's resolution offered some days
.ago regarding retaliatory legislation in the
United States to meet the exclusion of
American meats by foreign countries, was
brought up.
plred, the u
tbe
The morning hour having ex
Aftcr exeett
matter went over
House.—M r. Cobb, chairman of the
committee on public lands, reported a bill
declaring forteited certain grants of lands
made in certain states to aid
the construction of railroads,
ferred
Mr. King, chairman of the 'committee on
Mississippi levees, reported a bill to close
the gap of levees on tnc Mississippi river
and the improvement of navigation. It ap
propriates $1,000,000 to be expended in ac
cordance with tbe Mississippi river commis
sion. Referred to the committee on the
whole. Ms. Townsend introduced a bill
authorizing the president during a recess of
congress to prohibit any imports Injurious
to public health from countries which, on
the same ground, prohibit the importa
tion of American goods. Referred.
Mr. Wells chairman, of the committee on
rivers and harbors, reported a bill appropri
ating $1.000,000 for continuing the im
provement of tbe Mississippi
l erred to the committee of th
house then went into committee of the
whole (Cox, of Now' York, in the chair) for
Us consideration. Without action the com
mittee arose and the house adjourned
Further debate of the subject in the com
mittee will be limited to thirty minutes
4Senate.— Wednesday, Jan. 16.— The
senate, after slight amendment, adopted
the rules. Mr. Hoar called up his bill pro
vidlngfor the counting of the electoral vote,
being tbe same as the bill that passed the
senate of the Forty-seventh congress.
Mr. Miller (New' York) presented a
memorial from the committee of the na
tional stock convention, at Chicago, on the
subjectof European discrimination against
American cattle and meats, in connection
with the memorial, Miller presented
a bill which, he said, contained
the views of the cattle breeders' convention.
Mr. Plumb. by request, submitted a )< iat
resolution proposing an amendment to the
constitution prohibiting the manufacture
and sale of intoxicating liquors in theUnited
States. Referred. Mr. Beck submitted,
in order to be printed for consideration at
the proper time, an amendment to the reso
lution by At thony relating to European ex
clusion of American meat.
House.—M r. Hatch, chairman of the
committee on agriculture, reported a reso
lution requesting the president to transmit
to the house the correspondence had by the
state department with all foreign govern
ments on th« subject of the importation of
American bogs to their country. Mr. Nut
ting introduced a bill authorizing the con
struction of a ship canal around Niagara
(alls. Referred. Mr. Lamb, of the com
mittee on foreign affairs, reported a resolu
tion calling upon the secretary of state for
Information concerning the alleged arrest,
imprisonment and torture of E wheelock, a
citizen of the United States, by the govern
ment of Venezuela, In 1879. Adopted.
Mr. Cosgrove, from the committee on
postoffiees and post roads, reported a bill
to provide for a more speedy delivery of
letters from delivery offices. Placed upon
the bouse calendar.' The house went into
committee of the whole, (C'ox. of New
York, In the chair ) on the senate bill ap
propriating $1,U0",000 for continuing the
work on the Mississippi river, and without
action adjourned.
Senate. —Thursday, January 17.—
The chair laid before the senate a memorial
rom " m. Pitt Kellogg, denying all impu
talions against him contained in the recent
documents transmitted to the senate by the
Re
to committee of the whole.
1 river. Re
e whole. The
UUUUUICUia UailDUilllGU IW tut nenmo O» mu
secretary of the interior relating to the '
transferor the land grant of the Texas and
Pacific to the Southern Pacific, and asking
for an investigation. Van Wyck introduced
a bill to secure reasonable rates of trans
portation over the railroads aided by the
government. He said he introduced it as a
substitute for a similar one, which only cov
ered the Union and Central Pacific.
House.—T he senate bill appropriat
ing $1,000.000 for the improvement of the
Mississippi river was passed; yeas, 215;
nays, 84. The house bill of similar title and
import was laid upon the table. The bill
enabling the United States courts to nullify
patents fraudulently secured was passed.
The bill making all public roads and high
ways post routes, was also passed. Mr.
Springer, chairman of the committee on ex
penditures for the department of Justice,
reported back the resolution calling upon
the postmaster-general for the correspond
ence concerning frauds in star-routes.
Adopted. •
Senate.—F riday January 18.—The
chair laid before the senate a message from
the president transmitting for consideration
of congress communications from the
rctarles of war and navy on flic subject of
relief for the expedition of the Greely party,
and recommending Immediate action, as the
situation of the party Is most perilous; also
the correspondence relating to the execu
tion of the Chinese exclusion act. asked for
by the senate. Mr. Miller (Cal.), from the
committee on foreign relations, reported
favorably the bill for a supplemental com
mercial treaty with China, prohibiting tbe
importation and exportation of opium.
House.— Mr. Brumm offered a réso
lution. which was referred to the committee
on foreign affairs, instructing the committee
to make inquiry whether any foreign minis
ter accredited to Hie United Slates and en
deavored to nullify tbe effects of the unan
imous resolution of the house
by representatives reflecting upon
the honor and Integrity of its members.
The house then went into committee of the
whole on the private calendar, Mr. Springer
in the chair, the first hill being the relief of
Fits John Porter. Speeches were made on
both sides of tbe question, but, without ac
tion, the house adjourned.
■c
DISASTER AT SEA.
The si mi
ir City
ih » Ledge at Devil*« Bridge—Over One
Hundred Live« Frobubly Lent.
►f Coin mbit« Strike«
Boston, January 18.- F. W. Nicker
son A Son, agents of the Savannah steam
ship line, received the following dispatch:
New Bedford, Mass., January 18.—To
E. W. Nickerson & .Son:—The steamer City
of Columbus Is ashore on Devil's Bridge,
Gay Head, and is fast hroakli g up. About
a hundred lives arc lost. Will leave on tbe
early train in the morning. 1 was saved bjr
the cutter Dexter.
(Signed)
The City of Columbus left Boston at 3 p.
m. on the 17th (Thursday), carrying eighty
passengers and a crew of 450. At 3:45 a. m.
on Friday, at Gay Head light, she was bear
ing south, half east. The vessel struck on
the outside of t .e Devil's Bridge buoy.
The wind was blowing a gale west by north.
The vessel immediately filled and keeled
over, the water breaking in and flooding the
port side saloon. All the passengers, ex
cepting a few women and children, came on
deck, nearly all wearing life preservers. All
the boats were cleared away, but wore im
mediately swamped. A majority of the
passengers were washed overboard. Seven
passengers left the vessel on tbe life raft,
S. E. Wright, Master.
and about more rigging.
10:20 a. ra. the Gay Head life boat put off
and took seven persons. Another life boat,
put off between 12 and 1 o'clock, and the
along
The
revenue cutter, Dexter, came
about 2:30 and sent off two boats,
ledges on which the City of Columbus struck
are considered by manners one of the most
dangerous points on the coast. The ledges
consist of a formation of submerged rocks.
constituting a double ledge or outer strata,
which is called the "Devil's Back," both
ledges being called the "Devil's Bridge."
These ledges are abreast of Gay Headlight
on the mainland and extend a little south
ward of it. The outer ledge of Devil's
Back is about eight miles from tbe mainland
on either side at the outer ledge Is very deep
water. The upper part of the ledge is
formed like the gable of a bouse, so that a
vessel striking it diagonally would naturally
keel over onto her beam ends. Tbe course
of vessels going around Gay Head is to pass
by the outer ledge on the south.
The total number of persons saved is two
hundred and thirty, and five dead bodies
were recovered, and one hundred and nine
teen souls were unaccounted for.
John r L. Cook, one of the passengers
sayed, relates a heart-rending scene; John
Roach, a coal passer, dangled from the
main mast for two hours with his hands and
At length his
legs about the main stay,
struggles grew feebler, until he dropped
into the »ea. A passenger was astride the
stay and clung there from 5 until nearly 10
a. m., when he too relinquished his fight
for life and fell into the ocean.. All those
rescued gave the highest praise to the ufli
eers of the revenue cutter for the bravery
they manifested in saving them from the
All the survivors now aboard the
for by the offi
wreck,
cutter are being .cared
cers.
The City of Columbus was one of the
She was built iu
finest vessels on the coast.
1838 by John Roach A Son, of Chester, Pa.,
for the Ocean Steamship company, of Now
York, to run between that port and Havana.
She was purchased by the Boston and Sa
vannah Steamship company iu September,
1880, and has since been plying between this
city and Savannah, making fortnightly trips
alte roatlon with her sister ship, the Gate
The Columbus was built of iron and
was rated A 1
in
ity.
thoroughly equipped. She
for a hundred years and was of 1,997 tons
burden. She wtts 270 feet long and 30 feet
beam and bad passenger accommodation for,
84 first-class and 40 second-class passengers.
The steamship was insured at a lower rale
than any vessel on the coast and was valued
$300,000, and insured for $250,900; $170,
000 in English and $80,000 iu American
• ii
companies.
How many creditors miss their dues
111*.*« • in
when nature s debt is paid.
/
xml | txt | https://chroniclingamerica.loc.gov/lccn/sn86091092/1884-01-26/ed-1/seq-1/ocr/ | CC-MAIN-2022-05 | refinedweb | 5,208 | 62.17 |
I'm currently setting up a fairly complex bash configuration which shall be used on multiple machines. I try to find out if it is possible to determine whether I'm logged in via SSH or on a local machine. This way I could, for instance, set some aliases depending on that fact. Like aliasing halt to restart since stopping a remote server might not be the best thing to do.
halt
restart
What I know so far is, that the environment variable SSH_CLIENT is set when I logged in via ssh. Unfortunately, this variable is discarded when I start a super user shell with sudo -s. I also know that I can pass a parameter to sudo that instructs sudo to copy all my environment variables to the new shell environment, but if I don't want to do this, is there an other way?
SSH_CLIENT
sudo -s
You could use "w" or "who" command output. When you connect over ssh, they'll show your source IP.
ps afx
ps
who am i
hostname=$(who am i | cut -f2 -d\( | cut -f1 -d:)
who am i | sed 's/.*(\(.*\))/\1/g'
If you want to know if you bash shell is directly a child process of sshd (not n>1 layers deep) you can
cat /proc/$PPID/status | head -1 | cut -f2
cat /proc/$PPID/status | head -1 | cut -f2
it should give you sshd or whatever is the parent process name of your current shell.
sshd
You could add SSH_* to env_keep in sudoers so that this can be detected while switched to the other user.
SSH_*
env_keep
sudoers
I think you want to rethink the way you're thinking of the problem. The question isn't "am I logged in via SSH, because I want to turn off certain commands." It's "am I logged in at the console, because then I will enable certain commands."
Look for your shell's parent cmdline and recurse. Maybe something like the following:
#!/usr/bin/env bash
## Find out how I'm logged in
# Tested on RHEL5.5
PD=${1:-$$}
ME=`basename $0`
## Read the shell's PPID
PAR=`ps --no-headers -p $PD -o ppid`
## CMDLINE can contain stuff like the following:
# /sbin/getty-838400tty4 // logged in at a console
# gnome-terminal // logged in Gnome Terminal
# -bash // in a subshell
# su- // we became another user using su
# sshd: jc@pts/1 // logged in over ssh
# login // logged in terminal or serial device
eval `python - << __EOF__
import re
f = open("/proc/${PAR}/cmdline", 'r')
ln = f.readline()
if re.search(r'^ssh', ln):
print "echo Logged in via ssh"
if re.search(r'getty.*?tty', ln):
print "echo Logged in console"
if re.search("gnome-terminal", ln):
print "echo Logged in Gnome"
if re.search(r'^login', ln):
print "echo Logged in console"
if re.search(r'^-?bash', ln) or re.search(r'^su', ln):
print "./$ME $PAR"
f.close()
__EOF__
`
Edited to make it actually work :)
I've came up with the following, based on tips from others here:
[ -n "$SSH_TTY" ] || [ "$(who am i | cut -f2 -d\( | cut -f1 -d:)" != "" ]
Yes, as others noted, the info is in the presence of your IP in parentheses in the output of who am i.
You can use Bash regular expressions to detect it:
if [[ $(who am i) =~ \([0-9\.]+\)$ ]]; then echo SSH; else echo no;99 times
active
6 months ago | http://serverfault.com/questions/187712/how-to-determine-if-im-logged-in-via-ssh | crawl-003 | refinedweb | 569 | 71.65 |
stylo: test
_namespace _rule .html fails for several cases
RESOLVED FIXED in Firefox 56
Status
()
P2
normal
People
(Reporter: xidorn, Assigned: xidorn)
Tracking
(Blocks: 1 bug)
Firefox Tracking Flags
(firefox56 fixed)
Details
Attachments
(2 attachments)
There are 17 failures in layout/layout/test/test_namespace_rule.html, and it seems to me they are probably just bugs in how Servo handles @namespace rule, and this test doesn't use much the CSSOM namespace rule.
Priority: -- → P2
Assignee: nobody → ferjmoreno
Assignee: ferjmoreno → nobody
Created attachment 8885133 [details] testcase So there are 6 failures left, which all have empty namespace uri involves. See the testcase. I suspect these failures are all because we use empty atom for kNameSpaceID_None rather than an independent namespace as mentioned in comment inside nsNameSpaceManager.h [1] and bug 1292278 comment? [1]
Flags: needinfo?(bobbyholley)
> we use empty atom for kNameSpaceID_None rather than an independent namespace I believe this behavior is correct per spec. In every entry point (the XML parser, DOM APIs, CSS @namespace rules), using an empty string for a namespace URL is either an error or is normalized to "no namespace" a.k.a. "the null namespace". Relevant to the attached test case: > If there is no default namespace declaration in scope, the namespace name has no value. > […] > The attribute value in a default namespace declaration MAY be empty. This has the same effect, within the scope of the declaration, of there being no default namespace. > In CSS Namespaces a namespace name consisting of the empty string is taken to represent the null namespace or lack of a namespace.
So, that may be correct per spec, and in the attached testcase, replacing "foo|test" with "|test" also gets a green rect in all browsers, which indicates that "foo" namespace really just means null namespace in browsers. If empty string in xmlns attribute also indicates null namespace, I guess getting a green rect from the test case is the correct behavior per spec.
(In reply to Xidorn Quan [:xidorn] UTC+10 from comment ? If all the UAs currently agree and disagree with the spec, then nobody has demonstrated that the spec is web-compatible, and stylo is not the right vehicle to do that. Let's align with Gecko and the other UAs.
Flags: needinfo?(bobbyholley)
(In reply to Bobby Holley (:bholley) (busy with Stylo) from comment #4) > If all the UAs currently agree and disagree with the spec To be precise, I meant "agree with each other and jointly disagree with the spec".
That is not the case here as far as I can tell. There is interop and implementations agree with the spec and this test. It’s only Stylo that has a bug somewhere, but I believe that bug is not "empty string namespace should be distinguished from no namespace" as Xidorn suggested earlier.
All the UAs agree with each other an with the spec here. "" means null namespace. Per spec, the square in should be green, and is in browsers. It's possible the problem lies in whatever code maps servo namespace strings to Gecko namespace ids?
Note that servo gets a green square here too, so this is really pretty stylo-specific.
I know what's going on...
Assignee: nobody → xidorn+moz
Comment on attachment 8886481 [details] Bug 1355715 - Use empty atom rather than 'empty' atom for none namespace. Doh! r=me
Attachment #8886481 - Flags: review?(bobbyholley) → review+
Pushed by [email protected]: Use empty atom rather than 'empty' atom for none namespace. r=bholley
Status: NEW → RESOLVED
Last Resolved: a year ago
status-firefox56: --- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla56 | https://bugzilla.mozilla.org/show_bug.cgi?id=1355715 | CC-MAIN-2018-47 | refinedweb | 598 | 63.29 |
One of the challenges one faces when doing multi language support in WPF is when one has several projects in one solution (i.e. a business layer & ui layer) and you want multi language support. Typically each solution would have a resource file – meaning if you have 3 projects in a solution you will have 3 resource files.
Image1
For me this isn’t an ideal solution, as you normally want to send the resource files to a translator and the more resource files you have, the more fragmented the dictionary will be and the more complicated it will be for the translator.
This can easily be overcome by creating a single project that just holds your translation resources and then exposing it to the other projects as a reference as explained in the following steps.
Step 1
Step 1 - Add a class library to your solution that will contain just the resource files.
Image2
Your solution will now have an additional project as illustrated below.
Image3
Step 2
Reference this project to the other projects.
Image4
Step 3
Move all the resources from the other resource files to the translation projects resource file.
Step 4
Set the translations projects resource files access modifier to public.
Image5
Step 5
Reference all other projects to use the translation resource file instead of their local resource file.
To do this in xaml you would need to expose the project as a namespace at the top of the xaml file… note that the example below is for a project called MaxCutLanguages – you need to put the correct project name in its place.
xmlns:MaxCutLanguages=”clr-namespace:MaxCutLanguages;assembly=MaxCutLanguages”
And then in the actual xaml you need to replace any text with a reference to the resource file.
End Result
You can now delete all the resource files in the other projects as you now have one centralized resource file. | http://blog.markpearl.co.za/Centralizing-a-resource-file-among-multiple-projects-in-one-solution-(C-and-WPF) | CC-MAIN-2019-04 | refinedweb | 317 | 65.96 |
So I've had it in the back of my mind to give GLFW another look at some point for possible inclusion into Derelict 2. Today, I did. A new version was released late last year (2.7) and a new branch that streamlines the API (3.0) has been started. I really like the new branch. So, being the spontaneous sort of fellow I am, I decided I wanted a binding to it. I knocked one up in just over 30 minutes. It's now sitting in my local "scratch" copy of Derelict, waiting to be compiled and tested. Given that it's 1:30 am as I type this, I don't think I'm going to get to it just yet. Tomorrow for sure.
I won't be adding this new binding to the Derelict repository just yet. GLFW 3.0 is still in development. So, just as with the binding I've begun for SDL 1.3 (which will become SDL 2 on release), I'll wait until the C library is nearing a stable release before I check it in.
Making D bindings to C libraries is not a difficult thing to do. It's just tedious if you do it manually, like I do. I have a system I've grown used to now that I've done so many of them. It goes reasonably quick for me. Some people have experimented with automating the process, with mixed results. There are always gotchas that need to be manually massaged, and they might not be easily caught if the whole process is automated. One example is bitfields.
D doesn't support bitfields at the language level. There is a library solution, a template mixin, that Andrei Alexandrescu implemented in the std.bitmanip module. I don't know how compatible it is with C. I've only had to deal with the issue once, when binding to SDL 1.2, but that was before the std.bitmanip implementation. Besides, it's a D2 only solution and Derelict has to be compatible with both D1 and D2. So what I did was to declare a single integer value of the appropriate size as a place holder. The bits can be pulled out manually if you know the order they are in on the C side. I could have gone further by adding properties to pull out the appropriate bits, but I never did the research into how different C compilers order the bitfields on different platforms.
Another issue that crops up is dealing with C strings. For the most part, it's not a problem, but if you are new to D it's a big gotcha. Like C strings, D strings are arrays of chars (or wchars or dchars as the case may be). But, char strings in D are 8-bit unicode by default. Furthermore, D arrays are more than just a block of memory filled with array values. Each array is conceptually a struct with length and ptr fields. Finally, and this is the big one, D strings are not zero terminated unless they are literals. Zero-terminated string literals are a convenience for passing strings directly to C functions. Given a C function prototype that takes a char*, you can do this:
someCFunc("This D string literal will be zero-terminated and the compiler will do the right thing and pass the .ptr property");
If you aren't dealing with string literals, you need to zero-terminate the string yourself. But there's a library function that can do that for you:
import std.string;
// the normal way
someCFunc(toStringz(someString));
// or using the Universal Function Call Syntax, which currently only works with D arrays
someCFunc(someString.toStringz());
A lot of D users like the Universal Function Call Syntax and would like to see it work with more types instead of just arrays. Personally, I'm ambivalent. The way it works is that any free function that takes an array as the first argument can be called as if it were a member function of the array.
Going from the C side to the D side, you would use the 'to' template in std.conv:
// with the auto keyword, I don't need to declare a char* variable. The compiler will figure out the type for me.
auto cstr = someCFuncThatReturnsACharPtr();
// convert to a D string
auto dstr = to!(string)(cstr);
// templates with one type parameter can be called with no parentheses. So for to, this form is more common.
auto dstr = to!string(cstr);
Another gotcha for new users is what to do with C longs. The D equivalent of nearly all the C integral and floating point types can be used without problem. The exceptions are long and unsigned long. D's long and ulong types are always 64-bit, regardless of platform. When I initially implemented Derelict, I didn't account for this. D2 provides the aliases c_long and c_ulong in core.stdc.config to help get around this issue. They will be the right size on each platform. So if you see 'long' in a C header, the D side needs to declare 'c_long'. I still need to go through a few more Derelict packages to make sure they are used.
The issues that crop up when actually implementing the binding aren't so frequent and are easily dealt with. Sometimes, though, you run into problems when compiling or running applications that bind to C.
D applications can link directly to C libraries without problems, as long as the object format is supported by the compiler. On Linux, this is never an issue. Both DMD and GDC can link with elf objects. Problems arise on Windows, however. The linker DMD uses, OPTLINK, is ancient. It only supports OMF object files, while many libraries are compiled as COFF objects. If you have the source code and you can get it to compile with Digital Mars C++, then you're good to go. Otherwise, you have to use the DigitalMars tool coff2omf, which comes as part of the Digital Mars Extended Utilities Package. Cheap, but not free. Then you still might face the problem that the COFF format output by recent versions of Visual Studio causes the tool to choke. There are other options, but it's all nonsense to me. That's one of the reasons when I made Derelict I decided that it would only bind to libraries that come in shared form and they will be loaded manually. Problem solved. But there are other issues.
In a past update to DMD (not sure which), the flag '--export-dynamic' was added to the DMD config file (sc.ini) on Linux. So that means that every binary you build on Linux systems with DMD has that flag passed automatically to gcc, the backend DMD uses on Linux. Normally, not an issue. Until you try to build a Derelict app. The problem is that Derelict's function pointers are all named the same as the functions in the shared library being bound to. This causes conflicts when the app is built with --export-dynamic on Linux, but they don't manifest until run time in the form of a segfault. Removing the flag from sc.ini solves the problem. One of these days I need to ask on the D newsgroup what the deal with that is.
I know all of this could sound highly negative, giving the impression it's not worth the hassle. But, seriously, that's not the case. I have been maintaining Derelict for seven years now. Many bindings have come and gone. Version 2 currently supports both D1 and D2, as well as the Phobos standard library and the community-driven alternative, Tango. I can say with confidence that D works very well with C the large majority of the time. And for anyone planning to use D to make games, you will need to use C bindings at some level (Derelict is a good place to start!). As for binding with C++... well, that's another story that someone else will have to tell.
- Read more...
- 6 comments
- 2302 views | https://www.gamedev.net/blogs/blog/1140-d-bits/?page=1&sortby=entry_num_comments&sortdirection=desc | CC-MAIN-2018-09 | refinedweb | 1,364 | 74.9 |
Just as any other data adapter class, OleDbDataAdapter provides a Fill method that callers can use to add tables to a DataSet object. The OleDbDataAdapter class is defined in the System.Data.OleDb namespace, and you typically use it as follows:
OleDbDataAdapter da = new OleDbDataAdapter();
DataSet ds = new DataSet(); da.Fill(ds, "MyTable");DataSet ds = new DataSet(); da.Fill(ds, "MyTable");
The Fill method of the OleDbDataAdapter class has an interesting peculiarity. It can load into the target DataSet object not only the result sets created by the given command but also the contents of an existing ADO Recordset object. One of the overloads of the Fill method has the structure shown here:
int Fill(DataSet ds, Object adodb, String srcTable);
The data adapter copies each result set found in the ADO Recordset object into a table in the DataSet. As usual, the first table is named after the specified srcTable argument, and all the other result sets add a progressive index to the table name: MyTable1, MyTable2, and so on. The link between ADO and ADO.NET is a one-way binding that you can use only to copy data from ADO to the DataSet object. Any updates are performed by ADO.NET.
The Fill method loads the Recordset object into a given table in the specified DataSet object, regardless of how you actually obtained the Recordset object. The following code shows how to retrieve the Recordset object by using plain ADO code, load it into a DataSet object, and then bind it to a CheckBoxList control. The output is shown in Figure 8-3.
Recordset adoRS = new Recordset(); adoRS.Open(strCommand, strConn, CursorTypeEnum.adOpenForwardOnly, LockTypeEnum.adLockReadOnly, 1); // Transforms a Recordset into a DataSet using the OleDbDataAdapter OleDbDataAdapter oda = new OleDbDataAdapter(); DataSet ds = new DataSet(); oda.Fill(ds, adoRS, "MyTable"); // Bind to the CheckBoxList control chkOutput.DataSource = ds.Tables["MyTable"]; chkOutput.DataTextField = "lastname"; chkOutput.DataValueField = "employeeid"; chkOutput.DataBind();
The signature of the Fill method mandates that you specify the name of the table. When a table with the name you specified already exists in the target DataSet object, the contents of both tables are merged. If primary key information is present, duplicate rows are refreshed and appear only once. When creating a table, the Fill operation creates primary keys and constraints if the MissingSchemaAction property for the data adapter is set to AddWithKey. Filling a DataTable object with the contents of a Recordset object is just a special case of a more general filling mechanism.
Typically, you call Fill on an open Recordset object. If the Recordset object is closed prior to the beginning of the operation, no exception is thrown. The data adapter assumes that the current result set in the Recordset object is closed and attempts to move to the next one. If another result set is found, the execution continues; otherwise the method terminates.
When the Fill method terminates, all the result sets have been successfully loaded in the target DataSet object, and the original Recordset object is closed. The Fill method returns the number of rows successfully added to or refreshed in the DataSet object.
The Fill method can load the contents of the Recordset object directly into a DataTable object. The DataTable object might or might not be a member of a DataSet object. The signature of this overload is slightly different.
int Fill(DataTable dt, Object adodb);
When this implementation of the Fill method terminates, the input Recordset object is left open. In addition, when handling batch SQL statements that return multiple results, this implementation of the Fill method retrieves schema information only for the first result.
Loading the contents of a Recordset object into a DataTable object provides you with a bit more flexibility because you can handle the various result sets separately. You can decide whether to load them all or discard a few. You are responsible for closing the Recordset object when finished.
Being able to pump data out of an ADO Recordset object into an ADO.NET object is a big step forward on the road to integration between ADO and ADO.NET systems. Let’s review a common migration scenario and see how you can elegantly connect new ASP.NET pages to the existing middle tier and data back-end. You begin the renovation of the Web application with the presentation layer, then handle the more problematic changes to the middle tier. Let’s assume that, as shown in Figure 8-4, you have a typical Windows DNA 2000 system made of ASP pages populated by a middle tier that uses ADO for all data access needs.
Well written ASP pages do not include plain ADO code but rather call into COM objects, usually Visual Basic COM objects. Such middle tier and data tier components can have methods that return only ADO Recordset objects, which the page uses to fill out the HTML user interface.
From the .NET perspective, calling into the ADO library is not really different from calling into any other COM object. CIS is involved and you jump out of the CLR. You can make a Visual Basic COM object callable from within an ASP.NET page by following the instructions discussed earlier in this chapter.
Let’s suppose that you have a COM component whose progID is MyComp.NWData, which you use for data access. This object has one method with the following signature and code:
Public Function GetEmployees() As ADODB.Recordset Dim strCnn As String Dim strCmd As String strCnn = "PROVIDER=sqloledb;UID=sa; DATABASE=northwind;" strCmd = "SELECT firstname, lastname, title FROM Employees" Dim rs As New ADODB.Recordset rs.Open strCmd, strCnn, adOpenForwardOnly, adLockReadOnly, 1 Set GetEmployees = rs End Function
How do you use this code to retrieve a Recordset object from within an ASP.NET page? You get a .NET wrapper for the library and link the resulting assembly to the page. Visual Studio .NET can do this for you; otherwise, you use the tlbimp.exe utility.
Let’s assume that you successfully created a .NET assembly called MyComp.dll from the original Visual Basic COM DLL. You link this assembly to the ASP.NET page by using the following directives.
<%@ Assembly Name="MyComp" %> <%@ Assembly Name="ADODB" %>
Both the assemblies must be visible to the ASP.NET run time and placed, for example, in the bin directory of the application. Since the method of the MyComp component returns an ADO Recordset object, you must also link with ADODB if you want your code to remain strongly typed. Figure 8-5 shows how the Intermediate Language Disassembler (ILDASM) reads the .NET version of the MyComp component.
What happens from now on in the migration scenario is not really different from what we discussed earlier for plain ADO code. You create a new instance of the .NET class that represents the VB COM object. Next you invoke the needed method and get an ADO Recordset object.
MyComp.NWData mc = new MyComp.NWData(); ADODB._Recordset rs = mc.GetEmployees();
The OleDbDataAdapter class lets you convert this recordset into a DataTable object, making it bindable to any data bound ASP.NET control—for example, the DataGrid control.
OleDbDataAdapter da = new OleDbDataAdapter(); DataSet ds = new DataSet(); da.Fill(ds, rs, "MyTable"); grid.DataSource = ds.Tables["MyTable"]; grid.DataBind();
Figure 8-6 shows this technique in action in the AdoAdapterFromCOM.aspx sample application.
Most COM components work if invoked from within ASP.NET pages, even when you call them by using the late-bound Server.CreateObject method—the typical way of creating object instances in ASP. By creating a .NET wrapper, you also improve performance because the component is early bound to the managed code of the page.
Most of the COM components developed using Visual Basic 6 work in the single-threaded apartment (STA). If you plan to use these components from within an ASP.NET page, you should mark the ASP.NET page with the aspcompat attribute:
<%@ Page aspcompat="true" %>
The aspcompat attribute forces the page to be executed on an STA thread. In this way, the page can call STA components, such as those developed with Visual Basic 6. Calling into STA components is not as fast as calling into native .NET components or imported free-threaded components.
If you omit setting the aspcompat attribute and call STA objects, the run time is expected to detect the situation and throw an exception. However, an exception will not be thrown when you use the STA component through the services of a .NET assembly. The ASP.NET run time has no way to detect that it is going to make STA calls because the calls are not direct and are processed by an intermediate proxy. As a result, your application is walking a tightrope suspended between poor average performance and lethal deadlocks.
The range of the COM components you can call from within ASP.NET pages includes Visual Basic COM components registered with COM+ applications. Setting the aspcompat attribute to true also enables you to access the ObjectContext object so that you can interact with the COM+ run time. | http://etutorials.org/Programming/Web+Solutions+based+on+ASP.NET+and+ADO.NET/Part+III+Interoperability/Interoperable+Web+Applications/Adapting+Recordset+Objects+to+DataSet+Objects/ | crawl-001 | refinedweb | 1,512 | 56.86 |
Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
By SWATI GOENKA
(08BSHYD0871)
RELIANCE INDUSTRIES LIMITED
1
A REPORT ON PROJECT FINANCING OF RPL AND PRE AND POST MERGER VALUATION OF RIL - RPL
By SWATI GOENKA
(08BSHYD0871)
ICFAI BUSINESS SCHOOL
MAY 15, 2009
2
AUTHORISATION
This report has been authorized by PROF AJIT PATIL as a part of the evaluation for SUMMER INTERNSHIP PROGRAM. This report has been submitted as a part of partial fulfillment of the requirements of the MBA program of ICFAI Business School.
Authorizing Person (Prof. Ajit Patil)
Date: 15/5/2009
3
ACKNOWLEDMENT
I would like to take this opportunity to thank all those who have made working on this project feasible for me. I would first like to thank „Reliance Industries Limited’ for providing me with the opportunity to work with them and giving me my first taste of the real corporate and professional world. It gave me an opportunity to understand the real life situations and implement all those things which I had earlier only come across in textbooks as part of my course. I would also like to extend my sincere gratitude to my guides, Mr. K.R. Raja and Mr. Hariharan Mahadevan for allowing me to work under their able guidance. Without their guidance, help and support this project would not have been possible. I extend a special thanks to Mr. Ritesh, for helping us thoroughly in the day to day working and the project. Also, I would like to thank my faculty guide Prof. Ajit Patil for his able guidance and support. Last but not the least; I would like to thank my colleague Tanya Bhardwaj (Inte rn), without whose support, co-operation and suggestions, this project could not be completed.
4
TABLE OF CONTENTS PARTICULARS
Authorization Acknowledgement Abstract 1. Introduction 1.1 Purpose and Scope 1.2 Limitations 1.3 Methodology 2. Industry Overview 3. Business Overview 4. Long Term Sources of Finance 4.1 Capital Market 4.2 Different Kinds of Equity Issue 5. Initial Public Offer 5.1 Intermediaries Involved in IPO 5.2 Considerations Before Deciding for An IPO 6. IPO by an Unlisted Company 7. Pre Issue Obligations 8. Terms of Issue 9. Pricing by Companies Issuing Securities 10. Promoter’s Contribution & Lock In Period 11. The Issue 12. Objects of the Issue 13. Basis for Issue Price 14. SEZ & Tax Benefits 15. Stock Movement in 2006 16. External Commercial Borrowings (ECB) 17. Eligible Borrowers 18. Recognized Lenders 19. Average Maturities for ECB 20. RPL Debt 21. Interpretations 22. Portfolio Tracker Version 0.0 23. Merger of RPL with RIL
PAGE NO.
3 4 9 12 13 15 16 17 22 25 26 27 29 31 33 34 36 37 38 41 44 46 48 51 52 58 59 60 61 62 63 83 88
5
24.
25. 26. 27. 28. 29.
23.1 Synergy of merger 23.2 Swap Ratio 23.3 Investors Position 23.4 Stock Position 23.5 Analysts take on the merger Valuation of RIL and RPL 24.1 Valuation of RIL 24.2 Valuation of RPL 24.3 Study of stock prices of RIL & RPL 24.4 Weighted Average Cost of Capital 24.5 Free Cash Flow to Equity 24.6 Free Cash Flow to Firm Findings Conclusion Recommendations Declaration References
91 92 95 97 99 104 116 121 124 127 130 132 133 135 136 137
6
LIST OF ILLUSTRATIONS
LIST OF TABLES
PARTICULARS Table 1: World Refining Capacity in MMBD Table 2: World demand growth in MMBD Table 3: GDP Billion US$ (ON PPP BASIS) Table 4: Proposed Funding for RPL Table 5: Lead Managers to size of issue Table 6: Pre Issue Shareholdings Table 7: Capital Structure Table 8: Issue Details Table 9: Estimated Expenses Table 10: Estimated Expenses of Issue Table 11: Comparison with Domestic Peers Table 12: Debt Raised by RPL Table 13: Dilution of Promoters Table 14: Swap ratio
PAGE NO. 19 19 20 23 36 41 42 44 46 47 49 62 93 96
7
LIST OF FIGURES
PARTICULARS Figure 1: Refining Requirement Forecast Figure 2: Different Kind of Equity Issue Figure 3: Book Building Process Figure 4: MRPL Stock Movement Figure 5: BPCL Stock Movement Figure 6: HPCL Stock Movement Figure 7: RIL Stock Movement Figure 8: RPL Stock Movement Figure 9: RPL Details Figure 10: Stock Position of RIL Figure 11: Stock Position of RPL Figure 12: Operating Profit per Share Figure 13: Percentage Growth Figure 14: Book Value per Share Figure 15: Gross Profit Margin Figure 16: Gross and Net Profit Margin Figure 17: Return on Net Worth
PAGE NO. 21 27 40 52 53 53 54 55 56 97 98 105 105 106 107 108 108
8
ABSTRACT
Refinery project requires huge investments for the setting up the refining plant. Hence, the long term sources of finance like raising funds through equity shares and raising long term secured debt are more viable. Reliance Petroleum Limited (RPL) to fund its operations opted for long term sources of funds namely equity and debt. The report aims at understanding various guidelines and processes involved for raising the long term funds through a case study of RPL. The capital cost of RPL‟s project was estimated at Rs. 270 billion. The project was funded through debt (Rs. 157.5 billion) and equity (Rs 112.5 billion). As one of the means for raising Equity funds, RPL went for an Initial Public Offer (IPO) through the book building process. The Company issuing Equity through IPO has to fulfill certain guidelines (rules and regulation) issued by SEBI under Section 11 of the Securities and Exchange Board of India Act, 1992. The guidelines are called Disclosure and Investor Protection Guidelines (DIP). To raise funds through debt (External Commercial Borrowings in case of RPL) guidelines of the Reserve Bank of India for External Commercial Borrowings need to be complied with. ECB can be accessed under two routes. They are – Automatic Route and Approval Route. RPL raised debt through Approval Route for a new project having a term o f 9 years and 7 months.. Apart from long term sources of finance, the report explains the functionality of a software named Portfolio Tracker Version 0.0. Portfolio Tracker.
9
In the second part of the report, an attempt has been made to understand the facts of recent RIL – RPL amalgamation, which was announced on 27 th February 2009 and took place on 1st April, 2009. The report discusses the swap ratio of 1:16 that has been decided by the Board of Directors of RIL for the merger. This ratio is marginally in favor of the shareholders of RPL, would mean a dilution of 4.4% of RIL‟s equity. Moreover, the position of the shareholders of the two companies has also been looked at and the factors which can affect the shareholder‟s investments decisions have also been analyzed. RIL will be benefitted from certain financial and operational synergies arising out of the merger of RPL with RIL.
To understand the probable course of action for investors, the valuation of two companies has been done. This facilitates the investors in making the decision for investment in RIL after the amalgamation of RPL. The financial ratios of RIL state that the company is in good financial health. It shows Earnings per Share of Rs. 133 with a Dividend per Share of Rs 13 in 2007-08. The company‟s liquidity ratio reveals that the company has considerable amount of cur rents assets and can very well pay off its current liabilities through it.
The financial ratios of RPL, though do not canvas a very clear picture of financial health of the company. This is because the company has started its operation on 15 th March, 2009. Hence only 15 days operational data has been made public. Currently the company has been highly leveraged as D/E Ratio for the company is 0.95:1. This is because RPL is a new project and requires heavy machinery.
The free cash flows). RIL has positive figures for FCFF and FCFE which indicates sound financial position of the company. The figures for RPL are negative because the company has only recently started production (only 15 days of production till March 2009). Further, the weighted average cost of capital has been calculated for both the companies depending on the volatility of its shares at the stock market. Last one year‟s data has been taken to calculate the variance and standard deviation. The RIL stock prices are more volatile when
10
compared to RPL. Thus the cost of equity for RIL is higher than Reliance Petroleum. Similarly, the cost of debt is compared based on the financial charges based over the total value of debt. The cost of debt is little higher for RPL 5.96 than that of RIL‟s 4 .55. But as RPL is better leveraged than RIL and the cost of equity (Ke) is higher than cost of debt (Kd), the WACC for RIL is higher than RPL. As RIL has a higher WACC, the valuation of RPL is better when compared to RIL (in respect of WACC).
Thus, RIL seems to be a little riskier investment than RPL. But from the past records of RIL, it has proved to be an ever growing company. Its financial records speak very well of the financial health of the company and shows that RIL has been a profit making company since its inception. The credit for the profits of the company goes to its various subsidiaries, which have been compensating for each other‟s losses through their profits.
Thus, numbers might term RIL to be a riskier firm when compared to RPL. But on papers, RIL is a much stronger company and shows a bright future. The past profits of the company second the statement of RIL being an intelligent investment.
11
INTRODUCTION
ABOUT THE COMPANY The Reliance Group was founded by Late Sh. Dhirubhai Ambani a nd today it is India's largest private sector enterprise. The flagship company, Reliance Industries Limited, is a Fortune Global 500 company and is the largest private sector company in India. For Reliance, backward vertical integration has been the cornerstone of the evolution and growth. It started with textiles and ever since Reliance has pursued a strategy of backward vertical integration in polyester, fiber intermediates, plastics, petrochemicals, petroleum refining and oil and gas exploration and production. Exploration and production of oil and gas, petroleum refining and marketing, petrochemicals (polyester, fiber intermediates, plastics and chemicals), textiles, retail and special economic zones has been the core activities of the Group. Reliance enjoys global leadership as the largest polyester yarn and fiber producer. Also, it is amongst the top ten producers of the petrochemical products in the world. Major Group Companies are Reliance Industries Limited (including main subsidiaries Reliance Petroleum Limited and Reliance Retail Limited) and Reliance Industrial Infrastructure Limited. PURPOSE OF THE REPORT The Final Report is the written component of the evaluation of the internship. This report contains the work done by me during my three months o f internship with RELIANCE INDUSTRIES LIMITED. This report is an attempt to document the analysis and leanings made by me during the period. This report will help the Faculty Guide Mr. AJIT PATIL and the Company Guide Mr. K. R. RAJA and Mr. HARIHARAN to understand my work.
12
SCOPE OF THE REPORT As a part of the MBA program, I have undergone an industry internship with M/s Reliance Industries Limited to understand the practical applications of various financial instruments, transactions, processes and administration of the finance function. To achieve this objective, my company guide advised me to study the project financing for Refinery and Polypropylene plant being developed by Reliance Petroleum Limited (RPL), a subsidiary of RIL. Based on this study, I am expected to learn, in detail, the long term financing of the project and use this knowledge for related treasury and finance functions of the company like debt servicing, compliance of financial covenants, Accounting, MIS and filing of reports. The report discusses about Long Term Sources of Finance, factors which determine the requirement of long term sources of funding and sources of raising them. It also includes the study of equity and debt in detail. It discusses the pros and cons of different types of long term fund raising methods. The report further includes the detailed study of Initial Public Offering (IPO) including its preissue obligations as mentioned by Securities and Exchange Board of India (SEBI). It states the Book building process for deciding the issue price. Once there is a general understanding about the IPO, the report goes deep in the IPO process of Reliance Petroleum Limited and mentions specific SEBI guidelines for raising funds through IPO, which were compiled by RPL. The report also mentions the debt raising process and External Commercial Borrowing Guidelines. The report discusses the various covenants and clauses that are included in debt agreements. It broadly discusses the roles and responsibilities of various parties involved in debt agreement in light of RPL‟s Commercial Term Agreement. The report also includes the details of PORTFOLIO TRACKER VERSION 0.0 which helps investors to monitor various stocks in their portfolio at a specific period. The report contains the information about the various functionalities of the Microsoft Excel driven software. Further, to understand the current happenings in the Reliance Industries Limited, my company guide and the faculty guide suggested me to follow the recent merger of RPL with the company. This helped me in understanding the concept of merger and amalgamation. This also helped me
13
in analyzing the swap ratio decided by the board members of the Reliance Industries. The report also discusses the analysis of various analysts on the idea of merger and further the declared swap ratio. A calculation has been done to study the impact on the promoter‟s holding in the RIL after the amalgamation of RPL as the stated ratio of 1:16. In order to understand the various options available to the investors of RPL and RIL, the valuation of both the companies has been done. As a part of valuation the previous 5 year data has been used for RIL to estimate its future financial statistics. Percentage of Sales method has been used for estimation of the future financials. After the estimations, various financial ratios like liquidity ratios, leverage ratios, payout ratios, coverage ratios etc. have been calculated and analyzed in order to study the financial health of the company. However, due to the lack of the historic data for Reliance Petroleum Limited, 15 days production data has been extrapolated to estimate the year‟s data. This data has been used to estimate the future financials of the company. The growth of sales has been estimated by estimating the GDP growth rate of India and the utilization capacity of the refinery. The report also deals with the historic data of stock prices of RIL and RPL in consideration with Nifty Index for last year, starting from 1 April, 2008. This data has been used to find out the annual return, standard deviation and volatility of the stocks of these companies. This volatility helps in calculating the cost of equity. Further, by using the cost of equity and cost of debt, weighted average cost of capital is calculated for both the companies. The valuation of the company is also discussed based on the result of WACC. The free cash flows for both the companies have also been calculated which). These cash flows are indicators of the financial position of the company.
14
LIMITATIONS 1. While analyzing the prospectus of Reliance Petroleum Limited, I understand t hat the price band for the book building process was decided by management of the company in consultation with the merchant bankers to the issue. I am explained that the determination of the price band involves complex research, calculations and analysis o f various factors including market conditions, financing, project schedule, and project feasibility. This research and calculations are subjective in nature and no concrete data is available to us for study. 2. Some of the key documents and information provid ed by the company about the debt raising was technical and confidential in nature and hence, was not studied. 3. The software updates the prices at particular interval; hence the change in prices for less than a minute cannot be accommodated in it. This might become a shortcoming while deciding the arbitrage strategy as the stock prices changes at every fraction of seconds. 4. Non availability of the previous year‟s financial data (Profit &Loss account) was a major limitation in the process as the company hadn‟t started the production. Due to this, the exact percentage of the expenditure, incomes etc are not available. Thus, there are high chances that the projections made in the project might vary considerably from the actual results. 5. RPL‟s refinery has been des igned to have a Nelson Complexity Index of 14.0, which would make it amongst the most complex refineries in Asia. Similarly GRM of the company is estimated to be on higher side as compared to industry average. Due to this, the competitor‟s data cannot be of much help in estimating RPL‟s financials. Therefore, many assumptions have to be made while forecasting RPL‟s financial statements. 6. The comparison of Reliance Petroleum Limited with the competitors for the purpose of forecasting is not viable because RPL has been set up in a Special Economic Zone (SEZ), and therefore, the tax and other benefits are not available to other refineries. 7. The company cannot benchmark itself against the industry average and the industry leader.
15
METHODOLOGY 1) Study of Securities and Exchange Board of India (Disclosure and Investor Protection) Guidelines, 2000, to understand the legal framework to be followed while doing an Initial Public Offering (IPO). 2) Study and analysis of prospectus of RPL to understand the equity funding for the company. It helped in understanding the various methods through which equity can be raised for a company. It also provided an insight about the promoter‟s contribution, lock in period etc. 3) Study of Common Terms Agreement relating to the financing of a refinery and polypropylene plant in Jamnagar (Gujarat) by RPL. This helped in understanding the agreement clause between the company and its lenders, thus explaining the debt financing of a company. 4) Financial forecasting of RIL is done by using the historic data of the company and by analyzing the financial performance in the past. Percentage of Sales method is used for estimating these statistics. For RPL, the latest quarter results, released on 23 rd April 2009 are used to do the financial forecasting. 5) Financial ratios, stock volatility, weighted average cost of capital (WACC), free cash flow to equity (FCFE) and free cash flow to firm (FCFF) are calculated as a part of valuation of the company. 6) SECONDARY DATA: Using the secondary data available on internet like analysis done about RPL and other competitor companies. 7) CASE ANALYSIS: Case analysis as we are studying the RPL‟s funding process to understand the funding procedure of any company.
16
INDUSTRY OVERVIEW
GLOBAL OIL REFINING INDUSTRY:The oil refining industry is global in nature because crude oils, other feedstock and refined petroleum products can be transported at a relatively low cost by sea and by pipeline and there is worldwide demand for such products. The principal factors affec ting refining margins are the demand for and prices of refined petroleum products relative to the supply and cost of crude oils and other feedstock and the configuration, capacity and utilization rates of refineries. The range and quality of refined petroleum products produced by any given refinery depends on the types of crude oil used as feedstock and the configuration of the refinery. REFINED PETROLEUM PRODUCTS:LPG Naphtha Gasoline Middle distills Fuel oils Pet coke Bitumen Niche, high value added refined petroleum products
REFINING INDUSTRY CHARACTERISTICS:1. Economics of oil refining- Oil refining is primarily a margin-based business in which a refiner‟s goal is to optimize the refining processes and yields of all products in relation to feedstock used. The Gross Refining Margins (GRMs) of complex refineries are higher than those of simple refineries because complex refineries are able to generate a higher yield of light and middle distillates from lower cost heavier and sourer crude oils. Crude oil typically accounts for 90% to 95% of the total cost of refining. Because other operating expenses are relatively fixed, the goal of refineries is to maximize utilization
17
rates, maximize the yields of higher value-added products, minimize feedstock costs and minimize operating expenses. 2. Location of oil refine ries- The location of an oil refinery can have an important impact on its refining margin since the location influences its ability to access feedstock and distribute its products efficiently. The location dictates what proportion of the feedstock and products can be transported by tanker vessels by sea or via pipelines, rail or tank trucks. Refining companies seek to maximize their profits by placing their products in the markets where they receive the highest returns after taking into account delivery transportation costs and other expenses such as import duties in those markets. Due to their flexibility and lower logistics costs, coastal refineries typically have a competitive advantage over the oil refineries located inland. 3. Crude oil supply- In 2004, the global oil supply was estimated by the International Energy Agency (“IEA”) to be 82.1 million barrels per day. The Middle Eastern OPEC countries accounted for 27.8% and total OPEC countries accounted for 39.5% of this supply. IEA estimates that by 2020, global oil supply may reach 104.9 million barrels per day with Middle Eastern OPEC countries accounting for 33.7% and total OPEC countries accounting for 45.2%. SUPPLY AND DEMAND FOR REFINED PRODUCTS:The world‟s total refining capacity has remained at approximately the same level as it w as in the beginning of the 1980s. This trend has been enabled, in part, by upgrades and debottlenecking of existing refineries and combinations of adjacent facilities. However, it is believed that tightening petroleum product specifications are likely to result in further closures of low complexity and low economic size refineries.
18
Table 1: World Refining Capacity in MMBD 1994 World USA Europe including forme r Soviet Union Asia-Pacific India China Source: BP Statistical Review 2005 75.7 15.4 26.5 15.9 1.1 3.6 1999 81.9 16.5 24.8 21.4 2.2 5.4 2004 84.6 17.0 25.2 21.9 2.5 5.8 CAGR 19942004 1.1% 1.0% -0.5% 3.2% 8.9% 5.0%
Table 2: World demand growth in MMBD 1994 World USA Europe including forme r Soviet Union Asia-Pacific India China Source: BP Statistical Review 2005 68.2 17.7 19.8 17.1 1.4 3.1 1999 74.9 19.5 19.7 20.3 2.1 4.4 2004 80.8 20.5 20.0 23.5 2.6 6.7 CAGR 19942004 1.7% 1.5% 0.1% 3.2% 6.1% 7.8%
19
KEY INDUSTRY TREND:Economic growth and energy demand- Economic growth is a key driver of energy demand given the close correlation between total energy demand and economic output. In the World Energy Outlook 2004, the IEA estimated that in recent decades energy demand has risen in a broadly linear fashion along with gross domestic product. Table 3: GDP Billion US$ (ON PPP BASIS)
2002 World Asia-Pacific Middle East USA China Brazil India Russia 47227 17100 1431 10075 5494 1370 3160 1657 2010 65449 26158 2113 13084 9716 1783 5031 2543 2015 78947 33076 2594 15216 13003 2170 6524 3019 2020 94582 41253 3147 17634 16919 2638 8430 3579 2025 112752 51024 3789 20292 21699 3209 10807 4192 CAGR(%) 2002-2025 3.9% 4.9% 4.3% 3.1% 6.2% 3.8% 5.5% 4.1%
Source: EIA International Energy Outlook 2005 Economic regulations- Although industrialized countries continue to consume most of the world‟s petroleum products, growth in demand for refined petroleum products over the last few years has primarily been driven by non-OECD countries, most notably China. The general growth in consumption and the stricter specifications have contributed to an increased demand for lighter refined petroleum products, such as gasoline and middle distillates, and lower demand for heavier products, such as fuel oils, contributing to the larger price differentials between higher value and lower value refined petroleum products. Therefore, it is believed that complex refineries which can produce environmentally friendly fuel are better positioned to meet growing market demand for these light products. Increase in light-heavy spread- Over the years, the demand for light and middle distillates has increased, including new demand for products that meet stricter environmental standards, driving up the sales prices for light and middle distillates. The combination of these market factors has resulted in an increasing light heavy differential. Given these trends, it is believed that complex refineries that are able to convert heavy crude oils into light products can achieve significantly higher GRMs than simple refineries.
20
Shortage of complex refining capacity- As demand for fuel oil has been decreasing with the increase in demand for light products; existing simple refineries will either be phased out or will need to be upgraded. The chart below shows an incremental global refining requirement forecast to 2020, as reported by HART‟s World Refining and Fuels Services. While the additional crude distillation capacity is estimated at 18.8 million barrels per day (a 22% increase) from 2005 to 2020 levels, the conversion capacity additions are estimated at 12.4 million barrels per day (a 51% increase) from 2005 to 2020 levels and desulphurization capacity additions are estimated at 21.8 million barrels per day (a 54% increase) from 2005 to 2020 levels. Figure 1: Refining Require ment Forecast
Source: HART’S World Refining and Fuels Services, December 2005.
PROPELEYNE INDUSRTY:Polypropylene is a crystalline thermoplastic with a unique combination of physical, thermal and chemical resistant properties, produced by polymerization of propylene. According to CMAI, global consumption of polypropylene was estimated at 39.6 million tonnes and accounted for approximately 24% of all plastic demand in 2005. Polypropylene demand shows cycles that closely follow GDP cycles, growing as GDP increases. Global consumption of polypropylene is forecast to grow 5.4% per annum between 2005 and 2010 according to CMAI. Growth rates are expected to be higher in rapidly developing economies of the Asia region, including China and India, where current per capital consumption of polypropylene is low compared to more developed countries.
21
BUSINESS OVERVIEW
In 2006 Reliance Petroleum Limited was a startup company, formed to setup a Greenfield petroleum refinery and polypropylene plant. The plant is located in a Special Economic Zone in Jamnagar (Gujarat). The developed refinery has a complexity of 14.0, as mea sured using the Nelson Complexity Index. RPL has two major promoters, Reliance Industries Limited (RIL) and Chevron India Holding Pvt. Ltd. The RPL refinery and polypropylene plant is located adjacent to the existing refinery of RIL. RPL is 75% owned subsidiary of RIL. RPL has an agreement with Bechtel France S.A.S (“Bechtel”) to license the technology for the major process units of the refinery and polypropylene plant. Bechtel has also provided engineering, project management and other construction se rvices to the project. The refinery and polypropylene plant is locates in a Special Economic Zone (the “SEZ”) and hence receives certain tax benefits and concessions under SEZ regulations. PLANNED FINANCING FOR THE PROJECT AT THE TIME OF IPO The capital cost of the project was estimated at Rs. 270 billion. The project was funded through debt (Rs. 157.5 billion) and equity (Rs 112.5 billion). Before IPO, RPL entered into a preliminary term sheet with certain banks and financial institutions to provide for a syndicate term loan facility for approximately Rs. 67.5 billion. Additional financing through export credit agencies was proposed for approximately Rs. 45-67.5 billion. Another Rs. 22.5-33.75 billion were to be raised by further debt financing. As the total funds requirement for the projects is estimated at Rs. 270,000 million. The company has proposed to fund the Project through a mixture of debt and equity. The details of the proposed funding are as follows:
22
Table 4: Proposed Funding for RPL Source Total equity including proceeds from the issue Debt - Syndicate Loan - Exports Credits - Rupee Debt/Bonds Total Debt Total Source: RPL Prospectus 2006 pg.32 KEY COMPETITIVE STRENGTH Following are the major competitive strengths of RPL refinery and polypropylene plant: 1) RIL‟s superior project execution skills in constructing a complex refinery: One of the promoters of RPL, Reliance Industries Limited has core competency in conceptualizing and implementing the multi-billion dollar projects on time and in a cost efficient manner. Thus, RPL got benefits from the expertise of RIL in construction of its refinery. 2) Large and complex refinery capable of using heavier and sourer, low cost crude to produce high quality, premium petroleum products. 3) Benefits of low capital costs: RIL‟s has a prior experience in constructing and operating the Jamnagar refinery, especially in the areas of design and engineering, construction, labor and resource optimization, greater use of local material and resources and faster implementation. This resulted in a significant reduction in the capital cost for the project and enabled RPL to achieve lower cost per barrel, adjusted for complexity. 4) Strategic location with proximity to crude oil sources and target export market: The refinery is located on the west coast of India in close proximity to the Middle East, the largest crude oil producing region in the world. This will result in lower ship t urnaround time and crude freight costs. 5) Fiscal incentives by virtue of being located in a Special Economic Zone: An SEZ operates as a delineated area which is deemed to be a foreign territory for the purposes of trade operations, duties and tariffs. Being an export oriented refinery, RPL intend to Range Amount (in Rs. millions) 135000
45000-90000 45000-67500 22500-33750 157500 270000
23
export the bulk of their production. Hence they will benefit from an income tax deduction on export turnover for a period of five consecutive years following the commencement of commercial operations (with a scaled reduction in income tax deduction for the next five year period and, subject to certain reinvestment conditions, for a third five year period thereafter). They will also be exempt from customs duty for goods and services imported into or exported from the SEZ and also from excise duty on domestic procurement, for the purposes of our authorized operations.
24
LONG TERM SOURCES OF FINANCE
Long term sources of finance are those that are repayable over a longer period of time, generally for more than 12 months based on the feasibility of the Company‟s business/project. The sources of long term finance are equity shares, debentures, public deposits, term loans from banks etc. The need for the long term finance is determined by certain factors like the nature of the business, the goods produced and the type of technology that is required for the business to be carried out. Mainly, the long term sources of finance are used to finance fixed assets, to finance the permanent part of the working capital requirement and to fund the growth and expansion of the business as it is done for a longer period of time. The sources of long term finance can be broadly classified into: EQUITY- The amount of funds contributed by the owners or the stockholders including the retained earnings taken together is termed as the shareholder‟s equity. There are different methods of raising equity finance i.e. promoters co ntribution, initial public offering, private placement and right‟s issue. The company can issue its shares either on par i.e. at Face value or at a premium which is known as the share premium. The company‟s earnings which have not been distributed to share holders and have been retained in the business are known as the reserves and surplus. DEBT- An amount of money borrowed by one party from another is known as debt. A debt arrangement gives the borrowing party permission to borrow money under the condition that it is to be paid back at a later date, usually with interest. Generally, the startup companies often turn to debt to finance their operations. In fact, almost all the corporate balance sheets will include some level of debt. The debt is also referred to as “leverage.” The most popular source for debt financing is the bank and other financial institutions. A company can also raise debt by selling the debentures of the company to the lenders.
25
Refinery project requires huge investments for the setting up various processing units, pipelines, storages etc. This kind of project requires large amount of investment and short term means of finance like unsecured loans cannot meet such requirements. Hence, the long term sources of finance like raising funds through equity shares and raising long term secured debt is more viable. Moreover, the gestation period to start the operations of typical refining plant is also
very long. Accordingly huge amount of long term financing is envisaged in such projects.
CAPITAL MARKET
The Capital Market is the market for securities (broadly classified into debt and equity), where companies and government can raise long term funds. The securities decouple individual acts of savings and investment over time, space and entities thus allow savings to occur without concomitant investment. The capital market aids economic growth by mobilizing the savings of the economic sectors and directing the same towards channel of productive uses. The capital market acts as a break on channeling savings to low-yielding enterprises and impels enterprises to focus on performance. It continuously monitors performance through movements of share prices in the market and the threats of takeover. Financial market works as a conduit for demand and supply of long-term debt and equity capital. Money provided by savers and depository institutions are channeled towards the borrowers and investees through various financial instruments like securities. A capital market is a highly decentralized system comprising of three major parts, namely stock market, bond market, and money market. It also works as an exchange for trading existing shares.
26
DIFFERENT KINDS OF EQUITY ISSUE
Figure 2: Diffe rent Kind of Equity Issue
27
Initial Public Offering. A Further Issues is when a listed company proposes to issue fresh securities to its existing shareholders as on a record date. Here, existing shareholders have the privilege to buy a specified number of new shares from the firm at a specified price within a specified time. A rights issue is offered to all existing shareholders individually and may be rejected, accepted in full or (in a typical rights issue) accepted in part by each shareholder. Rights are often transferable, allowing the holder to sell them on the open market. A private placement is an issue of shares or of convertible securities by a company to a select group of persons under Section 81 of the Companies Act, 1956 which is neither a rights issue nor a public issue. It is a direct private offering of securities to a limited number of sophisticated investors. which include pricing, disclosures in notice etc, in addition to the requirements specified in the Companies Act. A Qualified Institutions Place ment is a private placement of equity shares or securities convertible in to equity shares by a listed company to Qualified Institutions Buyers only in terms of provisions of Chapter XIIIA of SEBI (DIP) guidelines. The Chapter contains provisions relating to pricing, disclosures, currency of instruments etc.
28
INITIAL PUBLIC OFFER (IPO)
An Initial Public Offer (IPO) may be termed as the maiden offer made by a non-public company to take up equity stake by the public. This offering is normally made by the company in order to raise public funds for its future projects. A corporate may raise capital in the primary market by way of an initial public offer, rights issue or a private placement by issuing either debt or equity instruments. An Initial Public Offer (IPO) is the selling of securities to the public in the primary market. It is the largest source of funds with long or indefinite maturity for the company. IPO is the first sales of stock by a company to the public through investment banking firms. A successful initial public offering increases the visibility and appeal of the company, thereby increasing the demand and value for shares of the company. Public companies have many shareholders and are subject to strict rules and regulations. They comprise of a BOD (Board of Directors) consisting of the requisite number of independent directors. This is for complying with the provisions of Corporate Governance. The management of the company is entrusted to update the BOD with all developments including the updated financial information on every quarter. In India, the regulatory body that guides these public companies is Securities and Exchange Board of India (SEBI). Reasons for Going Public: The main reasons for going public generally include: Raising funds to finance capital expenditure programs like expansion, diversification, modernization etc Arranging funds for increased working capital requirements Financing acquisitions like a manufacturing unit, brand acquisitions etc Debt refinancing It works as exit route for existing investors.
29
Pros and Cons of IPO: When a company has strong foothold in the market, it is easier for the company to raise funds at a cheaper cost. But when a private firm starts off at an initial stage its cost of acquir ing funds is generally high. To support its activities company requires long term funds at a cost which is lower than the return on capital employed. The benefits of an IPO are: A publicly traded company may tap a broader universe of investors as well as a larger pool of investment capital. It provides liquidity to the existing shareholders. A publicly traded company seeks more attention and hence increases its visibility. It may raise more capital through additional stock offerings if sufficient investor interest exists. A publicly traded company may be able to attract and retain highly qualified personnel if it can offer employee stock option plans, bonus shares or other incentives, which might be instrumental in reducing the high attrition rate from which the corporate world is presently suffering. A listed company is in the position of winning the confidence of the mass because of its transparency exercised through stringent regulations imposed by the stock exchanges.
However, the use of IPOs is limited because of the following reasons: It involves substantial expenses. A publicly traded company needs to make continuous disclosers. It involves complexity in complying with federal and state laws governing the sale of business securities. Thus, it has increased regulatory monitoring. Offering business's ownership for public sale does little good unless the company has sufficient investor awareness and appeal to make the IPO worthwhile Management must be ready to handle the administrative and legal demands of widespread public ownership. Therefore, it requires substantial amount of management time and efforts.
30
An IPO also means a dilution of the existing shareholders interests. IPOs can be a risky investment for the individual investor as it is difficult to predict what the stock will do on its initial day of trading.
INTERMEDIARIES INVOLVED IN IPO
Merchant Banker:". Merchant Bankers assists the company from the initial phase of preparing prospectus to the listing of securities at the stock exchange. It is mandatory for them to carry due diligence for all the information provided in the prospectus and they must issue a certificate to SEBI. A Company may appoint more than one Merchant Banker provided inter-Se allocation of responsibilities between the Merchant Bankers is properly structured. Unde rwriters: Underwriters are a company or other entity that administers the public issuance and distribution of securities from a corporation or other issuing body. An underwriter works closely with the issuing body to determine the offering price of the securities, In case there is under subscription (i.e., the company does not receive good response from public and amount received from is less than the issue size), underwriters subscribe to the unsubscribed amount so that the issue is successful. Bankers to an Issue: Bankers to an issue means a scheduled bank carrying on all or any of the following issue related activities namely:i. ii. iii. iv. Acceptance of application and application monies; Acceptance of allotment or call monies; Refund of application monies; Payment of dividend or interest warrants.
31
Registrar and Share Transfer Agents: They are responsible for processing all applications received from the public and prepare the basis of allotment. The dispatches of share certificates / refund orders are also handled by them. Stock Brokers & Sub-Broke rs: These are intermediaries who charge a fee or commission for executing buy and sell orders submitted by an investor. Depositories are the intermediaries who hold securities in dematerialized form on behalf of the shareholders. For RPL, the book running lead managers were: 1. JM Morgan Stanley Private Limited 2. DSP Merrill Lynch Limited 3. Citigroup Global Markets India Private Limited 4. Deutsche Equities India Private Limited 5. Enam Financial Consultants Private Limited 6. HSBC Securities and Capital Markets (India) Private Limited 7. ICICI Securities Limited 8. SBI Capital Markets Limited 9. UBS Securities India Private Limited The Registrar to the issue for RPL is KARVY Computershare Private Limited.
32
CONSIDERATION BEFORE DECIDING FOR AN IPO
Before deciding to launch an IPO, the management of the issuing company should pay considerable attention to its future business model. Deciding to come out with the IPO considering the brighter side of mobilizing resources is not enough, unless it is supported by an increase in the shareholder‟s value of the invested funds. When an inve stor invests in the stock of a company, he expects its stock value to grow constantly, subject to market conditions. Therefore, to have a successful IPO, the company must have a robust business model and a considerable future growth prospects. For a successful IPO, a company should take care of following point: It should have an effective risk management system. It should also have an internal audit department reviewing the procedures implemented and continuously improve the same. The company should abide by the corporate governance procedures to safeguard the interests of its stakeholders as well as its own interest. The management should be capable enough to make effective and efficient utilization of the resources. In case, the company is unable to fulfill the above the valuation of stock deteriorates and the company will lose market confidence. Decreased valuation of the company may affect the lines of credits, pricing of any follow-on public issue, ability to maintain morale of the employees, confidence and value of the shareholders‟ wealth. Thus a company should comply with the above mentioned points in order to have a successful IPO.
33
IPO BY AN UNLISTED COMPANY
Securities and Exchange Board of India (SEBI) came into existence in 1992, since Controller of Capital Issues was dissolved. To come up with an IPO the company needs to fulfill all guidelines (rules and regulation) issued by SEBI under Section 11 of the Securities and Exchange Board of India Act, 1992. The guidelines are called Disclosure and Investor Protection Guidelines (DIP). As per the SEBI guidelines, an unlisted company can make a public offering (IPO) of equity shares or any other security which may be converted into or exchanged with equity shares at a later date, only if it meets all the following conditions: a. The company has net tangible assets of at least Rs.3 crores in each of the preceding 3 full years of which not more than 50% is held in monetary assets. b. The company has a track record of distributable profits for at least 3 years out of immediately preceding 5 years. c. The company has a net worth of at least Rs. 1 crore in each of the preceding 3 full years. d. In case the company has changed its name within the last 1 year, at least 50% of the revenue for the preceding 1 full year is earned by the company from the activity suggested by the new name; and e. The aggregate of the proposed issue and all the previous issues made in the same financial year in terms of size does not exceed 5 times its pre- issue net worth as per the audited balance sheet of the last financial year. As RPL was a newly incorporated company hence the above conditions were not applicable. SEBI guidelines take care of the condition where a newly incorporated company might like to go for an IPO hence as per SEBI guideline, if the unlisted company does not comply with the above given conditions it may make an IPO of equity shares or any other security if it meets both the conditions: a. The issue is made through book building process with at least 50% of the net offer being allotted to qualified institutional buyers (QIB), failing which the full subscription monies
34
shall be refunded, OR, the project has at least 15% participation by Financial Institutions/ Scheduled Commercial Banks, of which at least 10% comes from the appraisers. In addition to this at least 10% of the issue size shall be allotted to QIBs failing which the full subscription monies shall be refunded. b. The minimum post issue face value capital of the company shall be Rs. 10 crores, OR, there shall be compulsory market making for at least 2 years from the date of the listing of the shares. RPL very well qualify in the above stated condition. The details of the same are elaborated while discussing the details of RPL’s IPO. Note: Net tangible assets mean the sum of all net assets of the company, excluding the intangible assets defined under the AS-26 issued by ICAI. Project means the object for which the monies proposed to be raised to cover the objects of the issue.
35
PRE ISSUE OBLIGATIONS
As per the SEBI Guidelines, there are certain pre- issue obligations that the Company should take into account before the issue. These obligations can be summarized as follows: 1. The Lead Merchant banker has exercised due diligence about all the aspects of offering, veracity and adequacy of disclosure in the offer documents as the liability of the merchant banker continues even after the completion of the issue process. 2. The documents that had to be submitted along with the Offer Document by the Lead Manager like the Memorandum of Understanding, a statement of the allocation of the responsibilities among all the merchant bankers to the issue and the list of promoters‟ group and other details. 3. The condition given by SEBI regarding the minimum number of lead manager to be appointed based on the size of the issue (given in the table below) had been complied with. Table 5: Lead Managers to size of issue Size of the Issue Less than Rs. 50 crores Rs.50 crores to Rs.100 crores Rs.100 crores to Rs.200 crores Rs.200 crores to Rs.400 crores Above Rs.400 crores No. of Lead Managers 2 3 4 5 5 or more
4. An agreement with the depositories should have been entered into by the issuer company
for the dematerialization of securities.
From the study of the prospectus of RPL, I understand that all the above obligations were fulfilled by the Company.
36
TERMS OF THE ISSUE
The Equity Shares being offered are subject to the provisions of the: o Companies Act o Memorandum of Association of the company o Articles of Association of the company o The terms of the Red Herring Prospectus, the Prospectus, the Bid cum Application Form and other documents/certificates. The Equity Shares are also subjected to: o Laws as applicable o Guidelines o Notifications o Regulations relating to the issue of capital and, o Listing and trading of securities issued by SEBI, Government of India, and the Stock Exchanges.
37
PRICING BY COMPANIES ISSUING SECURITIES
An unlisted company eligible to make a public issue and desirous of getting its securities listed on a recognized stock exchange pursuant to a public issue, may freely price its equity shares. Price Band: As per the SEBI guidelines, following are the points to be taken care off while deciding the price band of the issue: a. Issuer company can mention a price band of 20% (cap in the price band should not be more than 20% of the floor price) in the offer document filed with the Board and the actual price can be determined at a later date befo re filing of the offer document with the ROCs. b. The final offer document shall contain only one price and one set of financial projections, if applicable. Freedom to determine the denomination of shares for public/right issues and to change the standard denomination: a. In case of IPO by an unlisted company a. If the Issue Price is Rs 500 or more, the issuer company shall have discretion to fix the face value below Rs 10 per share, subject to the condition that the face value shall in no case be less than Re 1 per share. b. If Issue Price is less than Rs 500 per share, the face value shall be Rs 10 per share.
As the Issue Price of RPL is Rs 60, the face value of the RPL shares is Rs 10 per share.
b. The disclosure about the face value of the shares (including the statement about the issue price being “X” times of the face value) shall be made in the advertisement, offer document and in application forms.
RPL has complied with the above guideline as it has clearly mentioned the face value of the shares in all the above stated documents.
38
Book Building Process
Corporate can raise capital in the primary market either by way of an initial public offer, rights issue or private placement. An Initial Public Offer (IPO) is the selling of securities to the public in the primary market. Initial Public Offering can be made either through the fixed price method or the book building method or a combination of both. In case the issuer company wants to issue the securities through the book building process then as per SEBI guidelines, the securities can be issued in the following manner: a. 100% of the net offer to public through the book building procedure. b. 75% of the net offer to public through the book building process and the remaining 25% through the fixed price method. c. Under the 90% scheme, this percentage would be 90 and 10 for the book building process and fixed price method respectively.
THE PROCESS The issuer who is planning an IPO nominates a lead merchant banker as a „book runner‟. The issuer specifies the number of securities to be issued and the price band for orders. The issuer also appoints syndicate members with whom orders can be placed by the investors. Investors place their order with the syndicate member who incorporates the orders into the „electronic book‟. This process is called „bidding‟ and is similar to open auction. The Book should remain open for a minimum of five days. Bids cannot be entered for a price which is less than the floor price. The bidder can revise bids before the issue closes. On the close of the book building period the book runner evaluate the bids on the basis of the evaluation criteria which may include:
39
-
Price Aggression Investor quality Earliness of bids
The book runner and the company conclude the final price at which it is willing to issue the stock and allocation of securities. Generally, the number of shares is fixed; the issue size gets frozen based on the price per share discovered through the book building process. Allocation of securities is made to the successful bidders.
Price Band
Draft Red Herring Prospectus
SEBI + Stock Exchange
Stock Exchange Initial Listing Approval
Issue Opens
SEBI Filing
Red Herring Prospectus
SEBI Comment
ROC filing (3 days before the issue opens)
Issue Closes
Fixation of Price
Prospectus
ROC Filing
Figure 3: Book Building Process
40
PROMOTER’S CONTRIBUTION & LOCK -IN PERIOD
In accordance with the SEBI Guidelines, 2000; the promoter‟s contribution in any issue shall be in accordance with the following provisions as on: a. The date of filing the red herring prospectus (in case of a book built issue) or prospectus (in case of a fixed price issue) with ROC or letter of offer with designated stock exchange as the case may be in case of fast track issue and b. The date of filing draft offer document with the board, in any other case. In case of the public issue by unlisted companies, the promoter shall contribute not less than 20% of the post issue capital. The promoters have to bring in the full amount of the promoters contributio n including premium at least one day prior to the issue opening date, which shall be kept in escrow account with a Scheduled commercial Bank and the said contribution /amount shall be released to the company along with the public issue proceeds. In case t he promoter‟s contribution has already been deployed by the company, the company has to give the cash flow statement in the offer document disclosing the use of such funds received as promoter‟s contribution. Details of pre- issue shareholding and promoter‟s contribution in RPL are as under; Table 6: Pre Issue Shareholdings Name of Shareholders Pre-Issue (As on date of filing of the Red Herring Prospectus with ROC) Number of Equity Shares Equity shares capital (%) Post-Issue Number of Equity Shares Equity share capital (%) 75% 5% 10% 10% 100%
Promoter – RIL 2,700,000,000 Promoter – Nil Chevron Pre-IPO Investors 450,000,000 Public Nil Total 3,150,000,000 Source: RPL Prospectus 2006 pg. 24
85.71% 0.00% 14.29% Nil 100%
3,375,000,000 225,000,000 450,000,000 450,000,000 4,500,000,000
41
Table 7: Capital Structure Price/ Equity Shares (Rs) Dec 06,05 100,000 10 Jan 30,06 4,300,000 10 Feb 25,06 2,695,600,000 10 Apr 03,06 450,000,000 60 Apr12, 06 900,000,000 60 Source: RPL Prospectus 2006 pg.23 Date of Allotment No. of Equity Shares Consideration (Cash, bonus, other than cash) Cash Cash Cash Cash Cash Cumulative Share Premium (Rs) Nil Nil Nil 22,500,000,000 67,500,000,000 Cumulative Share Capital (Rs) 100,000 4,400,000 2,700,000,000 3,150,000,000 4,050,000,000
The entire pre-Issue capital (including the Pre-IPO placement, but excluding the minimum Promoter‟s Contribution) would be locked - in for a period of one year from the date of allotment in the Issue. Hence the capital brought-in on 6-Dec-2005, 30-Jan-2006 and 25-Apr-2006 by Reliance Industries Limited would be locked for 1 year. 900,000,000 Equity Shares which form the 20% of the post Issue paid- up capital. These Equity Shares represent the minimum Promoters Contribution pursuant to clause 4.1.1 of the SEBI Guidelines. In terms of clause 4.11.1 of SEBI Guidelines, these Equity Shares will be locked-in for a period of 3 years from the date of allotment in the Issue or the date of commercial production, whichever is later or as per the SEBI Guidelines. As per Clause 4.15.1 of the SEBI Guidelines, the locked- in Equity Shares held by the Promoter can be pledged only with banks or financial institutions as collateral security for loans granted by such banks or financial institutions, provided the pledge of shares is one of the terms of sanction of loan. Under Clause 4.16.1(a) of the SEBI Guidelines, the Equity Shares held by persons other than the Promoter prior to the Issue may be transferred to any other person holding the Equity Shares which are locked- in as per Clause 4.14 of the SEBI Guidelines, subject to continuation of the lock-in in the hands of the transferees for the remaining period and compliance with SEBI Takeovers Regulations.
42
Further, under Clause 4.16.1(b) of the SEBI Guidelines, the Equity Shares held by the Promoter may be transferred to and amongst the Promoter group or to a new Promoter or persons in control of the Company subject to continuation of the lock- in in the hands of the transferees for the remaining period and compliance with SEBI Takeover Regulations. Thus, RIL has
transferred 225,000,000 Equity Shares constituting 5% of the post-Issue Equity Share Capital to Chevron India Holdings Pvt. Ltd. on April 27, 2006. These Equity Shares are held by Chevron India as part of the minimum promoter‟s contribution.
43
THE ISSUE
Table 8: Issue Details Issue of Equity Shares Of which, Promoter‟s Contribution Net Issue to Public Of which the QIB Portion Of which Available for Allocation to Mutual Funds Balance for all QIB including Mutual Funds Non-Institutional Portion Retail Portion Equity Shares outstanding prior to the Issue Equity Shares outstanding after the Issue Source: RPL Prospectus 2006 pg.5 1350,000,000 Equity Shares 900,000,000 Equity Shares 450,000,000 Equity Shares At least 270,000,000 Equity Shares (allocation on proportionate basis) 13,500,000 Equity Shares (allocation on proportionate basis) 256,500,000 Equity Shares (allocation on proportionate basis) Minimum of 45,000,000 Equity Shares (allocation on proportionate basis) Minimum of 135,000,000 Equity Shares (allocation on proportionate basis) 3,150,000,000 Equity Shares 4,500,000,000 Equity Shares
The total Issue of equity shares by RPL is of 1,350 million equity shares, out of which the promoter‟s contribution is nearly 2/3rd of the issue, i.e. 900 million equity shares and 1/3 rd of the total issue (450 million equity shares) are the net issue to public. As equity shares are being offered to the public through 100% Book Building Process in Out of 450 million shares (net issue to the public), in accordance with the SEBI Guidelines read with Rule 19(2)(b) of the SCRR (The Securities Contracts (Regulation) Rules, 1957, as amended from time to time), wherein: 1) At least 60% of the Net Issue shall be allocated on a proportionate basis to QIBs (270 million shares out of 450 million shares offered to the public), including up to 5% of the QIB Portion that shall be available for allocation on a proportionate basis to Mutual Funds only (13.5 million shares out of 270 million shares offered to QIBs) and the remainder of the QIB Portion shall be available for Allocation on a proportionate basis to
44
all QIB Bidders, including Mutual Funds (256.5 million shares out of 270 million shares offered to QIBs); 2) Minimum of 10% of the Net issue shall be available for allocation on a proportionate basis to the Non-Institutional Bidders (45 million shares out of 450 million shares offered to QIBs) and 3) Minimum of 30% of the Net Issue shall be available for allocation on a proportionate basis to Retail Individual Bidders (135 million shares out of 450 million shares offered to QIBs), subject to valid Bids being received at or above the Issue Price.
The outstanding equity shares prior to the IPO were 3150 million shares, out of which 2700 million shares belonged to the Reliance Industries Limited (RIL) and 450 million shares were allotted to pre-IPO investors. FACE VALUE AND ISSUE PRICE The face value of the Equity Shares is Rs. 10 each and the Floor Price is Rs. 57 and the Cap Price is Rs. 62 per Equity Share. Issue Price is Rs 60 per equity share, which is 6 Times the face value. The above stated information has been clearly mentioned on the cover page of the RPL prospectus. Hence, it is in compliance with the SEBI guidelines, where the issuing company is asked to make a clear statement about the face value and the issue price of the shares. MARKET LOT AND TRADING LOT The equity shares of the company were only allotted in dematerialized form complying with existing SEBI Guidelines which states that the trading in the Equity Shares of the Company should only be in dematerialized form for all the investors.
45
OBJECTS OF THE ISSUE
The objects of the Issue by the company were to achieve the benefits of listing and raising capital for financing the proposed project. The company intended to utilize the proceeds of the Issue, after deducting underwriting and management fees, selling commissions and other expenses associated with the Issue (“Net Proceeds”), to partially finance the equity portion of the Project. As the SEBI guidelines, ask for clear information about the projected funds requirement of the company in the prospectus, RPL has included the project estimates in the following manner: Fund requirements: The company had proposed to set up the Project for an estimated cost of Rs. 270,000 millions (approx US$ 6 billion) as estimated by the company and intend to finance the Project through a combination of debt and equity. The Project was expected to begin commercial operations in, or around, December 2008. The estimated expenses expected to be incurred in connection with the Project are set forth below on a half yearly basis: Table 9: Estimated Expenses 1st half of CY 2006 2ndhalf of CY 2006 1st half of CY 2007 2ndhalf of CY 2007 1st half of CY 2008 2ndhalf of CY 2008 & beyond Total
Deposits for 5990 infrastructure including utilities etc. Equipment/Constr 25474 22176 uction costs etc. Technical fees 21678 7712 Interest during 7916 2715 construction, pre operating costs Contingency Margin money for working capital Total 61058 32639 Source: RPL Prospectus 2006 pg. 31
5990
40037 515 3874 44426
52770 2 5586 4875 63233
22295 650 7197 9750 39892
1088 9361 3892 4871 9540 28752
163840 39918 31216 19496 9540 270000
46
ISSUE RELATED EXPENSES The SEBI guidelines along with the projected cost of the Project also ask for the details of the other expenses that the company has to make in order to make an issue. Hence complying with that RPL prospectus includes the following information: The expenses of the Issue include, among others, underwriting and management fees, selling commission, printing and distribution expenses, legal fees, statutory advertisement expenses and listing fees. The estimated expenses of the Issue are as follows: Rs. in millions Table 10: Estimated Expenses of Issue Lead management, underwriting and selling commission Advertising and marketing expenses Printing, stationary including transportation of the same Others (Registrar‟s fees, legal fees, listing fees, etc.) Total estimated Issue expenses
Source: RPL Prospectus 2006 pg.34
202.5 80.0 92.0 93.1 467.6
This gives the investor a clear picture about the plan of the company, as to how they are planning to invest the funds which are raised by them. Thus, by following the SEBI guidelines, the issuing company becomes transparent and gives investors a chance to make an informed decision.
47
BASIS FOR ISSUE PRICE
The Issue Price has been determined by the company in consultation with the Book Running Lead Managers (BRLMs) on the basis of assessment of market demand for the offered Equity Shares by the Book Building Process. The face value of the Equity Shares is Rs. 10 and the Issue Price is 6.0 times the face value. The factors which influence the deciding of the issue price by the company can be broadly classified as qualitative and quantitative factors. These factors are discussed as under: QUALITATIVE FACTORS Factors Internal to the Company The company is promoted by Reliance Industries Limited (RIL), which is amongst the largest private sector companies, in terms of market capitalization, in India. Reliance Petroleum will benefit from economies of scale arising out of the size. The proposed refinery, having a capacity of 580 KBPSD, will be the sixth largest refinery globally based on current capacities. (Source: Oil and Gas Journal, December 2005). The company would derive significant advantages owing to higher complexity of the refinery. The higher complexity levels will enable the company to process lower cost, heavier and sourer crude oils and yet achieve superior yields of higher value products such as gasoline, aviation fuel and diesel. Nelson Complexity Index is a measure of secondary conversion capacity in comparison to the primary distillation capacity of any refinery. It is an indicator of not only the investment intensity or cost index of the refinery but also the value addition potential of a refinery. RPL has the NCI of 14.0 which is highest in India. Close proximity to the Middle Eastern crude oil sources would help the company in reducing turn-around time and crude freight costs. RPL will enjoy several fiscal incentives by virtue of being set-up in a Special Economic Zone.
48
RIL‟s expertise will be available for crude and other feedstock procurement, marketing of products, operation and maintenance of the refinery as well as risk management. Factors External to the Company As shown in the Industry analysis, the world economy is expected to grow at a CAGR of 3.9% per annum in terms of GDP on a purchasing power parity basis between 2002 and 2025. RPL is likely to benefit from this expected growth in world economy as there is a close co-relation between demand for petroleum products and economic activity. The company is also likely to benefit from the significant imbalances between demand and supply of different refined petroleum products that have developed in certain regions like the shortage of gasoline production capacity in the United States and the shortage of diesel fuel production capacity in Western Europe. QUANTITATIVE FACTORS Table 11: Comparison with Domestic Peers EPS (Rs.) P/E RONW (%) BOOK VALUE (Rs.) PRICE PER SHARE 562.48 46.33 734.28 177.10 67.62 232.96 RATIO OF PRICE PER SHARE TO BOOK VALUE 2.53 3.77 2.72 0.96 1.78 1.73
RPL IOCL HPCL BPCL MRPL RIL KRL BRPL CPCL
15.80 4.10 63.30 32.20 9.80 44.80
35.60 11.30 11.60 5.50 6.90 5.20
20.00 15.80 15.80 48.50 21.80 38.60 73.00 33.00
9.99 222.50 248.80 212.90 12.30 270.40 184.80 38.00 134.40
49
From the above table it can be seen that the ratio of the price per share to the book value per share for all the 8 companies ranges between 0.96 to 3.77 or we can say 1 to 4 (approx). Therefore, based on the findings above, the qualitative factors for the company and forecasted revenue based on internal calculations, the issue price was determined at Rs. 60 i.e., the issue price is 6 times the face value. The price band for the book building process was Rs. 57 to Rs. 62.
50
SEZ AND TAX BENEFITS
SEZs are those geographical regions of a country where the economic laws are more liberal as compared to the laws generally followed in the country. As per the requirements of the guidelines given by SEBI, the auditors of the issuing company should certify through the report about the tax benefits that would be available to the company and its shareholders are covered under the direct tax laws, Special Economic Zones Act, 2005 and Gujarat Special Economic Zones Act, 2004. The benefits available to RPL can be summarized as under: 1. The Company is entitled to deduction of 100% of the profits and gains from its unit set up in Special Economic Zone (SEZ) for a period of 5 consecutive assessment years and 50% of such profits and gains for further 5 consecutive assessment years. 2. The shares of the Company are not liable to Wealth Tax. 3. The Company is exempted from payment of Stamp Duty and registration fees payable on transfer of land within SEZ. 4. The Company is exempted from levy of stamp duty and registration fees o n loan agreements, credit deeds and mortgages executed by the SEZ or unit set up in the processing area of SEZ. 5. The Company is exempted from Sales Tax, Purchase Tax, Motor Spirit Tax, Luxury Tax, Entertainment Tax and other taxes and cess payable on sales and transactions within the SEZ. 6. The inputs (goods and services) purchased by the Company from Domestic Tariff Area shall also be exempted from sales tax and other taxes under the state laws of Gujarat.
51
STOCK MOVEMENT IN 2006
In an attempt to find out any movement in the stock prices of the competitors and the promoter of RPL due to the IPO of the company, a study of the competitor‟s stock prices has been done. Following are the stock movements of various related companies for year 2006 (the year in which the IPO of RPL has been made): Mangalore Refinery & Petrochemicals Limited (MRPL) The movement of stocks of MRPL over a period of one year starting from January 2, 2006 to December 2, 2006 has been constant except for a sudden increase followed by a steep fall around the month of May-June. Apart from this secondary fluctuation, the primary trend has been almost constant.
MRPL STOCK MOVEMENT
70 60 50 40 30 20 10 0
Figure 4: MRPL Stock Movement
Bharat Petroleum Corporation Limited (BPCL) The movement of stocks of BPCL over a period of one year starting from January 2, 2006 to December 2, 2006 has been constant except for a slight increase followed by a fall around the month of May-June. Apart from this secondary fluctuation, the primary trend has been almost constant.
52
BPCL STOCK MOVEMENT
600 500 400 300 200 100 0
Figure 5: BPCL Stock Movement
Hindustan Petroleum Corporation Limited (HPCL) The movement of stocks of HPCL over a period of one year starting from January 2, 2006 to December 2, 2006 has been constant except for a slight fall in the month of June followed by the recovery period till the month of September. Apart from this secondary fluctuation, the primary trend has been almost constant.
HPCL STOCK MOVEMENT
400 300
200
100 0
Figure 6: HPCL Stock Movement
53
Reliance Industries Limited (RIL) The movement of stocks of RIL over a period of one year starting from January 2, 2006 to December 2, 2006 has been constant except for a sudden increase followed by a fall around the month of May-June. Apart from this secondary fluctuation, the primary trend has been almost constant. The movement in the stock prices of RIL has been almost similar with the overall movement in the market.
RIL STOCK MOVEMENT
1400.00 1200.00 1000.00 800.00 600.00 400.00 200.00 0.00
Figure 7: RIL Stock Movement
Reliance Petroleum Limited (RPL) The movement of stocks of RPL over a period of one year starting from January 2, 2006 to December 2, 2006 has been constant.
54
RPL STOCK MOVEMENT
120 100 80 60 40 20 0
Figure 8: RPL Stock Movement
OVERALL MARKET TREND The primary trend of the market in the year 2006 was bullish in nature along with minor fluctuations in the opposite direction. During the months of May & June, there was a secondary trend which projected a bearish movement in the market. The stock market is generally influenced by many political and economical factors. During the period of secondary trend, the main incidents that took place were: The Left party took over in the state elections in Kerala & West Bengal. Pull out by foreign investors from the emerging markets like India, Taiwan and South Korea. Blowing up of the issue of higher education reservation. Hike in the fuel prices. Hiking the rates by 25 bps by RBI. During May-June, 2006, Sensitivity Index of BSE lost 28.14%. Experts said that the volatility during these two months was mainly due to the nervousness among the investors. Moreover, the apprehension related to the hike in the US Fed rates was pulling the fore ign investors away from the emerging markets like India.
55
This is the major cause of the secondary trend in the movement of the stock market during the two months and a related movement could be seen in the prices of the shares of the companies like HPCL, MRPL, BPCL and RIL. Conclusion: For the study of various related companies and the overall market trend, it is been concluded that the stock movements of the companies mainly depend on the overall market movement. In turn, the overall market movement depends on various economic and political factors. The IPO by Reliance Petroleum Limited does not show a major impact on the stock prices of the related companies because of the above stated factor. There is one more factor which explains the unrelated behavior of the stock prices of competitors and the IPO, that is: Oil and gas market in India is a supplier driven market and hence the entrance of a new competitor in the market does not pose a threat for other players in the market.
STOCK DATA AS ON APRIL 2, 2009
Market Capitalization is a measurement of corporate size of a public company. Thus capitalization can represent the public opinion of company‟s net worth. The stock price of Reliance Petroleum Limited along with the total number of outstanding shares helps in calculating the market capitalization of the company.
Figure 9: RPL Details
56
Current market price (April 2, 2009): Rs. 103.60 Reliance Petroleum Limited Key Data: Currency Fiscal Yr Ends Share Type Indian Rupees March Ordinary Market Capitalization Shares Outstanding Closely Held Shares 466,198,640,250 4,499,986,875 3,391,958,030
Where,
57
EXTERNAL COMMERCIAL BORROWINGS (ECB)
External Commercial Borrowings are being permitted by the Government for providing an additional source of funds to Indian corporate and PSUs (Public Sector Undertak ings) for financing expansion of existing capacity and as well as for fresh investment, to augment the resources available domestically. External Commercial Borrowings are approved within an overall annual ceiling, consistent with prudent debt management, keeping in view the balance of payments position and level of foreign exchange reserves. External Commercial Borrowings (ECBs) are defined to include commercial bank loans,
buyers‟ credit, suppliers‟ credit, securitized instruments such as Floating Rate Notes and Fixed Rate Bonds etc., credit from official export credit agencies and commercial borrowings from the private sector window of Multilateral Financial Institutions such as International Finance Corporation (Washington), ADB, AFIC, CDC, etc ECBs can be raised only through the internationally recognized source such as banks, export credit agencies, suppliers of equipment, foreign collaborators, foreign equity-holders, international capital markets etc. Any offers from non-recognized sources are not entertained. ECB can be accessed under two routes. They are – Automatic Route and Approval Route. Automatic Route External Commercial Borrowing for investment in real sector, especially infrastructure sector comes under automatic route. They don‟t require any RBI or government approvals. The maximum amount of ECB which can be raised under this route by an eligible borrower can be USC 500 million during a single financial year. However, NGOs engaged in micro- finance activity are allowed USD 5 million in a financial year. Approval Route Borrowings raised through this route require an approval from an Empowered committee set up by RBI. Any case which falls outside the purview of automatic route comes under approval route.
58
ELIGIBLE BORROWERS
AUTOMATIC ROUTE 1) Corporate registered under the Companies Act, 1956 (except financial intermediaries, such as banks, financial institutions (FIs) & housing finance companies), are eligible.
2) NGOs involved in micro-financing are also eligible, if they have a satisfactory borrowing relationship for at least 3 years with a scheduled commercial bank authorized to deal in foreign exchange.
3) Any individual, trust or non-profit making organization (except the NGO involved in micro- financing) is not eligible for raising ECB.
APPROVAL ROUTE 1) Financial institutions dealing exclusively with infrastructure or export finance are considered on a case by case basis.
2) Banks and financial institutions which had participated in the textile or steel sector restructuring package as approved by the Government are also permitted. However, they are only considered only to the extent of their investment in the package. Any ECB which has already been availed by the above stated entities are deducted from their entitlement.
3) NBFCs (Non Banking Financial Companies) are permitted to raise ECB under this route towards import of infrastructure equipment for leasing to infrastructure projects with a minimum average maturity of 5 years.
4) Foreign Currency Convertible Bonds (FCCBs) by Housing Finance Companies with strong financials satisfying criteria notified by RBI are permitted under the Approval Route. 5) Any other cases falling outside the purview of the automatic route limits.
59
RECOGNIZED LENDERS
Following are the recognized sources from where borrowers can take loans: 1) International banks, international capital markets, multilateral financial institutions (such as IFC, ADB, CDC etc.,), 2) Export credit agencies and 3) Suppliers of equipment, foreign collaborators and foreign equity holders. 4) NGOs engaged in micro- financing can also take loans from overseas organizations and individuals. But individual lenders from countries wherein banks are not required to adhere to Know Your Customer (KYC) guidelines are not permitted to extend ECB. „Foreign Equity Holder‟ is a foreign lender for ECB. They require minimum equity participation, in the capacity of equity holder, in the borrower‟s company. It is as follows:
1) If the ECB is up to USD 5 million, the overseas lender should directly hold minimum of 25% of the equity. 2) If the ECB is more than USD 5 million, the overseas lender should directly hold minimum of 25% of the equity and the debt-equity ratio should not exceed 4:1 (i.e. the proposed should not exceed 4 times the direct foreign equity holding). 3) If the debt-equity ratio exceeds 4:1 ratio, such case will be considered by RBI under Approval route.
60
AVERAGE MATURITIES FOR ECB
ECBs have the following minimum average maturities: 1) Minimum average maturity of five years for external commercial borrowings greater than USD 20 million equivalent in respect of all sectors except 100% Export Oriented Units (EOUs);
2) For external commercial borrowings of less than or equal to USD 20 million equivalent (for all sectors except 100% EOUs) has a minimum average maturity of three years.
3) 100% Export Oriented Units (EOUs) are permitted ECB at a minimum average maturity of three years for any amount.
61
RPL DEBT
Reliance Petroleum Limited, promoted by Reliance Industries Limited and Chevron India Holding Pvt. Ltd., rose at total of Rs157.5 billion. For this it has signed a Commercial Facilities Agreement with various Financial Institutions and Commercial Lenders. Following are the details of debt raised by Reliance Petroleum limited in the month of October in year 2006: Table 12: Debt Raised by RPL ECB / FCCB ECB ECB Borrowe r Equivalent Purpose Amount in USD 1,500,000,000 New Project 500,000,000 New Project Maturity Period (Aprox) 9 years 7 months 9 years 7 months
Reliance Petroleum Ltd. Reliance Petroleum Ltd.
RPL has raised debt through Approval Route..
62
INTERPRETATIONS
Commercial Lender Agreements The common terms agreements dates on or about the date of an agreement and entered into between the Commercial Facilities Agent and the Borrower. Availability Period It means the period from the Signing date to the earlier of the date falling 365 days after the Signing date and the date on which the Total Amount has been reduced to nil. Drawdown Date It is the date on which an Advance is made. Facility Office In relation to Commercial Facilities Agent or Commercial Lender, Facility Office is the Office identified with its signature in the debt agreement. LIBOR London Interbank Offered Rate (LIBOR) is a daily reference rate based on the interest rates at which banks offer to lend unsecured funds to other banks in the London wholesale money market . It is the interest rates that major international banks charge one another for loans. Majority Comme rcial Lende rs 1) Before any Advance has been made, a Commercial Lender or a group of commercial lender whose commitments amount in aggregate to more than sixty-six and two-third (66 2/3) per cent of the Total Available Amount. 2) After the Advances have been made, a Commercial Lender or a group of commercial lender to whom in aggregate more than sixty-six and two-third (66 2/3) per cent of the Total Loan is owed.
63
Transfer Certificate It is a certificate signed by the Commercial Lender and the Transferee whereby Commercial Lender seeks to procure the novation in favor of Transferee of rights and obligation under the Financial Documents. Weighted Ave rage Drawdown Date It is the date determined by the Commercial Facilities Agent on the earlier of i) ii) The date on which the Available Commitment is reduced to zero The last day of the Availability Period in accordance with below formula
Where N is rounded up to the nearest whole number. WAL is the weighted aggregate Advances being the aggregate of the number of days from the signing date to the date on which that advance was made, multiplied by the initial principal amount of the Advance. Purpose a) The Facility is intended to finance Project Costs. The borrower shall apply all amounts raised by it in compliance with the circulars on External Commercial Borrowings issued by the Reserve Bank of India and all other applicable laws and regulations of India. b) Without prejudice to the obligations of the borrower under the above clause, neither t he Commercial Facilities Agent, the Lead Arranger and the Commercial Lender nor any of them shall be obliged to investigate the application of amount raised by the Borrower. Nature of Comme rcial Lende r’s Right and Obligations 1) The failure of a Commercial Lender to perform its obligation shall not affect the obligation of the Borrower towards any other party. 2) Other party is not liable for the failure by a commercial lender to perform its obligation.
64
3) The amount outstanding at any time from the Borrower to each of the parties shall be a separate and independent debt. 4) If not expressly provided in the Finance Documents, each party is entitled to protect and enforce its individual rights.
Availability of the Facilities The Borrower may utilize facilities if: 1) The Commercial Facilities Agent receives a Notice of Drawdown from Borrower before a certain period from the proposed date for making the proposed Advances. The receipt of such notice obliges the Borrower to borrow the amount on a particular date subjected to certain terms and conditions. 2) The amount of Advance requested from each Facility is same when expressed in the percentage term of Available Amount from that Facility. 3) The proposed date of making the proposed Advances is a business day during the Availability period. 4) The aggregate principal amount of the proposed Advances is at least of a certain amount (Amount confidential to the company) or equal to Total Available Amount (if it is less than a certain amount (Amount confidential to the company)). If a Commercial Lender‟s Facility Commitment is reduced in accordance with the term mentioned in the agreement, after the Commercial Facilities Agent has received a Notice of Drawdown, then the amount of the proposed Advances shall be reduced accordingly. Security And Subordination The term loans from banks are secured by a first ranking pari passu mortgage over leasehold interests under the Land Lease Agreement and the fixed assets (including plant and machinery) of the Project of RPL; A first ranking pari passu charge over movable assets (other than current assets and investments) of the Project;
65
A floating second ranking charge over such of the company‟s current assets relating to the Project that are charged on a first ranking basis to the working capital lenders and an assignment of Company‟s right, title and interest under the key Project Agreements including agreements in respect of utilities. Inte rest Period The period for which an Advance is outstanding shall be divided into successive period each of which shall start on the last day of the preceding period. The duration of the interest period relating to an Advance can be one, three or six months (or such other period as the Borrower and the Majority Commercial Lenders may agree) as the Borrower may select in the Notice of Drawdown. If the borrower fails to give such notice of its selection of an Interest Period, the duration of that interest period shall be six months. These interest periods are required as it facilitate the Borrower to pay interest t o those Commercial Lenders who are listed by the end of the interest period. Inte rest 1) On the last day of each interest period the borrower shall pay accrued interest on the Advance. 2) After the start of each Interest Period, the Commercial Facilities Agent shall notify the Borrower of the amount of interest to be paid and the due date of payment. However, the failure of Commercial Facilities Agent to provide such notification, does not in any way affect on the Borrower‟s obligation to pay the same. 3) The date upon which the Borrower shall be obliged to make payments is determined in accordance with Common Term Agreement. 4) The rate of interest applicable to an Advance is the sum of LIBOR on the Quotation Date for the interest period and the Applicable Margin.
66
Repayment The Borrower shall repay the Loan by repaying (on each repayment date) an amount (confidential to the company) outstanding at the end of the Availability Period. Cancellation and Prepayment: 1) The Borrower may cancel the whole or any part of the Total Available Amount after giving a notice to Commercial Facilities Agent. This notice should be given in 7 business days in advance. Any cancellation under this clause shall reduce the Commercial Lender‟s Commitment on pro rata basis. 2) Any amounts undrawn at the end of Availability Period shall be cancelled and the Commercial Lender‟s Commitment shall be reduced to nil. 3) If the Borrower has given a notice of prepayment to Commercial Facilities Agent not less than 7 business days in advance, he has to prepay the whole or any part of Advance on the last day of Interest Period. 4) The notice of prepayment given by borrower is irrevocable in nature. It should specify the date upon which such repayment is to be made. 5) Any repayment made shall satisfy the borrower‟s remaining obligation under repayment clause. 6) The Borrower shall not repay or prepay all or any part of any Advance except at the times and in the manner expressly provided in agreement. The borrower shall not be entitled to reborrow any amount repaid or prepaid. 7) The Borrower is liable to obtain all approvals from all relevant authorities in India that may be required in connection with any repayment, prepayment or cancellation. This approval has been obtained by borrower at its own cost. Taxes The financing parties should receive a net amount which shall be free from any deductions based on tax. It implies that the amount receivable by the financing party should be before deducting the taxes.
67
If the financing party is required to make any payment on account of tax which should have been paid by the borrower, the borrower should indemnify the financing party with the required amount. The financing parties or its affiliates are not obliged to disclose their tax policies to the borrowers. When the financing party claims for indemnification, it should use its reasonable efforts to file the required documents and forms so as to avoid the increase in the amount and should notify the other party as well. The documents required to be done includes the fo llowing: a. Tax residency certificate issued by the tax authority of the country in which the Financing Party is the resident. b. Certificate of incorporation of the financing party. c. Certificate confirming no permanent establishment in India or if a permanent establishment exists in India, then a certificate confirming that income under the Finance documents is not attributable to the permanent establishment in India. d. Authority letter under Section 195 of the Income Tax Act 1961 of India authorizing the Borrower to make an application for an exemption from that Section 195. e. Any duly authorized translated copies of any original tax residency certificate or certificate of incorporation which is in a language other than English. Increased Cost If, by reason of any change in law or in its interpretation of for compliance with the request of central bank, the Commercial Lender or any holding company of Commercial Lender incurs an increased cost because of entering in to the agreement or performing obligations of this agreement or maintaining a commitment under this agreement then the Borrower shall from time to time (on the demand of Commercial Facilities Agent) pay to the Commercial Facilities Agent for the account of increased cost of Commercial Lender. No Commercial Lender is permitted to recover
68
1) Any cost which has accrued more than 180 days prior to the receipt by the Commercial Facilities Agent of the notice regarding the cost incurred. 2) Any cost which has accrued due to own gross negligence or willful misconduct.
Any Commercial Lender intending to pursue such claim should notify the Commercial Facilities Agent of the event by reason of which it is entitled to do so and a statement as to whether or not the Commercial Lender is making such claims against other Borrowers. Substitution of Commercial Lender: In the event that any Commercial Lender shall claim payment of any cost referring to Tax Gross up, Tax Indemnity or increase in cost, the Borrower has the right (provided that no Event of Default has occurred) to replace such Commercial Lender with another Commercial Lender or other Financial Institution. These replacements should be reasonable acceptable to the Commercial Facilities Agent and no other Commercial Lender should be liable to replace the claiming Commercial Lender. The claiming Commercial Lender to be replaced shall (on the date as specified by the Borrower in the written notice to the Commercial Lender) transfer the rights, interest and the other entire amount owing to such Commercial Lender. The relevant Commercial Lender should provide a Transfer Certificate for the same. However, if the claiming Commercial Lender fails to comply with its obligation, the Borrower shall not be obliged to make any payment to such Commercial Lender in respect of any cost or liability accruing after the date specified in the written notice. Illegality If, at any time, it is unlawful for a Commercial Lender make Advance to the Borrower, then Commercial Lender (after becoming aware of such fact) should deliver a certificate to the Borrower through Commercial Facilities Agent and 1) Such Commercial Lender shall not be obliged to make Advance to the Borrower thereafter and amount of its commitment shall immediately be reduced to zero.
69
2) The Borrower shall repay such Commercial Lender‟s share of outstanding Advance, the accrued interest and all other amounts owing to such Commercial Lender. Such payments should be made within 30 days of the date of that certificate.
Mitigation If, in respect of any Commercial Lender, circumstances arise which result in: 1) An increase in amount of payment to made to it under Tax Gross-up 2) A claim of indemnification pursuant to Tax Indemnity 3) The prepayment of part of the Advance under clause of Illegality Then, without in any way limiting the obligations of the Borrower, the commercial lender shall notify the Commercial Facilities Agent (who in turn shall notify the Borrower). Then, in consultation with Commercial Facilities Agent and the Borrower, the Commercial Lender should take steps to mitigate the effect of such circumstances. These steps might include transfer of Commercial Lender‟s Facilities Office to another jurisdiction or transfer of its right and obligation to another financial institution who is willing to take part in the Advance. This can only happen when Commercial Lender is under no obligation to avoid such step as in its bona fide opinion it would have an adverse effect on its business or financial condition. The above stated steps can only be taken when Commercial Lender is not obliged to disclose any information relating to its business Event of Default The following events describe the circumstances which constitute an Event of Default for the purpose of Finance Documents: 1. Failure to pay: The relevant amount should be paid in full within 5 business days of the due date for payment of such amount. 2. Misrepresentation: Any statement which is incorrect or misleading should be altered to the reasonable satisfaction of the Required Majority Commercial Lenders within 30 days of the notice from the Commercial Facilities Agent to the Company.
70
3. Financial Requirements: The Company‟s financial condition as at each 31 March and each 30 September commencing 12 months after the Commercial Operations Date do not satisfy the required ratio requirements. 4. Completion Date: The Completion Date does not occur on or before 31 December 2010. 5. Winding-up: The Company takes any action to start its winding up, dissolution or reorganization on liquidation. 6. Unsatisfied Judgment: A judgment, decree or order made against the Company is not stayed or complied with within 90 days and the failure to comply with such judgment, decree or order is reasonably likely to have or result in a Material Adverse Effect. 7. Changes of Control: At any timea. RIL and/or any of its Affiliates cease to be the direct legal and beneficial owner of at least 51% of the issued and outstanding equity share capital of the Company, or b. RIL (either directly or indirectly through its Affiliates) ceases to have control of the Company. 8. Repudiation: a. RIL repudiates the RIL Undertaking. b. The Company repudiates any Finance Document to which it is a party. c. Any counterparty to any Key Project Agreement or any other Agreement and such repudiation by the counterparty would have or result in a Material Adverse Effect. 9. Consents: Any Consent is suspended, cancelled, revoked, forfeited, surrendered or terminated or is varied or modified and such event will have or result in a Material Adverse Effect. 10. Abandonment: For any reason the Company abandons the construction or the operations and maintenance of the Plant. The advance and the other amount owed by the Borrower to the financial parties shall become immediately due and payable upon the declaration by Commercial Facilities Agent. In the Event of Default Commitment of each Commercial Lender shall be cancelled. Default Interest and Break Cost: Default Interest Period: If any sum due and payable by the Borrower to a Financing Party under the Agreement is not paid on the due date it becomes the unpaid sum. The period beginning on
71
due date and ending on the date upon which the obligation of the Borrower to pay such sum is discharged is divided into successive periods. The duration of each is selected by the Commercial Facilities Agent and these periods are called Default Interest Period. Default Interest: During each default period, the unpaid amount shall bear interest at the rate per annum.
If, for any such period, LIBOR cannot be determined then the rate per annum applicable to unpaid sum will be
This rate decided by Commercial Facilities Agent is the arithmetic mean of the rates notified by each Commercial Lender to the Commercial Facilities Agent before the last day of such Default Interest Period. Payment of Default Interest: Any interest accrued in respect to unpaid sum shall be due and payable by the Borrower at the end of the Default Interest Period or on any other date as specified by Commercial Facilities Agent in a written notice to Borrower. Notification of Default Interest: the Commercial Facilities Agent should promptly notify the Borrower and the Commercial Lender of determination of any default interest. If the calculation of Interest is error free then it would be conclusive and binding by both Borrower and Commercial Lender. Break Cost If the Commercial Lender receives any part of unpaid sum on the last day of an Interest Period, the Borrower shall pay an amount by which 1) The additional interest which would have been payable on the amount so recovered exceeds 2) The additional interest which in the reasonable opinion of the Commercial Facilities Agent ,
72
would have been payable to the Commercial Facilities Agent on the last date of the interest period in respect of a deposit of the unpaid sum equal to the amount so received. Indemnity Borrower‟s Indemnity: The Borrower undertakes to pay Commercial Facilities Agent: 1) An amount sufficient to indemnify each Financial Party against any reasonable cost which it may incur as a consequence of the occurrence of any Event of Default. 2) An amount sufficient to indemnify each Commercial Lender against any loss it may suffer as a result of its funding a portion of Advance. The Financial Parties should not assert against any director, officer or employee of the Borrower for any claim it has against Borrower. Currency of Account and Payment: A currency of Account and Payment is explicitly mentioned in the agreement. Payment On the date on which certain amount is to be paid by the Borrower to the Financial Parties, the Borrower shall make available to the Commercial Facilities Agent the due payment in the currency earlier mentioned in the agreement. Alte rnative Payment Agreement If, at any time, it shall become impracticable for the Borrower to make any payment then: 1) The Borrower may agree with the Financing Party an alternative arrangement under Indian Law. 2) The Borrower is obliged to notify the Commercial Facilities Agent of the agreement so reached between it and the Financial Party. No Set off: All payments required to be made by the Borrower under the Financial Document shall be calculated without any set-off or counter claim.
73
Claw back: When the sum is to be paid to the Commercial Facilities Agent for account of another person, the Commercial Facilities Agent is not obliged to pay the amount to the person unless it is sure that it has received the requires amount. Payment on Business day: If any payment is due on a non-business day then the payment should be made on the succeeding business day (if it is on the same month of the calendar) or on the preceding business day. Sharing Among the Comme rcial Lende rs If the amount recovered by the Commercial Lender from the Borrower is more than the amount it should have received 1) It shall pay the Commercial Facilities Agent an amount equal to such excess amount 2) The Commercial Facilities Agent should treat this a mount as the one received from the Borrower and shall pay it to the entitled person. Fees Commercial Facilities Agent: The Borrower shall pay to the Commercial Facilities Agent for its own account the agency fees specified in the letter of even date on a specific date as mention in letter. Commitment Fees: a) The Borrower shall pay Commercial Facilities Agent a commitment fees computed at the rate of certain per cent per annum on the Commercial Lender‟s commitment from day to day during the period beginning on the date falling 120 days after the Signing Date and ending on the last day of the Availability Period. b) Commitment fees shall be payable on the date falling six months after the Signing Date and thereafter quarterly in arrears until the last day of the Availability Period.
74
Cost and Expenses The Borrower from time to time (on demand of the Commercial Facilities Agent) reimburse the Commercial Facilities Agent and the Lead Arranger (on presentation and delivery of all details, invoice for all reasonable costs and expenses) incurred by it in connection with the proceeds of the Advance arrangement. The Borrower shall pay all Indian Stamp, registration and other taxes. Commercial Lender’s Liability to Cost If Borrower fails to perform any obligation in above stated clause, each Commercial Lender in proportion borne by its share of the Advance to the total amount of Advance, indemnify the Commercial Facilities Agent against any loss incurred by it as a result of failure from the Borrower. Assignment and Transfer by the Borrowe r The Borrower is not entitled to assign or transfer rights, benefits or obligation except on case of merger or consolidation. Assignment and Transfer by Comme rcial Lender Commercial Lender may at any time assign its rights and benefits under the Financial Document. For this they need the prior approval from Borrower unless: 1) Borrower has not shown its concern within 10 days after the Commercial Lender‟s request for the same. 2) Or an Event of Default has occurred. If a Commercial Lender wishes to transfer all its rights and interests, then it may effective after the delivery of a duly completed Transfer Certificate to the Commercial Facilities Agent and the Borrower. On the date at which the novation takes effect, the Transferee in respect of such transfer is liable to pay to the Commercial Facilities Agent for its own account transfer fees of certain amount (Confidential to the Company)
75
If the Commercial Lender is to be merged with any other person, such Commercial Lender at its own cost shall furnish Commercial Facilities Agent with: 1) An original copy of a legal opinion by a qualified legal counsel confirming that all such Commercial Lender‟s assets, rights and obligations have been duly vested in the succeeding entity. 2) A original copy of a written confirmation by legal counsel acceptable to the Commercial facilities Agent that English Law and the law of the jurisdiction in which the Facility Office of such Commercial Lender is located recognize such merger under the relevant foreign laws 3) A duly executed Transfer Certificate If the Commercial Lender, following any merger, does not comply with the requirement under then the Commercial Facilities Agent shall have the right to decline to recognize the succeeding entity. General Agency Provision Commercial Facilities Agent is not the trustee of any other person as per the Agreement. None of the Lead Arranger or the Common Facilities Agent is bound to account to any other Financial Party for any sum received by it for its own account. The Common Facilities Agent, the Lead Arranger and the Commercial Lenders: Each Lead Arranger and each Commercial Lender appoints the Commercial Facilities Agent to act as its agent in connection with the Finance Documents. The Commercial Facilities Agent may assume that 1) Any representation made by the Borrower or any other person in connection with any Transaction Document is true 2) No Event of Default has occurred 3) No change is circumstances has occurred 4) None of the parties under transaction document is in breach of its obligations 5) Any right, power vested upon the Commercial Lenders has not been exercised
76
unless its agency department has, in its capacity as agent for Commercial Lender, received actual notice to the contrary from any other party. The Commercial Facilities Agent shall: 1) Promptly inform each Commercial Lender of the content of any notice or document received by it in its capacity as Commercial Facilities Agent form the Borrower 2) Promptly inform each Commercial Lender of the occurrence of Event of Default, change of circumstance. Commercial Facilities Agent and the Commercial Lender are not bound to: 1) Account for any Commercial Lender for any sum received by it for its own account 2) To disclose to any other person any information if such disclosure might constitute a breach of any law Each Commercial Lender from time to time indemnify the Commercial facilities Agent in the proportion of its share of the Advance against any cost, claim, losses which Commercial Facilities Agent may incur. If the Commercial lender owes any amount to the Commercial Facilities Agent under the agreement, the Commercial Facilities Agent may (after giving a notice to that party) deduct the amount from the payment it is obliged to pay to that party. The Commercial Facilities Agent may resign from its appointment at any point of time without assigning any reason by not giving less than thirty days prior notice. But no such resignation shall be effective until a successor for the Commercial Facilities Agent is appointed. If the Commercial Facilities Agent gives resignation then any reputable Commercial Lender or
other financial institution may be appointed as a successor with the prior approval of the Borrower. But if no successor is appointed within the notice period, then Common Facilities Agent may appoint such a successor itself. Majority Commercial Lenders may remove the Commercial Facilities Agent from its appointment as agent at any time by giving not less than 30 days prior notice.
77
If the Majority Commercial Lenders remove the Commercial facilities Agent from its appointment any reputable Commercial Lender or other financial institution may be appointed as a successor with the prior approval of the Borrower by the Majority Commercial Lenders. If no Commercial Facilities Agent accepts its post as a successor by the date falling after 60 days of Commercial Facilities Agent‟s resignation, the resignation shall nevertheless become effective from that date and the Commercial Lenders shall perform all duties of the resigning Commercial Facilities Agent. In acting as Commercial Facilities Agent for the Commercial Lender, the Commercial Facilities Agent‟s agency division shall be treated as a separate entity from any other of its division or department. Calculations & Certificates Any interest or fee accruing under a Financial Document will accrue from day to day and is calculated on the basis of the actual number of days elapsed and a year of 360 days, provided that interest or fees in respect of any day shall accrue only once. The Commercial Facilities Agent shall maintain on its books a control account in which the following shall be recorded: i. The amount of the advance made or arising hereunder and each Commercial Lender‟s share therein, ii. The amount of all principal, interest and other sums due or to become due from the Borrower to any of the Commercial Lenders hereunder and each Commercial Lender‟s share therein and, iii. The amount of any sum received or recovered by the Commercial Facilities Agent hereunder and each Commercial Lender‟s share therein. Confidentiality The Commercial Facilities Agent and Commercial Lenders may disclose to any potential Transferee who may be considering entering into contractual relation with the Commercial Facilities Agent or Commercial Lenders only if
78
1) Required and permitted by applicable law or applicable regulation 2) Required for accounting purpose if the professional adviser execute and deliver a confidentiality agreement. 3) It is in connection with any legal proceedings taken against Borrower 4) It is in public domain Amendments and waivers The Borrower and the Commercial lenders should agree in writing to opt for any amendment, variation or waiver in the Commercial Facilities Agreement. Any such amendment, variation, supplement or waiver which changes or is related to the rights or obligations of the Commercial Facilities Agent shall require its agreement regarding the same. Decision Making If any consent, approval or decision is required by some or all the Commercial Lenders, then Commercial Facilities Agent shall promptly upon becoming aware of the requirement, advise each Commercial Lender specifying whether the decision is to be taken by all the Commercial Lenders or the Majority of the Commercial Lenders and shall also specify the time within which the decision should be taken. Each decision taken in accordance of the above stated clause shall be promptly notified to all the other parties by Commercial Facilities Agent. Governing Law The agreement also explicitly mentions the governing law that will be applicable for the agreement. Financial Condition and Covenants 1. FINANCIAL CONDITIONS Ratio Requirements: The Company shall ensure that its financial condition as at each 31 March and each 30 September commencing after the Commercial Operations Date, as evidenced by a Compliance Certificate and, in the case of its financial condition as at 31 March in any year.
79
The following ratios should be maintained as per the requirements of the Agreement: a. Tangible Net Worth b. Ratio of Total Long Term Secured Debt to Total Fixed Assets c. Ratio of Earnings Before Interest, Taxes, Depreciation and Amortization (EBITDA) d. Ratio of Total Long Term Debt to Tangible Net Worth e. The Debt Service Coverage Ratio 2. POSITIVE COVENANTS a. Construction, Operation and Maintenance: The Company shall ensure that in all material respects the Plant is designed, engineered, operated, maintained and repaired in accordance with Good Industry Practice and all the material Applicable laws. b. Obligations under Project Agreements: the Company shall comply with, in all material respects all of its obligations under all the Agreements. c. Security Documents: the company shall within 180 days of the signing date, execute and deliver the necessary documents to the Security Trustee. d. Application of proceeds: the company shall apply the proceeds of all utilizations under each of the commercial facilities agreement only for purposes permitted under the commercial facilities agreement. e. Compliance with law and environment standards: the company shall comply in every material respect with all the material applicable laws. f. Consents: the company shall obtain on a timely basis and do all that is necessary to maintain in full force and effect, all consents required by it and make all filings, notifications and notarizations, in each case, which at any time or from time to time it is required to obtain or make. g. Taxation: the company shall file all tax returns required to be filed by it and promptly pay all taxes to which it is assessed liable as they fall due.
80
h. Corporate existence: the company shall do or cause to be done all things necessary to preserve and keep in full force and affect its corporate existence and authority to conduct its business. i. Accounting systems: the company shall maintain adequate accounting, management information and cost accounting systems for the Project and shall engage Auditors to audit the financial statements annually. j. Insurance: the company shall effect and maintain or cause to be affected and maintained in full force and effect contracts and policies of insurance as stipulated in the agreements. 3. NEGATIVE COVENANTS a. Change of business: the company shall procure that no substantial change is made to the nature of its business. b. Shares: the company shall not issue any new shares or alter any rights attaching to any of its shares if the result of so doing would be that RIL and/or its affiliates cease to be the beneficial owners of at least 51% of its issued share capital. c. Investments: the company shall not make any investments or acquisitions in any unrelated business ot her than from the company‟s retained earnings or from the additional equity amount provided that such investment in any unrelated business is not of a nature that may result in the company incurring liability beyond the loss of the investment or acquisition itself. d. Mergers: the company shall not voluntarily take any steps intended to result in its merger, amalgamation or consolidation with any other person unless the legal entity into which the company is merged or consolidated agrees in writing with the commercial facilities agents that it will assume all the obligations and liabilities of the company under the finance documents and the permission of all the commercial lenders has been obtained.
81
e. Financial indebtedness: the company shall not incur or allow to remain outstanding any indebtedness for borrowed money other than permitted financial indebtedness.
82
PORTFOLIO TRACKER VERSION 0.0
Portfolio Tracker is software which is freely available on internet by different financial sites. I have made an effort to create similar software which can be used to keep a track of portfolio and which will also tell the user about any arbitrage opportunity which is available due to price variation at NSE (National Stock Exchange) and BSE (Bombay Stock Exchange) at different point of time during a single day. INTRODUCTION Portfolio Tracker is software that. GOAL AND OBJECTIVE 1) To calculate the profit or loss of an investor on a given portfolio based on the current market situation. 2) To find out the arbitrage opportunity based on price difference at NSE and BSE market. 3) As every investor does not have time to regularly keep a track of their portfolio at a single point of time, this software helps them to watch and monitor their portfolio as soon as they open this excel sheet along with the internet connection. STATEMENT OF SCOPE The software takes name of the share, total number of shares bought and the average purchase price as input and fetches the current market value of those shares. The software calculates the total profit or loss and percentage of the same. We can also extend the scope according to the use of investor or personalize the working of the program according to the needs of investor. Portfolio Tracker also compares the NSE and BSE prices every minute and recommends the arbitrage strategy to the user. We can also extend the scope of same for currency hedging. The Software is password protected; hence the user can prevent mishand ling of his personal portfolio and can keep the data confidential.
83
MAJOR CONSTRAINT 1) The software updates the prices at particular interval; hence the change in prices for less than a minute cannot be accommodated in it. This might become a shortcoming while deciding the arbitrage strategy as the stock prices changes at every fraction of seconds. 2) The software requires internet connection. 3) It takes average purchase price as an input instead of taking different purchase quantity at different purchase prices. USER PROFILE A common man dealing in share market can use the software for personal portfolio tracking. FUNCTIONAL MODEL AND DESCRIPTION 1) Profit and Loss Calculation: First function of the software is used to calculate the profit and loss for the entered portfolio of the user. This function requires following inputs from the user: a. Script code for BSE and script name for NSE b. Buy Date c. Total number of shares bought d. Average Buy Price Once these inputs are available to the software, it takes the name of the share and matches it which the data available from the NSE site using the VLOOKUP function. Buy price is compared which the current market price and If (Current Price > Buy Price) { GAIN; } Else { LOSS; }
84
Amount of Gain (Loss) per share is calculated by following formula: – Total Gain (Loss) is calculated by multiplying the Gain (Loss) per share with the total number of shares bought.
Percentage of Gain (Loss): Annualized:
Absolute:
VLOOKUP Function: The V in VLOOKUP stands for vertical. It searches for a value in the first column of a table array and returns a value in the same row from another column in the table array. Syntax: VLOOKUP (lookup_value, table_array, col_index_num, range_lookup) Where, Lookup_value is the value to search in the first column of the table array. Table_array is the array from where the value is to be matched. The first column of table_array is searched for lookup_value. Col_index_num is the column number of table_array from where the value is to be fetched.
85
Range_lookup can have any of the two values (False or True). For exact match false is used and true is used for approximate match, where the next largest value that is less than lookup_value is returned. 2) Recommendation of Arbitrage Strategy: An arbitrage. The function of the software recommends the users an arbitrage opportunity for his portfolio. It compares the NSE and BSE stock prices for shares of the portfolio and based on those prices it calculates the Arbitrage opportunity. If (NSE Price == BSE Price) { Arbitrage Opportunity = NO } Else { Arbitrage Opportunity = YES } Arbitrage Strategy: Now, following function is used for calculating the recommendation for arbitrage strategy: If (Arbitrage Opportunity == YES) { If (NSE Price > BSE Price) { Arbitrage Strategy = BUY BSE, SELL NSE; } Else { Arbitrage Strategy = BUY NSE, SELL BSE; } Else { Arbitrage Strategy = “- “; }
86
Arbitrage Gain/Loss: The software also calculates the Gain (Loss) per share because of the arbitrage. Following formula is used for the same:
ROAD AHEAD The software is currently launched with basic features of portfolio tracker. It can be further enhanced by making changes. Few of the enhancements suggested are: 1) Instead of the taking the hard coded value for the purchase price, a function can be developed which takes the purchase price and number of shares purchased as input and thereby calculate the Average Price. This will help the user the update his current purchase and hence automatically calculate his new Average Purchase Price. 2) This can be extended to monitor currency prices and take necessary actions at desired point of time to make profits. 3) Through this technique, we can not only calculate or monitor the stock prices but can also use it to retrieve data that can automatically be refreshed and takes no time to monitor the changes taking place.
87
MERGER OF RPL WITH RIL
Consolidation or amalgamation can be termed as the merger of two or more companies into one.". Therefore, the two concepts are, substantially, the same, but the term amalgamation is more common. The merging of two or more entities into a single entity is known as amalgamation. Such actions are commonly voluntary in nature and involve stock swap or cash payment to the target. Stock swap is often used as it allows the shareholders of the two companies to share the risk involved in the deal. There are certain reasons which are the major sources of motivation for the merger of two entities. These reasons are also common to the merger of RIL-RPL.
Synergy : It refers to the fact that the combined entity can often reduce the fixed costs and other operating costs by removing duplicate departments or operations, thereby, lowering the costs of the company relative to the same revenue stream, thus increasing profit margins. The merger of RIL-RPL is also said to provide the merged entity with the financial and operational synergies. This is because the nature of the existing refinery and the new refinery is same and also the improved capacity and the complexity would give RIL the necessary reduction in the costs of operations.
Increased revenue or market share : RIL will be amalgamating its subsidiary company (RPL) and by doing this, it would be increasing its market power by capturing increased market share to set prices.
Economy of scale : The merged entity would result in the creation of the refinery which would be the world‟s largest refining capacity at a single location. The increased complexity and capacity of the new refinery would help the company to maximize the GRM.
88
Taxation: Availing the tax benefits is another major incentive for the two companies to opt for merger. But in the case of the merger of RIL-RPL, the merger will be tax neutral as after the merger both the entities will continue to enjoy the same tax benefits which they are currently enjoying. Therefore, on a consolidated basis, RIL will get the benefits of being an Export Oriented Unit and RPL will enjoy the SEZ benefits.
In 2002, RPL, which was the first refinery project, merged itself with its parent company RIL. Now after 7 years, RPL is once again merging with RIL. Present RPL was incorporated in October 24, 2005 to set up the second mega refinery complex. Though the market was surprised by the announcement of the merger between RIL and its subsidiary RPL, it was expected to happen owing to the strategic fit and the changes in the global scenario. The merger of RPL with RIL seems to be a strategic move as it would enable RIL to enjoy Economies to Scale in production and refinery of petrochemicals. It would also help in minimizing the cost of capital and capitalize on the cash flow of RPL. The merger of Reliance Industries Limited (RIL) and Reliance Petroleum Limited (RPL) will result in the creation of a petrochemical behemoth which would encompass the entire value chain in the business of petroleum giving the merged entity the necessary clout to take on the competition on the national and international level. According to media reports, the merger will create a new entity which will have a combined market value of about Rs.233,384 crores. Reliance Industries Limited, after the merger, will become the world‟s largest refining capacity at a single location. Also, it would become the fifth largest polypropylene manufacturer. The two firms after the amalgamation will continue to function as separate entities from the accounting point of view and therefore, the tax benefits available to RIL as an Export Oriented Unit and to RPL as a Special Economic Zone will be independent from each other. In fact, this merger will give RIL greater flexibility in operational planning. RPL had its IPO in April 2006. At that time RIL had 75% stake in the company. RIL and Chevron were the joint promoters of the company with a stake of 15% and 5% respectively. But in November 2007 RIL sold its 4.01% equity stake in Indian Stock Exchange. With this RIL was left with 71% of equity shares in the company.
89
According to the contract between Chevron and RIL, Chevron has an option to hike its stake up to 29%. For this it will have to buy 24% equity shares from RIL. This leaves RIL with only 47% (71-24) stake in RPL, hence reducing it below 51%. Now, let us look at the probability of Chevron hiking its stake. Following are the reasons, which make it highly unlikely: Very High Cost to Chevron: If Chevron raises its stake to 29% then it has to buy the shares from RIL at 5% discount to market price. The amount involved would be very costly for the company, which would be almost the cost of RPL refinery at the current market prices. Hence to raise its stake by 29% it has to bear almost complete refineries cost. RIL’s stake sales: When in November 2007 RIL sold off its 4.01% stake in RPL, it became further more unlikely for Chevron to raise its stake in RPL. As it will leave RIL with less than 51% shares in Reliance Petroleum Limited. Now, it cannot be the case that RIL does not mind letting its stake go below 51% as the debt agreement clearly states that an Event of Default will occur if RIL‟s share goes below 51% in RPL. RPL can source crude on its own: As the existing refineries of RIL, RPL also seems confident of sourcing crude on its own. Hence the hike of stakes of Chevron seems rather unlikely. From the above stated points we can see that in November 2007 only it became almost clear that RIL might take a step of merging RPL. RIL has had a history of merging its subsidiaries involved in refining petrochemicals. The old RPL which started its operation in FY01 was finally merged with RIL in FY02. IPCL (acquired by RIL in FY03 as a part of privatization in India) was merged in RIL in FY07. Thus, when on March 2, 2009 RIL finally announced the merger of RPL into RIL, it shouldn‟t have come as a shock to the market.
90
SYNERGY OF THE MERGER
RIL to maintain its refining margin well ahead of its competitors, it has contracted with crude oil suppliers for the cheaper heavy sour crude. Moreover, the product blends from the two refineries will also help RIL to produce fuel which matches the Euro 4 and Euro 5 grades, which is a requirement of the western markets.
The RPL refinery will be one of the most complex refineries in the world with a Nelson Complexity Index of 14.0 which will enable the refinery to process various varieties of crude to produce superior quality products which are able to meet the stringent specifications and command price premiums. This is a significant competitive advantage in the current industry scenario of increasingly heavy and sour new crude finds. As the location of the two refineries is adjacent to each other, it provides a strong base for the company to explore the locked synergies between the two refineries. The ability of the merged entity to buy and process different forms of crude (including the heaviest crude) owing to the improved complexity will help the company in lowering its buying cost thereby increasing the overall GRM.
Apart from this, the merged entity will also capitalize on shipping freight flexibilities to overcome the hurdles posed by the Indian customs authorities that do not allow two companies to load products in a single vessel. Therefore, the merger will help the merged entity in unlocking the financial and the operational synergies that exist between RIL and RPL.
The merger would enable RIL to have improved Cash Flows and Balance Sheet along with a lower cost of capital which would in long run benefit the shareholders of the merged entity. The RIL-RPL merger will also result in the optimization of the supply chain as the combined refining capacity of the company will be 1.24 million barrels every day and the ability to deal with the supply chain, the ability to steadfast in transportation is more for the merged entity. Also, the merged entity will save on the dividend distribution tax paid by RPL on distributing dividends at 17.99% to its shareholders.
91
SWAP RATIO
The Board of Directors has approved the merger of Reliance Petroleum Limited with Reliance Industries Limited for a swap ratio of 1:16. This means that for every 16 shares of RPL, 1 RIL share would be issued. The RIL board has approved a scheme of amalgamation of Reliance Petroleum with the company under the provisions of Sections 391 to 394 of the Companies Act of 1956. Under this scheme, the RPL‟s shareholders will get 1 fully paid equity share of Rs 10 each of RIL for every 16 fully paid equity shares of RPL held by them. The merger follows the philosophy of RIL of creating enduring value for its stakeholders. CRISIL has affirmed an AAA rating for RIL post merger. Merger, however, will reduce the shareholding of institutional investors, while banks and mutual funds will get a higher share in the merged entity. According to some analysts, the merger would result in the fall in the promoter holding by 2% from 49% to 47% which would be due to the cancellation of the treasury stock. The merger would also lead to an increase in the retail shareholding from 16.1% to 19%. The swap ratio of 1:16, which is marginally in favor of the shareholders of RPL, would mean a dilution of 4.4% of RIL‟s equity. The treasury stock which has been created as a result of merger would be extinguished which would prove to be positive for the shareholders of RIL. Therefore, the merger can be said to have a neutral effect on the shareholders of both the company. After the swapping of shares, RIL will have 3.7 million shareholders and the promoter‟s holding would fall to 47%. RIL is currently holding 70.38% in RPL which would be cancelled on absorption. If the swap ratio would have been 1:15, RIL would have to issue 8.89 crore equity shares as against the outstanding 133.3 crore of RPL shares. As a result, RIL‟s equity would have risen to Rs 1,662.65 crore, which amounts to a dilution of just 5.6%. Similarly if the swap ratio would have been 1:17, the dilution of the promoter‟s share in RPL would have been 1.95%.
92
Table 13: Dilution of Promote rs
PARTICULARS
Total number of RPL Shares RIL's Stake (including Chevron's) Non-Promoters Stake Number of RPL Shares for 1 RIL Share Number of Shares to Promoter Number of Shares to Non-Promoters Total number of RIL Shares before merger Number of new RIL Shares to be issued in lieu of RPL Treasury Shares to be cancelled Total number of RIL Shares post merger Current Treasury Shares in RIL Net Equity excluding Treasury (Pre-Merger) Net Equity excluding Treasury (Post-Merger) Existing Promoter's Stake Promoter's Stake Post Merger Dilution of promoter's share due to Merger
(%)
75.38 24.62
In Crores
450.00 339.21 110.79 16 21.200625 6.924375 157.30 28.125 21.200625 164.224375
12.64
19.88272 137.41728 144.341655
49.03% 46.96% 2.07%
77.12419 77.12419
The total number of shares of RPL is 450 crores which are represented by the promoter‟s stake which is equivalent to 339.21 crores which accounts for 75.38% of the total number of shares. The remaining 24.62% of the shares are held by the non promoters which is equal to 110.79 crore shares. The swap ratio as decided by the Board of Directors of RIL is 1:16 which implies that for every 16 shares of RPL, 1 share of RIL will be issued. Thus, the number of shares to be issued to the promoters is 21.2 crores and to the non promoters is 6.92 crore shares. The total number of shares of RIL before the merger are 157.30 crores and after the merger the total number of shares will be 164.22 crore shares. This figure includes 28.125 crores shares
93
representing the new shares that will be issued in lieu of the shares of RPL and 21.2 crore shares representing the treasury stock which will be cancelled by RIL. Currently, the total number of treasury shares in RIL is 19.88 crores. Hence, after the merger, the promoter‟s stak e will be 46.96% instead of 49.03%. it means that the equity will be diluted due to the merger to the extent of 2.07%.
94
INVESTOR’S POSITION
It is believed that the RPL shareholders would benefit in the long term from the merger with RIL. This gain would be due to the following reasons:
1. Gains from upside in the petroleum sector, retail and SEZ: RPL shareholders, after the merger will be gaining from the upsides from RIL‟s petroleum business after merger as the company will be gaining from the strong value accretion due to exploration of the large unexplored acreage in the highly prospective areas. The shareholders of the merged entity will even gain from the upsides in RIL‟s diversification in the organized retail sector and SEZs. 2. Rise more likely in RIL’s earnings: A steady rise in RIL‟s earnings is more likely than in RPL‟s earnings. There has been a secular rise in RIL‟s earnings since its inception in 1964. Because there are a number of subsidiary companies under the group, any decline in earnings of one subsidiary is set off against the increase in the earnings of the other subsidiaries. That is, RIL‟s earnings decline in any one business is neutralized by strong growth in other businesses. This would not have been possible in the case of RPL as its earnings are mainly from refining only. 3. Diversification of risk of minority shareholde rs: The merger will diversify the risk of minority shareholders of RPL as it would be shifting from standalone refinery to an integrated unit like RIL. Due to the downturn of the refining sector post global slowdown had increased risks for RPL as a standalone unit. Therefore, the merger with RIL would expose RPL shareholders towards a relatively stable exploration business, integrated refining and petro-chemical business and emerging retail business. 4. Industry scenario: The demand is expected to rise through the year 2010 at a rate of about 2% per year for oil and 3% per year for gas. Moreover, in the recent past there has been an increase in the demand for oil and gas. It is expected that the demand for oil and gas will continue to increase as they are expected to remain the leading energy sources for some time to come. Also increase in the exploration and production is expected due to advancement in the technology. Therefore, it would be safe to assume that the shareholders of RIL will be able to earn regular a nd consistent dividends along with the capital appreciation in the stock prices.
95
5. Issue price: The RPL issue price was Rs 60 per share and the proposed swap ratio and the current market price is higher than the issue price. In a market which has seen so much of turmoil in the last 12 months, the swap ratio can be considered as appropriate for the shareholders of both the companies.
Table 14: Swap ratio DATE 27-Feb-09 2-Mar-09 3-Mar-09 4-Mar-09 1-Apr-09 2-Apr-09 6-Apr-09 8-Apr-09 CLOSING PRICE OF RIL 1,265.05 1,225.15 1,199.05 1,209.60 1,579.45 1,662.50 1,672.25 1,724.05 CLOSING PRICE OF RPL 76.20 75.15 73.35 74.00 98.30 103.60 104.20 107.80 RATIO 16.60 16.30 16.35 16.35 16.07 16.05 16.05 15.99
From the above table, we can see that the ratio of the market price of RIL and RPL on the date of the announcement of the merger i.e., February 27, 2009 was 16.6. On the date of the announcement of the swap ratio for the merger i.e., March 2, 2009, the ratio of the market prices of the two companies was 16.3. For the next week, the ratio remained approximately equal to 16. On April 1, 2009, the date from which the merger was to be effective, the ratio still remained 16 (approx). Only after April 8, 2009, the ratio fell slightly below 16. Therefore, it can be said that the shareholders who purchased the shares of RPL during this time made a breakeven decision as the ratio of the market price of the shares of the company is almost equal to the swap ratio that has been decided for the merger. Hence, the swap ratio will put the RPL shareholders in the same position after the merger as the shareholders of RIL as they are in now. Thus, the question that arises for the stakeholders is not whether to shift from RPL to RIL but whether to stay as the shareholders of RIL. To understand this, analyzing the volatility of the company is essential which has been done in the later part of the report.
96
STOCK POSITION
The closing stock prices for RIL and RPL for the period o f two months starting from February 27, 2009, the date on which the announcement for the merger of the two companies was made till April 29, 2009, has been recorded to understand the movement of the share prices due to the announcement of the merger. On Friday, February 27, 2009, the announcement of the merger of RPL with RIL was announced. Following the announcement the shares of Reliance Industries dropped to an intraday low of Rs 1,213.20, down four per cent on the following Monday i.e. March 2, 2009, while RPL scrip also showed a fall of over eight per cent to an intra-day low of Rs 70 on the Bombay Stock Exchange. When the market closed on Monday, RIL was down by 3.15% to Rs 1,225.15 and RPL was lower by 1.38% at Rs 75.15. In 2002, when the first merger of RPL with RIL was announced, the share prices of RIL shares had dropped by 2.85% to Rs 312.95.
Close Price for RIL
2000 1800 1600 1400 1200 1000 800 600 400 200 0
Close Price
Figure 10: Stock Position of RIL Moreover, the share prices of RIL continued to fall further for the next few days as it had fallen in the following week of the merger announcement in 2002. Despite the fall in prices of the shares, analysts have termed the merger positive for RIL this time around. Another similarity that can be seen in both the mergers of 2009 and 2002 is that RIL has decided to cancel its
97
shareholding in RPL as a part of the deal, which is resulting in a dilution of the promoter holding in the merged entity.
Close Price for RPL
120 100 80 60 40 20 Close Price
0
Figure 11: Stock Position of RPL The Board of Directors of Reliance Industries and its refinery subsidiary RPL on Monday, March2, 2009 approved the merger of the two firms, thus creating one of the world's largest oil refinery. They offered the shareholders of RPL one RIL share for every 16 shares held by them i.e., the swap ratio was decided to be 1:16. On Saturday, April 4, 2009, the shareholders and the creditors of Reliance Industries Limited approved the Scheme of Amalgamation of Reliance Petroleum Limited with RIL. It was a Court convened meeting of the equity shareholders, secured creditors and unsecured creditors of RIL. 98.86% of the shareholders present in person/proxies, representing 99.9998% of the total value of the equity shares held by them, voted in favor of the Scheme of Amalgamation. 100% of the Secured and Unsecured Creditors present in person/proxies voted in favor of the Scheme of Amalgamation.
98
ANALYSTS TAKE ON THE MERGER OF RIL-RPL
The major points of the analysis given are summarized as under: CNBC TV 18 1 1. The swap ratio for the merger which is 1:16 would mean that there will be a dilution of 4.4% of fully diluted RIL equity. 2. The treasury stock which has been created due to merger will be extinguished which would be a positive step for Reliance. 3. The merger is believed to be an EPS accretive merger because the equity dilution will be only for 4.4% of RIL‟s expanded equity. But the expected contribution from RPL will be more and hence, it would be more accretive to the RI L‟s shareholders. 4. The merger is said to be tax neutral and after the merger both the entities will continue to enjoy the same tax benefits which they are currently enjoying. Therefore, on a consolidated basis, RIL will get the benefits of being an Export Oriented Unit and RPL will enjoy the SEZ benefits. 5. According to RIL, over the next 12-18 months, the demand for fuel would go down and therefore, the cost efficiency attained due to merger of the new refinery with the existing one will help the company to sell the product. 6. One of the main reasons for the merger of RIL-RPL is that since the two refineries are adjacent to each other, it would allow operational synergies for the Company. SANJIV AGRAWAL, E&Y2 1. The merger of RIL-RPL can be considered as a move in the right direction as it would help the merged entity to take advantage of the financial and operational synergies. 2. The company will get benefits of scale because of the integration and also it can prove beneficial in the bargaining of the crude prices given the level of complexity of the new refinery. The merger will give the company the operational flexibility.
1
CNBC TV 18: ia/news/business/ril-rpl-merger-a-co mprehensive-cnbc-tv18analysis/387462 [Accessed on April 19] 2 E&Y, SANJEEV A GA RWA L: m/ 500/news/“the -merger-doubles-rils-cashflow-overnight” sanjiv-agrawal-ey [Accessed on April 21]
99
3. It is expected that the cash flows would increase tremendously given the capacity of the new refinery along with the high Gross Refining Margin (GRM). 4. The shareholders of RIL will benefit in the long run as the new refinery merges with the existing one. DIMENSIONS CONSULTING3 Ajay Srivastava of Dimensions Consulting believes that the merger of RIL and RPL will be positive for RIL and the company would be benefitted from the increased cash flows. Moreover, a non-operating asset i.e., a 70% shareholding equivalent is becoming an operating asset for the company. THE FINANCIAL EXPRESS 4 1. Post merger, RIL‟s standalone balance sheet will show an increase by 13% as the sale of stake by Chevron will affect the balance sheet of the company marginally. 2. As told by ICICI Securities, the key reason for the merger cannot be attributed to the operational cost savings as believed by the company as the management of RIL has control over that of RPL and therefore, RIL managed operations of both the refineries. 3. Tax benefits won‟t increase post merger as both the companies have their own set of tax benefits. However, the huge amount of positive free cash flow from RPL of up to $1.4 billion would be utilized by RIL. BUSINESS STANDARD 1. RIL will benefit from the merger in the form of increased cash flows from RPL.
2.
The swap ratio of 1:16 has been in the favor of the shareholders of RPL. Moreover, the treasury stock created on account of merger will be cancelled which would mean a small dilution (4.4%) in the equity base which makes the merger earnings accretive.
3
DIM ENSIONS CONSULTING: india/news/market-outlook/ mergerrpl-positivefor-ril-d imensions-consulting/387374 [Accessed on April 20] 4 FINA NCIA L EXPRESS: [Accessed on April 16]
100
3. The merger will be tax neutral for the merged entity which implies that the carry forward of unabsorbed depreciation cannot be set off against RIL‟s profits. 4. Due to the merger of the new refinery with a greater complexity and capacity, processing of cheaper and heavier crude oil is made easier which would lead to higher GRMs. DNA INDIA5 1. Analysts believe that the cash flows generated by RPL will help the capital expenditure plans of RIL as the former is better structured in terms of cash flows. 2. The merger would also increase RIL's operational synergies. Its cost efficiencies would help the company to optimize fiscal incentives, enhance financial strength and flexibility. It would also eliminate transfer pricing issues. 3. The deal would impart the company with much needed liquidity in the short term. 4. Since the revenues of RIL from refining are about two-third, the merger would double the capacity of the refinery making the revenues from the other business of RIL negligible. 5. Though the merger is unlikely to have any impact on the tax benefits available to RPL, RIL will be able to use the depreciation from the plant of RPL to lower the profits of the merged entity to save on tax. ANGEL BROKING6 1. Angel Broking believes that the RIL-RPL merger is likely to be Earnings accretive for RIL shareholders. FY2010 EPS is likely to be higher by 1.66% due to the merger. 2. It is expected by RIL that the merger will provide synergies in the procurement of crude and product placement. But the researchers believe that synergies might be low as the two companies share the facilities.
3.
As per the analysis, there will be a dilution of RIL‟s equity due to the RIL-RPL merger by 4.4% on account of issuance of 6.92 crore shares to RPL shareholders. This is so because RIL has decided to extinguish the treasury share that it had created on the account of the merger. The merger will dilute RIL promoters' effective stake by 2.3%.
5 6
DNA INDIA: m/report.asp?newsid=1235366 [Accessed on April 15] ANGEL BROKINGS: [Accessed on April 14]
101
4. As per the analysis, the deal is a case of win-win situation for both RIL and RPL. This is because post merger, RIL will have improved Cash Flows and Balance Sheet along with a lower cost of capital. 5. For RPL, the merger is expected to reduce the volatility in Earnings which would allow the shareholders to participate in RIL‟s full energy value chain. BRICKS SECURITIES7 1. As per the firm‟s analysis, RIL-RPL merger ratio of 16:1 is in favor of RPL shareholders. BRICS intrinsic value estimates gave the swap ratio to be 21:1 based on the calculations done by their researchers for the valuations of the two companies. 2. Due to the merger, the company will be issuing 70 million shares to the shareholders of RPL which would increase the share capital of RIL to Rs. 16423 million. 212 million shares of RIL representing the treasury stock in lieu of its 75.4% stake (post Chevron stake purchase) will be extinguished. KHANDWALA SECURITIES8 1. The swap ratio recommended by the boards of RIL and RPL is 1:16. For this, RIL will be issuing 6.92 crore new shares which would in turn increase the equity share capital to Rs. 1643 crore. 2. Analysts believe that the merger is a strategic move made by the promoters of RIL mainly to enjoy economies of scale, capitalizing the cash flows of RPL and minimizing the cost of capital. 3. The downturn of the refining sector due to the global slowdown had increased the risks for RPL as a standalone unit. Therefore, due to the merger, the risk of the minority shareholders will be diversified.
7 8
BRICKS: march_2.pdf [ Accessed on April 13] KHANDAWALA SECURITIES: merger_03Mar09.asp?ArtCd=142668&Cat=E&Id= [Accessed on April 22]
102
4. The cash generated by RPL, after the merger, can be deployed in exploration and retail business of the merged entity. Similarly, the short term requirements of funds for RPL can be fulfilled by the surplus cash in balance sheet of RIL. Hence, the cost of capital for the combined entity can be reduced. 5. The merged entity would not have any extra tax benefits and the benefits available to the two entities separately will hold good. 6. Merger can help RIL to improve the product slate and facilitate refining of various types of crude oil which would help the company to reduce cost and improve the refining margins. 7. The merger of RIL-RPL will help in unlocking the operational and financial synergies that exist between the two companies.
103
VALUATION OF RIL AND RPL
VALUATION OF RIL Reliance Industries Limited, India‟s largest private sector conglomerate, reported an annual turnover of Rs 1, 33,443 Crores and net profit of 19,458.29 Crores in year 2007-2008. With 1453648601 outstanding shares, RIL has an EPS Rs 133.86. Earnings per Share serve as an indicator of a company's profitability. Higher the EPS, higher is the profitability of the company. Reliance Industries have maintained increasing earnings per share from last 5 years. In 20072008 it showed a 63% of increment from last year‟s 82.16. Thus, with the data of EPS, the growth of RIL looks promising. DPS: Dividend per Share is the amount that shareholder will receive for the each share they own. Dividend is distribution of a portion of a company's earnings, decided by the board of directors, to a class of its shareholders. RIL has been continuously paying dividend to its share holders from last five years. The value of DPS has increased from 5.25 in 2003-04 to Rs 13 in 2007-08. As per my observations and understanding, there has been a change in calculating the DPS for the company since 2006-07. Since this year the DPS is calculated on total outstanding shares minus shares held by subsidiary companies on which no voting rights are exercisable and petroleum trust. Hence, total shares for calculating DPS are:
By doing so, the total amount of equity dividend is lowered and hence a lesser amount of tax has to be paid as dividend tax. Thus the total retained earnings of the company are increased at a particular Dividend per Share (DPS). OPERATING PROFIT PER SHARE: It is a measure of company's earning power from ongoing operations. This is equal to earnings before deduction of interest payments and income
104
taxes. Operating profit is also called EBIT (earnings before interest and taxes) or operating income. Reliance Industries Limited has been showing a considerable growth in the operating profits. In 2007-08 it has shown a total growth of 13%, it has increased from 163.90 in 2006-7 to 185.74 in 2007-08.
Figure 12: Operating Profit per Share
Figure 13: Pe rcentage Growth
BOOK VALUE PER SHARE: It is the total value of the company's assets that shareholders would theoretically receive if a company were liquidated. It is also called the net asset value of a company, calculated by total assets minus intangible assets (patents, goodwill) and liabilities. RIL has shown an increasing book value of the company for last 5 years. Thus is reduces the risk factor for the investors. Since the inception of the company, it has been trying to increase the net worth of the investors and in the process it has today become the second largest private sector conglomerate in the world. The Book Value per share has increased from 246.73 in 2003-04 to 649.12 in 2007-08, hence showing a growth of 163%.
105
Figure 14: Book Value per Share PROFITABILITY RATIOS: Coming to the Profitability Ratios of the company, Reliance Industries has an operating margin of 17.47% in 2007-08..
For operating margin to increase, the difference between sales and cost has to increase. So, if a company is able to sell a given quantity at a higher price without a corresponding increase in expenses, margins are likely to expand. In 2007-08, sales of RIL increased by 19% and the expenses also rose by approximately 19%. Hence the operating margin of the company is pretty similar to its previous year performance. Thus the company has successfully maintained its operating margin over the years. Gross Profit Margin: It is a financial metric used to assess a firm's financial health by revealing the proportion of money left over from revenues after accounting for the cost of goods sold. Gross profit margin serves as the source for paying additional expenses and future savings. Firms that have a high gross profit margin are more liquid and thus have more cash flow to spend on research & development expenses, marketing or investing. Generally, investors
106
avoid investing in firms that have a declining Gross Profit Margin over a time period, example over 5 years.
Looking into the Gross Profit Margin of RIL, we notice that the company has been able to increase its profit margin by a nominal value. Hence, the company seems to be a nice investment. The gross profit margin of RIL has improved from 12.74% to 13.83% in last five years.
Figure 15: Gross Profit Margin Net Profit Margin: Net Profit Margin tells exactly how the operations of a business are performing. Net Profit Margin compares the net income of a firm with total sales achieved. The formula for Net Profit Margin is:
If Gross Profit Margin of a company is very high when compared to Net Profit Margin of the company, this means that a huge amount of earnings is being allotted to marketing or administrative expenses. When we see the Net Profit Margin and Gross Profit Margin of the company, we realize that only a considerable amount of revenue is being used for administrative and marketing expenses. In 2006-07 the Gross Profit Margin and Net Profit Margin of the company were 10.69% and 13.64% respectively.
107
Figure 16: Gross and Net Profit Margin
Return on Net Worth: It is also known as return on equity (ROE). It is an indicator of profitability and investors use ROE as a measure of how company is using its money.
RIL has shown a considerable stability in the RONW of the company. Thus, the company has been using the shareholders money efficiently. In 2007-08 the RONW of the company grew to 23.89% from previous years 18.67%.
Figure 17: Return on Net Worth
108
LEVERAGE RATIOS Total debt to equity Ratio: The debt to equity ratio is used for measuring solvency of a company. It indicates how much the company is leveraged, in other words it measures the company‟s ability to borrow and repay the money. Debt to equity ratio is closely watched by creditors and investors because it reveals the extent to which company management is willing to fund its operations with debt. Lenders are particularly sensitive to this ratio as an excessive high ratio value will put their loans at risk of not being repaid. By using debt in the company finances, it is able to have interest and depreciation tax shield due to the tax paid on the debt interest. However no such tax benefit is achieved when fund raising is done through equity.
Ratios D/E Ratio
2008 0.45
2007 0.44
2006 0.44
2005 0.46
2004 0.61
Thus from the above table it is clear that RIL has maintained its debt to equity ratio around 0.45 over last 5 years. This means that neither the ratio is too high (risky for the lenders) nor it is too low (entitling the business to leverage and earn higher returns on equity).
Fixed Asset Turnove r Ratio:.
Ratios Fixed assets turnover ratio
2008 1.28
2007 1.12
2006 0.96
2005 1.20
2004 0.97
109
RIL has a very high fixed assets turnover ratio, thus the company has been efficiently using the fixed assets to generate its revenues.
LIQUIDITY RATIOS It is used to determine a company's ability to pay off its short-terms debts obligations. Higher the value of the ratio, larger is the margin of safety that the company possesses to cover shortterm debts. Curre nt.
Ratios Current Ratio
2008 1.78
2007 1.61
2006 1.49
2005 1.66
2004 1.75
Reliance Industries Limited has a current ratio of 1.78, thus it can easily pay off its short term liabilities with its short term assets. This states that the company is in good financial health and it shows no signs of bankruptcy.
Quick Ratio: The quick ratio measures a company's ability to meet its short-term obligations with its most liquid assets. Higher the quick ratio better is the position of the company.
110
Inventory is excluded because some companies have difficulty turning their inventory into cash. In the event that short-term obligations need to be paid off immediately, there are situations in which the current ratio would overestimate a company's short-term financial strength.
Ratios Quick ratio
2008 0.99
2007 0.94
2006 0.86
2005 1.23
2004 1.19
The company has a decent quick ratio; this shows that company can repay almost all its current liabilities from its most liquid assets. This states a very fine financial health of the company. Inventory Turnover Ratio: Inventory turnover ratio states how many times a company‟s inventory is sold and replaced over a period of time. A low turnover implies poor sales and, therefore, excess inventory. A high ratio implies either strong sales or ineffective buying.
High inventory levels are unhealthy because they represent an investment with a rate of return of zero. It also opens the company up to trouble should prices begin to fall.
Ratios Inventory turnover ratio
2008 6.98
2007 8.97
2006 7.85
2005 8.91
2004 7.16
BPCL has the ratio of 11.64 while HPCL has the ratio of 9.47. Similarly, the ratio of MRPL and IOCL are 10.53 and 9.09 respectively. The competitor‟s inventory turnover ratio is more than that of RIL. This indicates that the ratio of RIL in comparison to its competitors is low. Reliance Industries Limited does not show a very high inventory turnover ratio; hence it does not deal with ineffective inventory buying problem. The company has very well managed its inventory level and reduces the risk in case of price fall.
111
PAYOUT RATIOS
Dividend Payout Ratio: It is the fraction of net income a firm pays to its stockholders in dividends:
Reliance Industries Limited has been giving out 10-15% of its net income as dividend to its share holders. The payout ratio provides an idea of how well earnings support the dividend payments. More mature companies tend to have a higher payout ratio. Keeping this is mind we can say that Reliance has a dividend payout ratio on a little lower side. But if the company has plans of expansion or if it wants to invest the retained earning elsewhere, which is beneficial for the company, the low dividend payout ratios are justified.
Earning Retention Ratio: It is the percent of earnings credited to retained earnings. In other words, the proportion of net income that is not paid out as dividends forms the retention ratio.
Ratios Earning retention ratio
2008 91.62%
2007 88.73%
2006 84.63%
2005 86.20%
2004 85.73%
Reliance has been showing such high retention ratio because of its continuous expansion and investment plans.
COVERAGE RATIOS Financial Charges Coverage Ratio: It is a ratio that indicates a firm's ability to satisfy fixed financing expenses, such as interest and leases. Reliance Industries have been showing an ever increasing financial charge coverage ratio over the last five years. It has increased from 7.66 in 2003-04 to 26.86 in 2007-08. Thus the company has been able to very well cover the fixed expenses and hence making it a nice investment option, as these ratios suggest of a good financial health of the company.
112
ESTIMATIONS FOR RIL The future financials for Reliance Industries are calculated by using percentage of sales method. The Percentage of Sales Method is a Financial Forecasting approach which is based on the premise that most Balance Sheet and Income Statement Accounts vary with sales. Therefore, the key driver of this method is the Sales Forecast and based upon this, Pro-Forma Financial Statements (i.e., forecasted) can be constructed and the firms needs for external financing can be identified. Thus the estimations have been made using following assumptions: 1) Year 2009 figures are not public. Hence the latest available figures are for 2007-2008. 2) Company would not be involved in any mergers or amalgamations. Hence the RPL – RIL merger has been overlooked by estimating the financial data for Reliance Industries Limited. 3) Sales in coming years are estimated by using following formula:
4) There has been no changes equity, secured loans, unsecured loans and investments of the company as there is no official declaration about it. 5) There is no occurrence of any unexpected circumstances leading to exceptional profit or loss to the company. 6) New revenues from Oil and Gas business from KGD6 fields are not considered for estimation.
Following figures have been generated by taking above states assumptions and by using percentage of sales method.
113
Profit loss account
(Rs crore) Income: Sales Other Income Expenses: Material consumed Manufacturing expenses Personnel expenses Selling expenses Administrative expenses Depreciation Interest Cost of sales Operating profit Other recurring income Adjusted PBDIT Financial expenses Depreciation Adjusted PBT Tax charges Adjusted PAT Other non cash adjustments Reported net profit Earnings For appropriation( Pys) Equity dividend Dividend tax Retained earnings
Average (% of Sales)
1.00
Estimated Figures
Mar'10 Mar'09
214,174.52
169,056.37
70.24% 4.21% 1.50% 3.73% 2.03% 4.81% 1.59%
150,431.11 9,021.81 3,213.83 7,991.83 4,356.09 10,296.07 3,403.19 175,014.67 39,159.85 4,231.73 43,391.58 3,403.19 10,296.07 29,692.32 9,798.47 19,893.86 (19.04) 19,874.81 46,533.02 (3,058.60) (446.31) 50,037.93
118,741.19 7,121.27 2,536.80 6,308.27 3,438.44 8,127.10 2,686.27 138,145.96 30,910.41 3,340.27 34,250.68 2,686.27 8,127.10 23,437.32 3,893.92 19,543.39 (15.03) 19,528.36 23,891.65 (2,414.27) (352.29) 26,658.21
18.28% 1.98% 1.59% 4.81% 16.61% -0.01%
-1.43% 14.59%
114
Balance sheet
(Rs crore) Sources of funds Owner's fund Equity share capital Reserves & surplus Loan funds Secured loans Unsecured loans Deferred Tax Liability Total Uses of funds Fixed assets Gross block Less : revaluation reserve Less : accumulated depreciation Net block Capital work-in-progress Investments Net current assets Current assets, loans & advances Less : current liabilities & provisions Total net current assets Total
Estimated Figures
Mar'10 Mar'09
INR 3,135.79 155,008.95 6,600.17 29,879.51 7,872.54 202,496.96 101,067.90 39,184.27 61,883.63 65,163.73 22,063.60 83,309.57 29,923.57 53,386.00 202,496.96
INR 3,135.79 104,971.02 6,600.17 29,879.51 7,872.54 152,459.03 101,067.90 39,184.27 61,883.63 37,659.17 22,063.60 48,145.94 17,293.31 30,852.63 152,459.03
115
VALUATION OF RPL Reliance Petroleum Limited was set up to harness an emerging value creation opportunity in the global refining sector by Reliance Industries Limited (RIL), one of India's largest private sector company with a significant presence across the entire energy chain and a global leadership across key product segments.
With an annual crude processing capacity of 580,000 barrels per stream day (BPSD), RPL is the sixth largest refinery in the world. It has a complexity of 14.0, using the Nelson Complexity Index, ranking it amongst the highest in the sector. The polypropylene plant has a capacity to produce 0.9 million metric tonnes per annum. The refinery project is being implemented at a capital cost of Rs 27,000 crore being funded through a mix of equity and debt. Reliance Petroleum Limited (RPL) has recently commissioned SEZ refinery at Jamnagar and processed 3.6 million tonnes of crude during the quarter ended 31st March 2009. The unaudited financial results for the quarter / year ended 31st march 2009 revealed the 15 day production results. Based on that data, the valuation of RPL has been done.
Per Share Ratios:
Financial Ratios
Per share ratios EPS (Diluted) (Rs) Adjusted cash EPS (Rs) Dividend per share Operating profit per share (Rs) Book value (incl rev res) per share (Rs.) Net operating income per share (Rs)
Mar ' 09
0.19 0.44 0.50 30.07 8.17
The earnings per share, operating profit per share and net operating income per share are so low for RPL because for year 2008-09 the reported net profit is 84 crores and is based on only 15 days production period. Dividend per Share is NIL for the company as there has no dividend paid till date by RPL.
116
Similarly for profitability ratios, the figures do not show the accurate financial health of the company, because the production has not been carried out for complete year leading to a nominal profit.
Financial Ratios
Profitability ratios Operating margin (%) Gross profit margin (%) Net profit margin (%) Adjusted return on net worth (%) Reported return on net worth (%) Return on long term funds (%)
Mar ' 09
6.17% 3.10% 2.28% 0.62% 0.62% 0.32%
Debt to Equity Ratio: The current debt to equity ratio for Reliance Petroleum Industries is 0.95. This means for every 1 Rupee of equity RPL has taken 0.95 Rupee of debt. Debt to equity ratio for Reliance Industries Limited (RIL) is 0.45. The difference between the D/E Ratio of two companies is because RPL is a new project and hence it requires huge investment for its machine. The debt raised from the market has not been paid off and hence forms a major portion of total funding. However, RIL had 0.45 D/E Ratio as it is an old company and has paid of major portion of its debt over the past years. Fixed Asset Turnove r Ratio: Being a new project, RPL has a very high value of Fixed Asset Turnover Ratio. For 2008-09 its value is 15.54. Curre nt Ratio and Quick Ratio: Current Ratio is a measure of the degree to which current assets cover current liabilities (Current Assets / Current Liabilities). A high ratio indicates a good probability the enterprise can retire current debts. However, the quick ratio measures a company's ability to meet its short-term obligations with its most liquid assets. Higher the quick ratio better is the position of the company.
117
Quick Ratio draws a more realistic picture of a company's ability to repay current obligations than the current ratio as it excludes inventories that may hardly be liquidated at their book value. RPL shows a big difference in the values of current and quick ratios. The value of current ratio comes out to 2.51 whereas quick ratio is only 0.51. This is because of the huge inventory that the company holds. The inventory (Stores, Chemicals and Catalysts) are approximately worth 748 crores.
ESTIMATIONS FOR RPL The future financials for Reliance Petroleum are calculated by using percentage of sales method, the same way it has been done for RIL. The data of 15 days of production, which was published on 23rd April 2009, has been extrapolated to show a year production. The factors which have been taken into consideration are growth rate depending upon the GDP (Gross Domestic Product) growth rate of India and working capacity of the refinery which keeps on increasing per year till it reaches the maximum. The current working capacity of the Reliance Petroleum‟s Jamnagar refinery is 40%. It can be explained as follows: Particulars Current production Actual Capacity Capacity Utilization Per Day Yearly Production Production 240000 580000 0.4137931 87600000 211700000 0.4137931
The current production is 240000 barrels per day but the actual capacity is 580000 barrels per day. By dividing the two figures, we can get the capacity utilization for RPL for 15 days production. The figure is coming out to be approx 40%. This means it was operating of 40% capacity of its installed capacity. We assume that the working capacity will keep on increasing in the coming year and will be 70%, 85% and 100%
118
in 2009-10, 2010-2011 and 2011-12 respectively. The GDP growth has been assumed to be 6%, 8% and 10% for the coming three years.
Assumptions made for making an estimation of the financial figures of RPL: 1) The 15 day production data has been extrapolated to make a year‟s data. 2) The growth of RPL sales depends on the working capacity of the refinery and the GDP of the country. 3) The GDP growth rate is assumed to be 6%, 8% and 10% in the coming years. Similarly the working capacity increases to 70%, 85% and 100%. 4) Percentage of Sales method has been used to estimate the future figures. Thus the percentage of sales of all the expenses and incomes will remain same in the coming years. 5) There is no merger planned for the company. 6) There has been no changes equity, secured loans, unsecured loans and investments of the company as there is no official declaration about it. 7) There is no occurrence of any unexpected circumstances leading to exceptional profit or loss to the company.
Profit loss account
(Rs crore) Growth Rate (Assumption) Installed Capacity Working Capacity Income: Sales Other Income Expenses: Material consumed Personnel expenses Selling expenses
Average Percentage of Sales (%)
Actual Figures
IV Q Mar ' 09
Estimated Figures
Mar ' 10 6% Mar ' 11 8% 100% 85% 218,352.50 217,721.78 630.72 214,110.85 173,502.60 947.13 19,392.51 Mar ' 12 10% 100% 100% 282,451.40 281,757.60 693.79 277,084.63 224,532.77 1,225.70 25,096.19 119
100% 40% 3,702.00 1.00 3,678.00 24.00 3,617.00 79.69% 0.44% 8.91% 2,931.00 16.00 327.60
100% 70% 166,602.79 166,018.79 584.00 163,265.35 132,300.46 722.21 14,787.32
Administrative expenses Depreciation Interest Cost of sales Operating profit Other recurring income Adjusted PBDIT Financial expenses Depreciation Adjusted PBT Provision for Current Tax Adjusted PAT Reported net profit Earnings For appropriation( Pys) Transfer To Reserve Retained earnings
4.80% 3.07% 1.44%
176.40 113.00 53.00 3,451.00
7,962.40 5,100.63 2,392.33 155,772.39 10,246.40 584.00 10,830.40 2,392.33 5,100.63 3,337.44 39.26 3,298.17 3,298.17 84.00 3,298.17 3,382.17
10,442.12 6,689.11 3,137.37 204,284.36 13,437.42 630.72 14,068.14 3,137.37 6,689.11 4,241.66 49.90 4,191.76 4,191.76 3,298.17 4,191.76 7,489.93
13,513.33 8,656.50 4,060.13 264,368.00 17,389.61 693.79 18,083.40 4,060.13 8,656.50 5,366.77 63.14 5,303.63 5,303.63 4,191.76 5,303.63 9,495.39
6.17% 0.39% 1.44% 3.07% 1.18%
227.00 24.00 251.00 53.00 113.00 85.00 1.00 84.00 84.00 84.00 84.00
Balance sheet
(Rs crore) Sources of funds Owner's fund Equity share capital Reserves & surplus Loan funds Secured loans Deferred Tax Liability Total Uses of funds Fixed assets Gross block Less : revaluation reserve Less : accumulated depreciation Net block Capital work-in-progress Investments Net current assets Current assets, loans & advances Mar'10
Estimated Figures
Mar'09
4499.986875 12,415.13 12,827.53 29742.65 212.48 28.51 183.97 26,473.87 2,438.32 1,074.50
4499.986875 9,032.96 12,827.53 26360.4762 212.48 28.51 183.97 23,172.32 2,438.32 940.50
120
Less : current liabilities & provisions Total net current assets Total
428.01 646.49 29742.65
374.63 565.87 26360.4762
STUDY OF STOCK PRICES OF RIL AND RPL
To understand the stability of the stock prices of Reliance Industries Limited, a study of stock prices of the company was done by using last one year‟s data. Percentage standard deviation was calculated by using following formula:
Where, RL = percentage change in share price on daily basis, determining the change in return Rf = Risk-Free Rate of 364 day T-bills as on 13/04/2009 (%) = 4.4
Variance is calculated through:
Where N is the total number of days of which the data has been considered.
Annual Returns of a stock are calculated by using the formula:
121
By using the above formulas, Deviation in the stock prices of Reliance is 3.3523 and the deviation for RPL is 3.0302. This shows that RPL‟s stock show more stability when compared to Relia nce Industries Limited‟s stocks.
Particulars Annual Returns (%) Standard Deviation (%) Variance (%)
RIL -26.21 3.35 11.19
RPL -34.81 3.03 9.15
S&P CNX Nifty Index -28.75 2.62 6.85
Annual Returns percentage is negative because the prices of the companies and the Nifty index were higher on 1 April 2008 when compared to the current indexes. This is because of the current slow down or economic recession as it may be called.
If we observe the Annual Returns for Nifty Index, we see that it has gone down by 28.75 however the RPL stock‟s annual return has gone down up to 34.81%. Thus, RPL has shown more loss when compared to the Nifty Index. RIL stock‟s annual returns have gone down by only 26.21%, this value is better than that of Nifty Index. Hence even after the economic slowdown, RIL has managed to show better results.
However when we talk about the standard deviation and the variance in the prices of the stock market of the two companies and the Nifty Indexes, we can say that standard deviation and variance of RIL is more when compared to RPL and Nifty. Thus RPL proves to be a more stable stock when compared to RIL. Similarly, the variance of RIL is very high as compared to that of Nifty.
Particulars Covariance Correlation Coefficient Volatility (β)
RIL 3.68 0.42 0.54
RPL 3.29 0.41 0.48
Covariance , in probability theory and statistics, is a measure of how much two variables change together. In finance, it is a measure of the degree to which returns on two risky assets
122
move in tandem..
So when we calculate the covariance of RIL with stock market, we see that the value comes out to be positive. Same is the case for RPL. But looking the value of covariance we realize that RIL has been more correlated to stock market when compared to RPL. Hence, the stock prices of RIL very closely follow the stock market indexes. Thus the changes in the RIL stock prices are very much governed by the broader factors like economic condition and political scenarios.
Correlation coefficient is a measure that determines the degree to which two variable's movements are associated. The correlation coefficient will vary from -1 to +1. A -1 indicates perfect negative correlation, and +1 indicates perfect positive correlation. Following formula is used for calculating the correlation coefficient:
If we see the values of correlation coefficient we can say that the values for RIL and RPL are approximately the same. This means, both the stock prices vary in association with the variance of Nifty Index. But strictly speaking, the association of RIL stock prices is higher with Nifty Index when compared to RPL‟s.
Volatility is a statistical measure of the dispersion of returns for a given security or market index. It is normally denoted by β (beta).
123
volatility means that a security's value does not fluctuate dramatically, but changes in value at a steady pace over a period of time. When we compare RIL stock‟s volatility with RPL stock‟s volatility, we understand that Reliance Industries Limited has higher volatility. This means RIL stocks have higher risk of changing value over a short period of time. Thus, RIL stocks are more volatile then RPL share prices.
Observation: Higher returns are accompanied by high risks. RPL and RIL stocks very well justify the statement. RIL has less stable stock as compared to RPL and hence are riskier. But at the same time the returns of RIL stocks are better when compared to RPL stocks.
WEIGHTED AVERAGE COST OF CAPITAL (WACC)
It is the rate that a company is expected to pay to finance its assets. It is a calculation of a firm's cost of capital in which each category of capital is proportionately weighted. All capital sources - common stock, preferred stock, bonds and any other long-term debt - are included in a WACC calculation. The WACC of a firm increases as the beta and rate of return on equity increases. With an increase in WACC decrease in valuation takes place and a higher risk is noticed.
Now, for the valuation of the company let us find out the WACC for RPL and RIL and then compare the results.
The cost of capital can be calculated by using the following formula:
Where, Kc = Cost of capital We = Weight of equity
124
Ke = Cost of equity Wd = weight of debt Kd = Cost of debt Now, further calculations are done using below stated formulas:
The cost of debt comes out to be 4.55 for RIL and 5.96 for RPL. The financial charges mentioned in the quarterly report of RPL are only 53 crore rupees. But the one mentioned in the RPL‟s balance sheet is 774.213252 crores, so the calculations are made on that statistics. Thus we can see that cost of debt for RIL is lesser when compared to RPL.
The cost of equity is 17.48 for RIL and 16.10 for RPL. Here Rf is the risk free rate, which is rate of 364 day T-bills as on 13/04/2009, i.e. 4.4%. Rm is market return, which comes out to be 28.75 for the considered period. The value of β for RPL is 0.48 and for RIL it is 0.54.
125
WACC For RIL
Beta(β) Risk free rate (Rf) Market return (Km) Risk Premium Required Rate of Return (Cost of Equity) Cost of Debt Tax Rate After Tax Cost of Debt Total Debt Shareholders' Equity Total capital WACC 0.54 4.40 28.75 24.35 17.48
(in percent) (in percent) (in percent) (in percent)
WACC For RPL
Beta(β) Risk free rate (Rf) Market return (Km) Risk Premium Required Rate of Return (Cost of Equity) Cost of Debt Tax Rate After Tax Cost of Debt Total Debt Shareholders' Equity Total capital WACC 0.48 4.40 (in percent) 28.75 (in percent) 24.35 (in percent) 16.10 (in percent) 6.04 (in percent) 1.18 (in percent) 5.96 (in percent) 49% (in percent) 51% (in percent) 100% (in percent) 11.15 (in percent)
5.38 (in percent) 15.44 (in percent) 4.55 (in percent) 29% (in percent) 65% (in percent) 94% (in percent) 13.48 (in percent)
From this we can say, that WACC for RIL is higher than that of RPL. This shows that RIL is a riskier firm to invest when compared to RPL. This is majorly due to the equity percentage in RIL. Reliance Industries uses almost 65% of equity, as equity is costlier than debt, WACC for RIL is higher for the company.
As the WACC for RIL is 13.48%, this means that only those investments should be made that give a return higher than the WACC of 13.48%. Similar ways RPL should invest only in investments which pays higher returns than 11.15%. Thus, because of the higher portion of equity in capital structure, WACC of RIL is higher than RPL‟s WACC.
After the calculation of WACC, calculations for FCFE and FCFF were done to analyze the financial position of the two companies. These are discussed in detail as under:
126
FREE CASH FLOW TO EQUITY
Free Cash Flow to Equity (FCFE) means a measure of how much amount of cash can be paid to the equity shareholders of the company after the payment of all the expenses, reinvestment and debt repayment.
It is calculated as:
FCFE is many a times used by analysts to determine the value of a company. In the valuation of RIL, the two stage FCFE model has been used. According to this model, the value of a stock is the present value of the FCFE per year for the extraordinary growth period plus the present value of the terminal price at the end of the period. It can be written as:
Where,
FCFE (t) = Free Cash flow to Equity in year t Pn = Price at the end of the extraordinary growth period r = required rate of return to equity investors in the firm The terminal price is generally calculated using the infinite growth rate model,
Where, gn = Growth rate after the terminal year forever. In the case of RIL for the current period:
Current Earnings per share= (Capital Spending - Depreciation)*(1-DR) Change in Working Capital * (1-DR) Current FCFE 133.86 1.2 59.87 75.19
127
And in the case of RPL,
Current Earnings per share= (Capital Spending - Depreciation)*(1-DR) Change in Working Capital * (1-DR) Current FCFE 0.19 -0.14 1.4 -1.06 depreciation/amortization is added back to the cash flows because free cash flow is meant to measure money being spent right now and the not transactions that happened in the past. This makes FCFE a useful instrument for identifying growing companies with high up-front costs, which may eat into earnings now but have the potential to pay off later. Through FCFE we can find out the total amount that the co mpany could have paid in the previous years. The Free Cash flow to Equity (FCFE) is a measure of how much cash is left in the business after non-equity holders (debt and preference shareholder) have been paid, and after any reinvestment needed to sustain t he firm‟s assets and future growth. If FCFE>Dividends, this means that company is paying too less to its shareholders. But when FCFE<Dividends then company is paying too much to its shareholders. In case of RIL, the dividend per share for the current period is 13. From the above given figures, it can be seen that the dividend paid by the company is less than FCFE. Hence management is in pressure to pay more to its shareholders. But it is also because the company has invested the earnings for the expansion and further investments which help the company to increase the shareholders wealth in the long run. The future plans of action for the company are: 1. Basic studies of new material development in Hazira. 2. Treatment of plant waste water streams for re-use.
128
3. Development of nano metal/metal oxides composites of polyolefin. 4. High shrinkage fiber development in Dhenkanal. 5. Nano structured catalysts for hydrogenation and dehydrogenation processes in Vadodara. 6. Nano structured adsorbents for purification and recovery of monometers. 7. Installation of SSM air texturing pilot machine in Silvassa. 8. PFF silicon finish oil consumption to be reduced by 0.5 Kg/MT in Hoshiarpur. In the case of RPL, the company has not been paying any dividends because RPL is yet to fully commission its refinery and generate the revenues which can be distributed to the shareholders. Though the figure for the FCFE is negative, it does not mean that financially the company is not sound. It is negative because the company has only recently started production (only 15 days of production till March 2008). The estimations for both the companies can give a better view about the future of the company. The estimations for the next four years for RIL are:
Particulars Earnings - (Cap Ex-Depreciation)*(1-DR) -Chg. Working Capital*(1-DR) Free Cash flow to Equity Present Value 1 169.97 1.2 0 171.17 145.71 2 215.83 1.2 0 217.03 157.26 3 274.06 1.2 0 275.26 169.77 4 348 1.2 0 349.2 183.33 Terminal Year 481.66 2.48 0 479.18
The estimations for the next 3 years for RPL are as follows (these figures are only 15 days figures):
Particulars Earnings - (Cap Ex-Depreciation)*(1-DR) -Chg. Working Capital*(1-DR) Free Cash flow to Equity Present Value 1 0.19 -0.14 0 0.34 0.29 2 0.2 -0.14 0 0.35 0.26 3 0.21 -0.14 0 0.36 0.23 Terminal Year 0.23 0 0 0.23
129
FREE CASH FLOW TO THE FIRM Free cash flow for the firm (FCFF) is a measure of a company's profits after it has laid out money for all expenses and reinvestments. FCFF means a measure of financial performance that expresses the net amount of cash that is generated for the firm, consisting of expenses, taxes and changes in net working capital and investments. It is calculated as:
This is a measurement of a company's profitability after all expenses and reinvestments. It's one of the many benchmarks used to compare and analyze financial health. It is one of the most important criteria to check the financial condition of a company. A positive value would indicate that the firm has cash left after expenses. A negative value, on the other hand, would indicate that the firm has not generated enough revenue to cover its costs and investment activities. It indicates bad financial health of the company. In the valuation of RIL, the two stage FCFE model has been used and t his model is designed to value the equity in a firm, with two stages of growth, an initial period of higher growth and a subsequent period of stable growth. In case of RIL,
Current EBIT * (1 - tax rate) = - (Capital Spending - Depreciation) - Change in Working Capital Current FCFF 24468.28 150.81 7512.8 17106.29
And in case of RPL:
Current EBIT * (1 - tax rate) = - (Capital Spending - Depreciation) - Change in Working Capital Current FCFF 248.05 -64.77 627.82 -315.01
130 shareholders generally don‟t prefer negative cash flows for the company in any year until they get good returns. Negative cash flow does not necessarily mean loss, and it may be due only to a mismatch of expenditure and income. Having positive FCFF implies that the company has free cash flows, which is good news for the investors. Therefore, before investing in any company, the investors should also have a look at the FCFF of the company. The value of FCFF for RIL is positive, which implies that company has a sound financial health. In case of RP L, the value is negative but it does not imply that the company‟s financial health is low. It is negative because the company has only recently started production (only 15 days of production till March 2008). The estimations for the future can give a better picture of the company‟s financial position. The estimations for the next 4 years are given as under:
Particulars EBIT * (1 - tax rate) - (Cap Ex-Depreciation) -Chg. Working Capital Free Cash flow to Firm Present Value 1 28,391.43 -174.99 3121.97 25544.45 2250.32 2 32943.59 -203.05 3506.50 29640.14 23017.09 3 38225.62 -235.60 4068.72 34392.51 23535.28 4 44354.56 -273.38 4721.08 39906.86 24065.12 Terminal Year 56098.14 -11112.98 0 67211.12
Similarly, the estimations for RPL for the 4 years are as follows (these figures are only 15 days figures):
Particulars EBIT * (1 - tax rate) - (Cap Ex-Depreciation) -Chg. Working Capital Free Cash flow to Firm Present Value (Rs. In Crores) 1 258.47 -67.49 23.68 302.27 272.24 2 269.32 -70.32 24.68 314.96 255.49 3 280.63 -73.27 25.71 328.19 239.77 4 292.42 -76.35 26.79 341.98 225.02 Terminal Year 332.12 -151.3 0 483.42
131
FINDINGS
1. The financial ratios of RIL reveal a good financial health of the company, as the company has an increasing book value in last five years. The company has strong payout and liquidity ratios. 2. The stocks of RPL are more stable when compared to RIL. Thus, the volatility of RPL stock (0.48) is lesser than RIL‟s stock (0.54) 3. Weighted average cost of capital (WACC) of Reliance Industries Limited (13.48%) is more than Reliance Petroleum Limited (11.15%). This is because RPL is better leveraged than RIL. 4. In case of RIL, the dividend paid by the company is less than FCFE but it is so because the company has invested the earnings for the expansion and further investments which help the company to increase the shareholders wealth in the long run. While in the case of RPL, the company has not been paying any dividends because it has not yet become due to the investors. 5. The positive value of FCFF for RIL implies that company has a sound financial health. In case of RPL, the value is negative because the company has only recently started production (only 15 days of production till March 2009).
132
CONCLUSION
The capital cost of RPL‟s project was estimated at Rs. 270 billion. The project was funded through debt (Rs. 157.5 billion) and equity (Rs 112.5 billion). RPL went for an Initial Public Offer (IPO) through the book building process and it fulfilled certain guidelines issued by SEBI called Disclosure and Investor Protection Guidelines (DIP) and to raise funds through debt, guidelines of the Reserve Bank of India for External Commercial Borrowings were complied with. Also software using the functions of Microsoft Excel 2007 called the Portfolio Tracker was created which helps in calculating the gain or loss on the stocks of a portfolio. By analyzing the merger of RPL with RIL, it can be concluded that the swap ratio for the merger which is 1:16 would mean that there will be a dilution of 4.4% of fully diluted RIL equity is in favor of the shareholders of RPL.
The deal is believed to be a win-win situation for the shareholders of both the companies. This can be attributed to the fact that post merger RIL will have improved Cash Flows and Balance Sheet along with a lower cost of capital. Moreover, the merger will help in unlocking the operational and financial synergies that exist between the two companies.
The financial ratios and free cash flows of RIL state that the company is in good financial health. RPL has started its operation on 15th March, 2009. Hence only 15 days operational data has been made public. Currently the company has been highly leveraged as D/E Ratio for the company is 0.95:1 because RPL is a new project and requires heavy machinery.
The weighted average cost of capital (WACC) for RIL is higher than RPL and hence the valuation of RPL is better when compared to RIL (in respect of WACC).
But considering the fact that RPL is a new project and the estimations of the company statistics may show high variations from the actual results, RIL seems to be a good investment option. This is because the Reliance Industries Limited has managed to be a profit generating company
133
since last 30 years. Thus, past data suggests that though RPL is a relatively stable stock option for investme nt, RIL‟s past data promises to increase the shareholders wealth.
We can also conclude that buying the shares of RIL will yield better results for the shareholders even though RIL stocks are more volatile as compared to the shares of RPL. This is because the former has a record of sustained earnings since its inception.
134
RECOMENDATIOS
Based on the study, it can be said that the RPL shareholders would benefit in the long term from the merger with RIL and the stakeholders should not exit the market by selling their shares of RPL. This is because after the merger, RPL shareholders will be gaining from the upsides from RIL‟s petroleum business as the company will be gaining from the strong value accretion due to exploration of the large unexplored acreage in the highly prospective areas. Also, a steady rise in RIL‟s earnings is more likely than in RPL‟s earnings. There has been a secular rise in RIL‟s earnings since its inception. Because there are a number of subsidiary companies under the group, any decline in earnings of one subsidiary is set off against the increase in the earnings of the other subsidiaries. Moreover the downturn of the refining sector post global slowdown has increased risks for RPL as a standalone unit. Therefore, the merger with RIL would expose RPL shareholders towards a relatively stable exploration business, integrated refining and petro-chemical business and emerging retail business. The swap ratio of 1:16 will put the RPL shareholders in the same position after the merger as the shareholders of RIL as they are in now. Thus, we can say that the merger will benefit t he shareholders in the both the long and the short run.
135
DECLARATION
The reports and notes on merger and valuation of the company is completely my work and it is as per my understanding. Same are not vetted or authorized by the Company.
136
REFERENCES
FROM PRINT MATERIAL 1. IM Pandey, 2008, Financial Management, Vikas Publishing House, 9th Edition, pp.432434 2. IM Pandey, 2008, Financial Management, Vikas Publishing House, 9th Edition, pp.438440 3. Taxman‟s SEBI Manual, Volume I, 12th Edition, June 2008
FROM WEB PAGES AND ONLINE BOOKS 1. 2. 3. 4. 5. 6. 7. 8. [Accessed on 8 March 2009] 9. [Accessed on 8 March 2009] 10. Morgan Stanley, “Strategy Chart book January 5, 2007” [online] retrieved from : [Accessed on 15 March 2008] 11. [Accessed on 18 March 2009] 12. [Accessed on 25 March 2009] 13. [Accessed on 27 March 2009] 14. [Accessed on 27 March 2009] 15. [Accessed on 27 March 2009] 16.- loans/complete- guide-to-debt- financing.htm [Accessed on 1 April 2009] 17. [Accessed on 1 April 2009]
137
18.- financing/ [Accessed on 2 April 2009] 19. [Accessed on 2 April 2009] 20. [Accessed on 2 April 2009] 21. 9573 [Accessed on 3 April 2009] 22. [Accessed on 3 April 2009] 23. [Accessed on 4 April 2009] 24.- market.html [Accessed on 4 April 2009] 25. [Accessed on 5 April 2008] 26. [Accessed on 5 April 2008] 27. [Accessed on 5 April 2008] 28. [Accessed on 5 April 2008] 29. [Accessed on 10 April 2009] 30. [Accessed on April 23 till April 29] 31. 5 [Accessed on May 4] 32.- Lynch [Accessed on May 4] 33. [Accessed on May 5] 34. [Accessed on May 5] 35.- merger-to-unlock-synergiescrudesourcing-ril/387348 [Accessed on May 6] 36. [Accessed on May 6] 37. [Accessed on May 7]
138
38.- rpl- merger-what-should- investors-do.html [Accessed on May 7] 39. [Accessed on May 8] 40.- merger-gets-shareholders- nod/444294/ [Accessed on May 8] 41. [Accessed on May 8] 42. [Accessed on May 11] 43. [Accessed on May 11] 44. [Accessed on May 11]
FROM LEGAL MATERIALS 1. Reliance Petroleum Limited Prospectus April 28, 2006. S.I. 2006 2. GOVERNMENT OF INDIA. August 2005. Guidelines on External Commercial Borrowings Policies & Procedure. GOI Ministry of Finance. S.I. 2005 3. RELIANCE PETROLEUM LIMITED. 2007 -08. Annual Report. S.I. 2008 4. RELIANCE INDUSTRIES LIMITED. 2007-08. Annual Report. S.I. 2008 5. Other Internal Documents of the Company referred.
139 | https://www.scribd.com/doc/161597253/Share-Capital-Rpl | CC-MAIN-2018-17 | refinedweb | 32,739 | 51.38 |
, and some other Super Simple Python projects we’ve done, plotting a random dataset will make use of the
random library. We’ll also introduce a new library,
matplotlib.
matplotlib is a critical library for data scientists, and the default plotting library for Python.
Before we start with the program, we’ll need to use
pip to install
matplotlib in the terminal. We can do that with the following command:
pip install matplotlib
Generating a Random Dataset
As always, we’ll begin our program with our imports. We’ll import the
random library to generate our random dataset and
matplotlib.pyplot to plot it. That’s all for imports. To plot any two-dimensional dataset, we’ll need a list of
x and
y values. In this example, we’ll generate 100 random integers between 0 and 10 for each axis. We’ll save our
x values in a list called
xs and our
y values in a list called
ys.
import random import matplotlib.pyplot as plt xs = [random.randint(0, 10) for _ in range(100)] ys = [random.randint(0, 10) for _ in range(100)]
Plotting a Random Dataset using MatPlotLib
Once we’ve generated our
xs and
ys all we need to do is use
matplotlib to plot them. Earlier we imported
matplotlib.pyplot as
plt by convention. This allows us to call the module by using the name
plt instead of its full name. We’ll call the
scatter function to plot the
xs and
ys. It’s not strictly necessary to put in the
xlabel,
ylabel, and
title, but I did because it makes the graph look nicer. Once we plot the dataset, we just have to call the
show function to see it.
plt.scatter(xs, ys) plt.xlabel("X") plt.ylabel("Y") plt.title("Random plot") plt.show()
When we run our program we should see something like “Super Simple Python: Plot a Random Dataset” | https://pythonalgos.com/super-simple-python-plot-a-random-dataset/ | CC-MAIN-2022-27 | refinedweb | 323 | 76.01 |
The nosey programmer’s guide to Kotlin and Dart
Kotlin and Dart are the new kids of the block that have actually been around for years. Recently they have come into the mainstream as the languages of choice for Android and Flutter development respectively.
This article is for the ‘nosey programmer’ — those of us who are curious enough to want to know the basics, but not ready for a deep dive yet.
Specifically, this article is for Kotlin folk who are curious about Dart, and Dart developers who are wondering why Kotlin is causing a buzz.
If you are feeling nosey, then you can dip into the following Kotlin/Dart topics below.
- Variables
- Typing (Static vs Strong)
- Collections
- Classes & OOP
- Miscellaneous Features
Side Note — embedding Gists into Medium articles was severely broken when I wrote this. If this gets resolved I might update this article to use Gists for code examples.
Quick History
Kotlin was unveiled by JetBrains in 2011. It was started out of a frustration with existing languages targeting the JVM and the sense that there could be something better. Design of the Kotlin language was lead by Andrey Breslav and the language became open sourced in 2012.
Version 1.0 was release in 2016. It is the language used to write JetBrain’s IDEs and since 2017 it has been embraced language of choice by Google for the Android platform. The language is now steered by the Kotlin Foundation which is largely a joint initiative by JetBrains and Google.
Dart was created by Lars Bak and Kasper Lund. First publicly announced in 2011 it reached 1.0 in 2013. The original goal was to create a more structured language for writing web applications. From it’s early days Dart could run in a VM or be compiled to JavaScript. Unfortunately the Dart VM never made its way into the Chrome browser as had been hoped.
Inside Google Dart has been used in large-scale production apps for several years already via Angular Dart, including the web UI for Ad-Words, Google’s main money-making product. Recently Flutter has brought Dart more into the limelight.
Variables
Kotlin and Dart both support variables with inferred and explicit types, final and (compile time) constants.
// KOTLIN
var stringOne = "String One"
var stringTwo : String = "String Two"
val stringThree = "String Three"
const val stringFour = "String Four"//DART
var stringOne = "String One";
String stringTwo = "String Two";
final stringThree = "String Three";
const stringFour = "String Four";
Kotlin and Dart are pretty similar when declaring variables.
var and
val in Kotlin is a nice way of doing the more ‘traditional’
var and
final in Dart.
const val is a bit more verbose in Kotlin compared to just
const in Dart.
The semicolon!
If you are going to write Dart code you just have to accept that you are going to be hitting the semicolon key a lot. In Dart you need to add a semicolon at the end of each expression.
Kotlin by convention is written without a semicolon. Yay! If you feel compelled, you can use it too.
Statically vs Strongly Typed
Kotlin is statically typed. Dart is strongly typed. What does this mean in principal and in practice?
Statically typed languages are constrained to only support a known set of types. You cannot fake, hide, ignore or disable checking of these types.
Dart is described as being strongly typed. There is less consensus about what a strongly typed language really means (). Let’s not get side-tracked on this debate . The key point here is that with a strongly typed language you can circumvent the type system if you want. (In Dart this is via the
dynamic type.)
What does this mean in practice. Well, probably not too much in that both languages will help avoid type errors at compile and run type.
One quick example can illustrate that in both Kotlin and Dart you can’t coerce a variable to accept different types.
//KOTLINvar aStringVar = "A String"
aStringVar = 3 //compile errorvar anAnyVar : Any = "An Any"
anAnyVar = 3 //no error - Any is a super-type//DARTvar aStringVar = "A String";
aStringVar = 3; //compile errorObject anObjVar = "An Object";
anObjVar = 3; //no error - Object is a super-type
We only touched on this topic and there is probably enough for an entire article. Let’s move on.
Collections
With collections we start to see how Kotlin is a more feature-rich language. But let’s not feel sorry for Dart — it has very good collection classes and capabilities.
Basics
Both Kotlin and Dart have Lists, Sets and Maps.
Kotlin however goes further by offering read-only and mutable variations of these.
Listand
MutableList,
Setand
MutableSetx,
Map and
MutableMap.
Dart collections are mutable. (Collections declared with
const are immutable, but these are compile time constants so not equivalent to Kotlin’s immutable types.)
Let’s look at some code to see some of basic collection declarations.
//KOTLIN
val fixedNumbersList : List<Int> = listOf(1, 2, 3)
val numbersList : MutableList<Int> = mutableListOf(1, 2, 3)
numbersList.add(4)val fixedNumbersSet : Set<Int> = setOf(1, 2, 3)
val numbersSet : MutableSet<Int> = mutableSetOf(1, 2, 3)
numbersSet.add(4)
val fixedNumbersMap : Map<Int, String> = mapOf(1 to "one", 2 to "two")
val numbersMap : MutableMap<Int, String> = mutableMapOf(1 to "one", 2 to "two")
numbersMap.put(3, "three")//DARTfinal List<int> numbersList = [1, 2, 3];
numbersList.add(4);
final Set<int> numbersSet = {1, 2, 3};
numbersSet.add(4);final Map<int, String> numbersMap = {1: "one", 2: "two"};
numbersMap[3] = "three";
Both Kotlin and Dart have typed collections. These can either be explicitly given (as above) or they will be inferred.
Kotlin and Dart also supports specific collection implementations, such as LinkedList or TreeSet.
Whereas the Kotlin syntax is a little more verbose, Dart provides convenient literal syntax for lists, sets and maps.
Collection Methods
Collections in both Kotlin and Dart offer methods for applying functions over collections. Here are some examples to show some of these capabilities in the two languages.
Filtering
//KOTLINval words = "The quick brown fox jumps over the lazy dog".split(" ")// filter, partition, any, all, none
val longWords = words.filter { it.length > 3 }
//[quick, brown, jumps, over, lazy]
val partitionedWords = words.partition { it.length > 3 }
// ([quick, brown, jumps, over, lazy], [The, fox, the, dog])
words.any { it.length > 3 } //true
words.all { it.length > 2 } //true
words.none { it.length == 1 } // true...//DARTfinal words = "The quick brown fox jumps over the lazy dog".split(" ");
// where, retainWhere, firstWhere, lastWherefinal longWords = words.where((word) => word.length > 3);
//(quick, brown, jumps, over, lazy)final longerWords = longWords.toList()..retainWhere((word) => word.length > 4);
//[quick, brown, jumps]final firstLongWord = words.firstWhere((word) => word.length > 3);
//quickfinal lastLongWord = words.lastWhere((word) => word.length > 3);
//lazy
Predicate based filtering is pretty much the same aside the use of
filter vs
where. Kotlin’s
predicateseems more useful that Dart’s
retainWhere and the testing predicates (any, none, all) would also be nice additions to Dart.
Transformations
//KOTLINval words = "The quick brown fox jumps over the lazy dog".split(" ")// map, zip, associate, flattenval wordLengths = words.map { it.length }
//[3, 5, 5, 3, 5, 4, 3, 4, 3]
val germanWords = "Der schnelle braune Fuchs springt über den faulen Hund".split(" ")val matchedWords = words zip germanWords
//[(The, Der), (quick, schnelle), (brown, braune), (fox, Fuchs),
//(jumps, springt), (over, über), (the, den), (lazy, faulen), (dog, Hund)]val wordLengthAssociation = words.associate { it to it.length }
//{The=3, quick=5, brown=5, fox=3, jumps=5, over=4, the=3, lazy=4, dog=3}
val wordsFlattened = listOf(words, germanWords).flatten()
//[The, quick, brown, fox, jumps, over, the, lazy, dog,
//Der, schnelle, braune, Fuchs, springt, über, den, faulen, Hund]// DARTfinal words = "The quick brown fox jumps over the lazy dog".split(" ");// mapfinal wordLengths = words.map((word) => word.length);
//(3, 5, 5, 3, 5, 4, 3, 4, 3)
Compared to Kotlin, Dart’s collection transformations are limited.
zip,
associate and
flatten in Kotlin are useful and would be great to see in Dart.
Aggregating
//KOTLINval wordLengths = words.map { it.length }
//[3, 5, 5, 3, 5, 4, 3, 4, 3]// reduce, fold, sum, max, min, average
val totalLetterCount = wordLengths.reduce { sum, wordLength -> sum + wordLength }
//35
val totalLetterCountFold = words.fold(0) { sum, word -> sum + word.length }
//35
val totalLetterCountSum = wordLengths.sum()
//35
val longestWordLength = wordLengths.max()
//5
val shortestWordLength = wordLengths.min()
//3
val averageWordLength = wordLengths.average()
//3.888888888888889//DARTfinal wordLengths = words.map((word) => word.length);
//(3, 5, 5, 3, 5, 4, 3, 4, 3)// reduce, foldfinal totalLetterCount = wordLengths.reduce((sum, wordLength) => sum + wordLength);
//35final totalLetterCountFold = words.fold(0, (sum, word) => sum + word.length);
//35
Again, Kotlin provides more aggregation collection methods, although notably
reduce and
fold are also found in Dart.
Dart’s Mutating Methods
It’s worth noting that some collection methods in Dart actually change the collection rather than returning a new one. Specifically these are
retainWhere,
sort and
shuffle. This is a shame and not friendly for functional programming. However, as in the example above creating copy of a collection is quite easy and with Dart’s cascade operator (see below) the code can be kept concise.
Language specific collection features
Kotlin has a collection type called a Sequence which can be used for efficiently executing multiple processing steps on larger collections.
Also in Kotlin is the nice ability to use plus and minus operators to add or remove elements from a collection.
Recently added to Dart as of 2.3 is the handy spread operator (…) which makes adding multiple elements to a list a one-liner.
Also worth checking out is this contribution from Pascal Welsch which brings some of Kotlin’s collection features (like immutable collections) to Dart.
Classes and OOP
Classes and object oriented programming is part of both languages. As with collections, Kotlin offers more features and capabilities. Let’s look a some code to get a feel for both languages.
//KOTLINfun main() {
val job:InterruptableJob = InterruptableJob("100")
job.log() //Job(100 :: Status: IDLE)
runJob(job)
job.log() //Job(100 :: Status: RUNNING)
pauseJob(job)
job.log() //Job(100 :: Status: PAUSED)
}fun runJob(job:Job) { job.run() }
fun pauseJob(job:InterruptableJob) { job.pause() }interface Job {
val id:String
fun run()
}enum class JobStatus {
IDLE, RUNNING, PAUSED, CANCELLED, FAILED, COMPLETE
}abstract class BaseJob(override val id: String) : Job {
var jobStatus = JobStatus.IDLE
protected set
fun isIdle(): Boolean = jobStatus == JobStatus.IDLE
fun isRunning(): Boolean = jobStatus == JobStatus.RUNNING
fun isFailed() : Boolean = jobStatus == JobStatus.FAILED
fun isComplete(): Boolean = jobStatus == JobStatus.COMPLETE
override fun run() { jobStatus = JobStatus.RUNNING }
fun log() { println("Job(${id} :: Status: ${jobStatus})") }
}class InterruptableJob(id: String) : BaseJob(id) {
internal fun pause() { jobStatus = JobStatus.PAUSED }
internal fun cancel() { jobStatus = JobStatus.CANCELLED }
override fun run() { super.run() /*do something*/}
}
Above shows a few core OO concepts. Kotlin supports interfaces and abstract classes and inheritance. Classes can have overloaded constructors and can pass values or call methods of a super class.
The code also makes use of some visibility modifiers (e.g. internal, protected) and an enum class.
There are lots of OO features not seen here, some of which we will touch on below.
Let’s look at an implementation in Dart.
//DARTvoid main() {
final job = InterruptableJob("100");
job.log(); //Job(100) :: Status: JobStatus.IDLE
runJob(job);
job.log(); //Job(100) :: Status: JobStatus.RUNNING
pauseJob(job);
job.log(); //Job(100) :: Status: JobStatus.PAUSED
}runJob(Job job) => job.run();
pauseJob(InterruptableJob job) => job.pause();class Job {
String id;
run() {}
}enum JobStatus {
IDLE, RUNNING, PAUSED, CANCELLED, FAILED, COMPLETE
}abstract class BaseJob implements Job {
String id;
JobStatus _jobStatus = JobStatus.IDLE;
BaseJob(this.id);
bool isIdle() => _jobStatus == JobStatus.IDLE;
bool isRunning() => _jobStatus == JobStatus.RUNNING;
bool isComplete() => _jobStatus == JobStatus.COMPLETE;
run() => _jobStatus = JobStatus.RUNNING;
log() => print("Job(${id}) :: Status: ${_jobStatus}");
}class InterruptableJob extends BaseJob {
InterruptableJob(String id) : super(id);
pause() => _jobStatus = JobStatus.PAUSED;
cancel() => _jobStatus = JobStatus.CANCELLED;
run() => super.run();
}
This example illustrates that Dart has some interesting omissions when it comes to OOP.
Dart does not have an interface type. Instead every class implicitly defines an interface. This means you can use the interface of a class (without its implementation) via the implements keyword.
Dart does not support constructor overloading. Instead it is possible to used named constructors. For example we could add to the BaseJob a constructor method like
BaseJob.fromJson().
Similarly, Dart does not support method overloading.
There are no visibility modifiers in Dart. However declaring a variable as method with a ‘_’ prefix makes it private to a class or namespace.
Whilst Dart clearly offers simplified OO capabilities, it does support key constructs needed for object oriented programming, namely polymorphism, inheritance, encapsulation and class introspection with dynamic invocation (dart:mirrors). (Perhaps method overloading is an obvious omission.)
Language specific features
We only scratched the surface of Kotlin’s OO features. Dive deeper and you will find:
- Extensions : add functionality to a class without having to inherit or decorate.
- Data classes: data classes with automatic implementations of equals(), hash() methods and more.
- Sealed classes: prescribe a limited set of types in a class hierarchy.
- Nested classes, Type aliases, Delegation
Dart also provides the facility to use Mixins as an alternative way to share implementations across classes.
Other language features
It’s impossible to fully cover both languages here, however I wanted to mention a few features that will be seen if writing Kotlin or Dart.
Annotations
Both languages provide support for writing annotations to provide additional meta-data for IDEs and code analysis.
Threading & Asynchronicity
Kotlin can use the JVM’s threading mechanisms either via the
thread method, or by implementing the
Runnable interface and calling start on an instance of
Thread, as with Java.
kotlinx.coroutines are lighter-weight mechanism for aynchronous code that avoids the relatively heavy-weight and error-prone JVM threading mechanics.
Dart does not support shared memory threading. Instead Dart has a construct called an Isolate to take advantage of multi-core CPUs. An Isolate has it’s own memory heap — in this way they are isolated from each other. Communication between Isolates is achieved via message passing instead.
TypeAlias and TypeDef
Both Kotlin and Dart support creating named functions so they can be referenced as Function objects.
Kotlin uses typealias to do this, whilst Dart uses typedef.
Dart Cascades
The cascades notation allows chaining method invocation or property access on an object in a very concise syntax.
//DARTfinal planetBuilder = PlanetBuilder();
planetBuilder
..setName("Earth")
..setMass(5.97)
..setSurfaceType(Surface.Terrestrial)
..setGravity(9.8);
final Planet earth = planetBuilder.build();
Reflection / Introspection
Kotlin supports reflection and also enables use of Java’s reflection mechanisms.
Dart supports reflection, however via a separate package which is currently marked as being unstable. In essence, a big problem with reflection in Dart is that Dart is most often run outside it’s VM, i.e. for Flutter or JS.
Code Organisation, Packages & Imports
Kotlin and Dart take different approaches to organising programs. Let’s look at these by starting with two code examples and then highlighting the differences.
Packages
Kotlin and Dart have very different concepts when it comes to a package.
In Kotlin a package is similar to Java. It’s basically a namespace. All Kotlin code must be in a package. If none is specified the package is ‘default’.
However Kotlin is more flexible in that the package structure does not have to be mirrored by the directory structure. In other words you can have all Kotlin files in one directory but still organise the code in multiple different packages.
In Dart the term package is used to describe a library or module. It is a way to distributed and share functionality. There is no package keyword in the language.
The directory structure for a Dart programme is significant.
Imports
In Kotlin you can use code from other namespaces via imports. Code in the same namespace can be referenced without an import. Importing using the package name is the same for code from external libraries as it is for code in the same program. This will be very familiar to Java or .NET developers.
In Dart you can import features from other files, other packages or the core language. This means you will see three variations of import in Dart:
- Relative imports — these import private code in the same package. For example:
import 'src/utils.dart';
- Package imports — these imports are prefixed with “package:” and import features from an external package. For example:
import 'package:http/http.dart';
- Language imports — importing core language features are prefixed with “dart:”. For example:
import 'dart:math';
Notice how relative and package imports reference a
.dart file whereas language imports do not. Namespaces in Dart are based on the a file path.
Read More:
Tooling and Runtimes
Tools
Kotlin has fantastic support inside Intellij-based tools — which is to be expected. The IDE actively suggests improvements and optimisations to the Kotlin code you write and can convert Java to Kotlin.
Maven and Gradle plugins link Kotlin into build tools for JVM-based applications. Command-line compilation is available and support for Eclipse is still on offer.
There are a few Dart tools available. The Dart/Flutter teams provide plugins for IntelliJ and Visual Studio Code. These both provide very good Dart syntax checking and refactoring support. Tight integration with Flutter tools are also baked in.
A bunch of command-line tools are also on offer, including for code formatting, static analysis and documentation generation.
Runtimes
Kotlin primarily targets the JVM and is compiled into JVM bytecode. As such, it can be used in JVM-friendly environments (Android, servers etc).
However Kotlin is breaking free of the JVM with the Kotlin Multiplatform project. The approach is to allow Kotlin to have extensions that allow platform-specific implementations. This is in contrast to code generation.
Dart was originally conceived to run in a VM, and that option still exists today. Indeed you can still write server-side Dart. More typical, however, is for Dart to be compiled into JavaScript or ARM instructions for Flutter.
TL;DR
Kotlin and Dart have risen in popularity over the last couple of years, propelled by their use for mobile app development (Android and Flutter respectively).
Both languages support features you would expect to see today; functional and OO programming constructs, strict/strong typing and type inference, generics, annotations. All this is accompanied by high quality tooling.
This article takes a quick look at both languages. There is a lot that is not covered.
Hopefully this will help Kotlin developers understand that Dart is not just an attempt to fix JavaScript. Likewise Dart developers can get an insight into what the buzz around Kotlin is about. | https://medium.com/snapp-mobile/the-nosey-programmers-guide-to-kotlin-and-dart-cce36d8b35cc?source=---------2------------------ | CC-MAIN-2019-47 | refinedweb | 3,087 | 59.09 |
User:SPIKE/PLS-09
From Uncyclopedia, the content-free encyclopedia
This file concerns my entries in the 8th Poo Lit Surprise competition in 2009.
edit Clip-on tie
Since you're a noob, I'm just dropping by to note that your entry, User:SPIKE/Clip-on tie, as it currently stands, is invalid. It's exactly the same as Clip-on tie, which was created before Poo Lit started. If you're entering that page as a rewrite in Poo Lit, it must be rewritten. Sir Modusoperandi Boinc! 17:36, October 17, 2009 (UTC)
- Thank you for the heads-up. The article in my namespace is an exact copy of Clip-on tie, because the instructions were to move it there for the submission. I did rewrite Clip-on tie as described in the table. If my rewrite of Clip-on tie in the public namespace started before the competition started, that is a valid complaint and I agree the entry is invalid. Spıke ¬ 23:58 17-Oct-09
edit Rules of baseball
Discussion partly copied from another talk page ¬ 03:56 18-Oct-09
- Spike, here's a suggestion: Rewrite Rules of Baseball on your user space and cut out whatever would disqualify you. I don't think anyone's going to think you're violating any ethics by doing whatever version you want on your own user space. If you're concerned about it being changed in mainspace, simply ask that it's not. (And by the way, stop being noble--that's my schtick). WHY???PuppyOnTheRadio 04:05, October 18, 2009 (UTC)
- See? Aren't you glad you followed my suggestion? Congrats on your PLS win! WHY???PuppyOnTheRadio 00:53, October 26, 2009 (UTC)
edit FORTRAN
Copied from User talk:Modusoperandi ) | http://uncyclopedia.wikia.com/wiki/User:SPIKE/PLS-09 | CC-MAIN-2014-52 | refinedweb | 295 | 64.61 |
IEnumerable.Equals seems to call the wrong Equals method
IEnumerable.Equals seems to call the wrong Equals method - You probably want to use SequenceEqual. (new[]{1,2,3}).SequenceEqual(new[]{ 1,2,3}) // True (new[]{1,2,3}).SequenceEqual(new[]{3,2,1})
Enumerable.SequenceEqual Method (System.Linq) - Determines whether two sequences are equal according to an equality comparer . according to the default equality comparer for their type; otherwise, false .
Object.Equals Method (System) - true if the specified object is equal to the current object; otherwise, false . . Equals method calls the GetType method to determine whether the run-time types of .. However, they appear to have ToString, Equals(Object), and GetHashCode
How to Implement Java's equals Method Correctly - It is determined by a class's equals method and there are a couple of Now, some and other point to different instances and are no longer identical, so identical is false. A variable's Identity (also called Reference Equality) is defined by . There seems to be a way out of this: Employee.equals could check
Variants of the equals() method - This equals() method tests for object identity and returns false. dynamically (all methods are virtual in C++ terms) this doesn't seem to make a difference. .. So when you call some method of the collection class that says "is
The 10 Most Common Mistakes in C# Programming - Using the Equals method signature that includes a comparisonType every time you . to see a method called Sum() on the definition of the IEnumerable<T> interface. the class or interface which will then appear to implement this method.
How to Implement Unit Tests for Equals and GetHashCode Methods - Implementing Equals method and its supporting methods, such as Compiler will then be able to link calls to Equals to strongly typed Equality operator returns false if one argument is null and the other one is non-null. private static void ReportVehicles(IEnumerable<Vehicle> vehicles) { var query
Pro LINQ in VB8: Language Integrated Query in VB 2008 - NET Framework that may seem trivial, but can help improve your code by to look at a handy extension method defined in the Enumerable class that let's us The basic definition of two equal sequences is if they have: 6: return false; done this for you in an extension method called SequenceEqual().
c# override equals
C# Basics: Why We Override Equals Method - Object class and, by default, the Equals and == operator perform reference equality. Later in this post, we're going to be overriding Equals and
Overriding Equals in C# (Part 1) - The following example shows a Point class that overrides the Equals method to . If the method used a check of the form obj is Point in C# or TryCast(obj, Point)
Object.Equals Method (System) - This code generation applies to: C#. What: Lets you generate Equals and GetHashCode methods. When: Generate these overrides when you
Generate C# Equals and GetHashCode Method Overrides - To check whether both objects contain the same property value or not. This can be done by overriding the Equals method. We can override the Equals method as listed next: class Product.
Correct way to override Equals() and GetHashCode() - You can override Equals() and GetHashCode() on your class like this: public override bool Equals(object obj) { var item = obj as RecommendationDTO; if (item
The Right Way to do Equality in C# – Aaronontheweb - One of the pitfalls of doing development in C#, Java, C++, or really any predominantly Equals(object o) ; and; [Special cases] Override object.
Part 58 C# Tutorial Why should you override Equals Method - Introduction. It's essential to know how to override the Equals and GetHashCode methods to properly define the equality of types we create.
Overriding Equals and GetHashCode Laconically in C# - In C# objects can be compared with the == operator, with the . name, so let's use it for equality. public override bool Equals(Object obj) { return
.NET == and .Equals() - Equals method is marked as virtual you can override it in any class that derives from Object , which is well,
How to compare two objects (testing for equality) in C# - Tags compare two object values c# compare 2 object c# check if two objects are the same c
c# iequatable
IEquatable - C# Copy. public interface IEquatable<T> .. If you implement IEquatable<T>, you should also override the base class implementations of Equals(Object) and
What's the difference between IEquatable and just overriding - The main reason is performance. When generics were introduced in .NET 2.0 they were able to add a bunch of neat classes such as List<T>
Implementing IEquatable Properly - Explains how to properly implement the IEquatable interface. Overriding Equals and GetHashCode Laconically in C# · PDF File Writer C#
The Right Way to do Equality in C# – Aaronontheweb - Implement IEquatable<T> for your class (where T is the class;); Override object. Equals(object o) ; and; [Special cases] Override object.
Introduction to IEquatable<T> interface in C# - In this article, you will learn about IEquatable interface and Equality in C#.
C# Interfaces - IEquatable - Provides an equality check when there is only one way of comparing the objects (implemented inside the class) IEqualityComparer - Allows you to
Optimizing C# Struct Equality with IEquatable and ValueTuples - C# struct can easily be optimized with awesome C# 7 features with ValueTuple and IEquatable and GetHashCode.
Implementing the IEquatable of T interface for object equality with - Implementing the IEquatable of T interface for object equality with C# .NET. In this short post we'll see a way how to make two custom objects
C#: Classes implementing "IEquatable<T>" should be sealed - C#. 372 rules. All rules .. When a class implements the IEquatable<T> interface, it enters a contract using System; namespace MyLibrary { public sealed class Foo : IEquatable<Foo> { public bool Equals(Foo other) { // Your code here } } }
C# Journey into struct equality comparison, deep dive - C# Journey into struct equality comparison, deep dive no boxing; [Option 3] implement via IEquatable<T> to be fully compliant with some of .
c# equals
.NET == and .Equals() - However, Point3D.Equals calls Point.Equals because Point implements Object. Equals(Object) in a manner that provides value equality. C# Copy. using System
Object.Equals Method (System) - Plain Vanilla Operator == The most common way to compare objects in C# is to use the == operator. For predefined value types, the equality operator (==) returns true if the values of its operands are equal, false otherwise. For the string type, == compares the values of the strings.
Difference Between Equality Operator ( ==) and Equals() Method in C# - Both the == Operator and the Equals() method are used to compare two value type data items or reference type data items. The Equality Operator ( ==) is the comparison operator and the Equals() method compares the contents of a string. The == Operator compares the reference identity
C# difference between == and Equals() - Equals is just a virtual method and behaves as such, so the overridden version will be . If I want to compare references in C#, I use Object.
C# - In C#, Equals(String, String) is a String method. It is used to determine whether two String objects have the same value or not. Basically, it checks for equality.
C# - C# | Uri.Equals(Object) Method. Uri.Equals(Object) Method is used to compare two Uri instances for equality. Syntax: public override bool Equals (object
The Right Way to do Equality in C# – Aaronontheweb - In C#, for instance, you have the following methods that are built into every object : object.Equals; the == operator; ReferenceEquals , for
C# Journey into struct equality comparison, deep dive - Each object has virtual Equals(object obj) method. So, Equals for class types uses referential comparison, but for ValueType uses structural
C# String Equals Examples - This C# example page uses the string.Equals method. Equals is benchmarked against the equality operator.
Overriding Equals in C# (Part 1) - Overriding Equals in C# (Part 1). November Object class and, by default, the Equals and == operator perform reference equality. Later in this
sequenceequal c#
Enumerable.SequenceEqual Method (System.Linq) - C# Copy. public static bool SequenceEqual<TSource> (this System. examples demonstrate how to use SequenceEqual<TSource>(IEnumerable<TSource>,
SequenceEqual - Using C# LINQ - A Practical Overview - If the two sequences contain the same number of elements, and each element in the first sequence is equal to the corresponding element in the second sequence (using the default equality comparer) then SequenceEqual() returns true . Otherwise, false is returned.
C# SequenceEqual Method (If Two Arrays Are Equal) - This C# example program uses the SequenceEqual extension method. It requires System.Linq.
SequenceEqual - Equality Operator - Learn about LINQ SequenceEqual method with try it yourself examples. Linq SequenceEqual Example: SequenceEqual in Method Syntax C#. IList<string>
Why SequenceEqual for List<T> returns false? - Your problem is that one new Sentence { Text = "Hi", Order = 1 } is not equal to another new Sentence { Text = "Hi", Order = 1 } . Although the
SequenceEqual method in C# - SequenceEqual method in C# - The SequenceEqual method is used to test collections for equality Let us set three string arrays string arr1 This
C#/.NET Little Wonders: The SequenceEqual() Method - NET Little Wonders: The SequenceEqual() Method. Once again, in this series of posts I look at the parts of the .NET Framework that may seem
LINQ SequenceEqual Method - Example of LINQ SequenceEqual Method C# Code SequenceEqual(arr2
Determine if two sequences are equal with LINQ C# - If you'd like to check whether the two sequences include the same elements then you can use the SequenceEquals LINQ operator:
sequenceequal - This C# example program uses the SequenceEqual extension method. It requires System.Linq.
c# gethashcode
Object.GetHashCode Method (System) - GetHashCode()); } } } // The example displays output like the following: // n . Since bits are discarded by the left-shift operators in both C# and Visual Basic, this
C# - Object.GetHashCode() Method with Examples. This method is used to return the hash code for this instance. A hash code is a numeric value which is used to insert and identify an object in a hash-based collection. The GetHashCode method provides this hash code for algorithms that need quick checks of object equality.
Why is it important to override GetHashCode when Equals method is - Yes, it is important if your item will be used as a key in a dictionary, or HashSet<T > , etc - since this is used (in the absence of a custom
c# - What is hashCode used for? Is it unique? - The default implementation of the GetHashCode method does not guarantee unique return values for different objects. Furthermore, the .NET Framework does
GetHashCode Made Easy - Learn how to implement GetHashCode as quickly and as simply as of little used C# operators and magic numbers with little explanation for
Overriding Equals in C# (Part 2) - The Importance of GetHashCode. As soon as you polish off your Equals implementation, Visual Studio will start whining at you about the fact
Why is string.GetHashCode() different each time I run my program in - Check out the home page for up to 50 C# blog posts in December 2018! In this post I describe a characteristic about GetHashCode() that was
The Right Way to do Equality in C# – Aaronontheweb - The Right Way to do Equality in C#. and; [Special cases] Override object. GetHashCode() using a high-entropy function, with caveats.
The origin of GetHashCode in .NET - NET platform with the help of the GetHashCode method. Enter method or the C# construct lock), CLR searches a free synchronization block
C# String GetHashCode() method - C# String GetHashCode() method for beginners and professionals with examples on overloading, method overriding, inheritance, aggregation, base,
c# default equality comparer
EqualityComparer<T>.Default Property (System - The first search does not specify any equality comparer, which means FindFirst uses EqualityComparer<T>.Default to determine equality of boxes. That in turn uses the implementation of the IEquatable<T>.Equals method in the Box class. Two boxes are considered equal if their dimensions are the same.
What is the default equality comparer for a set type? - 2 Answers. It means it will use the comparer returned by EqualityComparer<T>.Default for the element type T of the set. As the documentation states: The Default property checks whether type T implements the System.IEquatable interface and, if so, returns an EqualityComparer that uses that implementation.
Understanding equality and object comparison in .NET framework - the object equality. Here is a equality comparer for our employee class. Assume the Employee class doesn't override Equals and provide default reference comparison. You want to . Posted in .NET. Tagged articles, C#.
The Right Way to do Equality in C# – Aaronontheweb - One of the pitfalls of doing development in C#, Java, C++, or really any predominantly Equality by reference is the correct default in that case.
EqualityComparer - Posts about EqualityComparer written by Sean. By default, you can't use equality operators on objects whose type is a type argument within a generic class.
C# Journey into struct equality comparison, deep dive - Deep dive on struct equality comparison, stepping into . var equalityComparer = EqualityComparer<T>.Default; for (int index = 0; index < this.
How to compare two objects (testing for equality) in C# - It's common to compare two objects in C# for equality, such as that we get the object class and everything that comes along with it by default.
EqualityComparer<T>.Default, IEquatable<T> and boxing · Issue - The implementation of EqualityComparer.Default makes me think that calling innocent ContainsKey of Dictionary will produce garbage even for value The C# code that you see here is not what actually runs in UWP apps.
c# - Declarative type comparer - Ordinal; case StringComparison.OrdinalIgnoreCase: return (IEqualityComparer< TType>) StringComparer.OrdinalIgnoreCase; default: throw
Distinct with a custom equality comparer – Levi Botelho's Coding Blog - C# does not offer a Distinct overload that takes a Func<T, T, bool> to handle the ReadLine(); } class DoubleComparer : EqualityComparer<double> { public OrderBy(x => x); var lastElement = default(T); var result = new | http://www.brokencontrollers.com/article/10727606.shtml | CC-MAIN-2019-26 | refinedweb | 2,296 | 54.52 |
Beat Any Website into Shape with Greasemonkey
Greasemonkey is a Firefox extension by Aaron Boodman, which allows you to run personal DHTML scripts (known as "user scripts") on any Website. User scripts are essentially bookmarklets that run automatically, but because they do, they’re a whole lot more useful.
Ever used a site that doesn’t work in Firefox, even though it could with just a few minor tweaks? possible applications, and discuss any issues that arise.
We’ll go from simple to increasingly complex examples, and I’ll be assuming you’re already familiar with JavaScript and the DOM; but all except one are working user scripts, so you can install and use them whatever your JavaScript skills. You will of course need Firefox and the Greasemonkey extension; if you don’t already have them, you can download them from the following links:
You might also want to download the code archive for this article — it contains all the scripts we’ll use here.
Before we start, I want to take a brief overview of how to install and write user scripts for Greasemonkey; if you’re already familiar with the extension, you may want to skip straight to the section headed Putting it to use.
Installing User Scripts
Greasemonkey user scripts are .js files with the name convention "scriptname.user.js". You can install the scripts in a couple of ways:
- Right-click on a link and select Install User Script… from the context menu.
- View the script in Firefox and select Install User Script… from the Tools menu..
Writing User Scripts
Greasemonkey user scripts run on the document once the DOM is ready, but before the onload event. They have no special rules, but it’s considered best-practice to wrap them in anonymous functions, so they don’t interfere with other scripting:
(function()
{
... scripting ...
})();
Apart from that, I have one suggestion to make for writing user scripts: wherever possible, use only well-supported DOM scripting techniques.
Specifically, I try to avoid technologies like XPath, which are not very widely supported and, likewise, non-standard properties like
document.body and
innerHTML, which may not exist in XHTML mode (on pages served as application/xhtml+xml or equivalent).
These guidelines may seem pointless since we’re only writing for Firefox, but we’ll improve the chances of wider compatibility this way (Firefox is not the only browser in which user scripting is available; but we’ll talk more about that later).
Meta-data Comments
Greasemonkey has a simple comments syntax, which is used to define the sites that a script should or shouldn’t run on, the script’s name, it’s description and namespace (a URI namespace as with XML, in which the name of the script must be unique). Here’s a summary:
// ==UserScript==
// @name Script name
// @namespace
// @description Brief description of script
// @include*
// @exclude*
// ==/UserScript==
The syntax is pretty self-explanatory, but the Greasemonkey authoring guide goes into more detail.
Putting it to Use
The development of Greasemonkey was heavily inspired by Adrian Holovaty’s site-specific extension for All Music Guide, which is designed to fix what many people consider are serious problems with the new site design. The developer’s aim was to make it as easy to write a site-specific extension as it is to write DHTML. And so he has.
Simon Willison’s Fixing MSDN with Greasemonkey was among the first to make popular use of Greasemonkey, presenting a script that reveals information that would otherwise only be visible to IE, while Mihai Parparita’s Adding Persistent Searches to Gmail is by far the most sophisticated example I’ve seen to date.
But we’re going to dive in with a few simple site enhancements, to introduce the possibilities, as well as the practicalities of user scripting. I made a blank template for Greasemonkey scripts that you may find useful; it also includes some handy methods.
Customising Sites you Visit Often
Auto-complete Forms
Our first example auto-completes a single form element on a specific page. In this case, the page is the SitePoint Forum homepage, and the value is my username:
View or install the script auto-complete.user.js from the code archive.
Even though the forum remembers me with a cookie, I might not have it, since I clear all my cookies quite often. There are plenty of other times where this could come in handy: a cookie might have expired, or a site might have more basic functionality that doesn’t remember people.
Let’s extend the principle by adding another site: the Yahoo! mail login page. First we would add the @include path to the top of the comments section:
// @include
We then add the appropriate scripting; in this case, the login field doesn’t have an id, so we’ll have to find it through
document.forms:
//look for the yahoo mail login form
var yahoo = typeof document.forms['login_form'] != 'undefined'
? document.forms['login_form']['login'] : null;
if(yahoo != null)
{
//write your username to the field
yahoo.value = 'brothercake';
}
You can add further sites as you like, defining for each an @include domain, and a chunk of code to find and complete its form field (changing the username to yours, obviously!).
Change the Layout
A user script can do much more than write a single element: it could rewrite the entire DOM of any page, making it look and behave exactly the way you want. In this next example, I’m going to modify the default Slashdot front page. I’ll remove the ad frame from the top and replace it with a duplicate of the site-search form, which is otherwise displayed only at the bottom:
View or install the script slashdot-restructure.user.js from the code archive.
As with the auto-complete example, or, for that matter, any user script that modifies a specific page, I only know which elements to change because I looked at the source code. But I don’t control it, and somewhere down the line it will probably change.
That’s an obvious point, but nonetheless, it’s important: anything we write that makes assumptions about the DOM will stop working one day, and we can’t do anything about that.
However, we can at least prepare for it by making sure we test for the existence of something before we modify it. That’s good practice anyway, but it becomes doubly-important here. In fact, it’s the purpose of the tests against null in the Slashdot example script; things like this:
//the first center element contains an ad frame
var adframe = document.getElementsByTagName('center')[0];
//remove it
if(adframe != null) { adframe.parentNode.removeChild(adframe); }
Now, if the script does fail, at least it won’t throw any errors.
Customise Search Results
I’ve often though it would be cool if search results could come back as a tree menu, with +/- icons to allow users to show and hide the summaries. Thankfully, Google results have a predictable (if soupy) structure that’s easy to iterate over, so it’s a fairly simple task to extract the relevant information and reformat it into a dynamic list:
View or install the script google-tree.user.js from the code archive.
The only serious complication we’ll face in parsing the HTML is that the format varies when the result includes a translate link (it has an additional containing
<table>). But, provided we allow for that, we can reliably identify the summary of each result, for which we then add show/hide behaviours that are triggered from a dynamically-created icon.
The behaviours come from an onclick handler on the icon, which works with both mouse and keyboard navigation. The use of javascript:void(null) in the link href simply provides an href value; otherwise the link wouldn’t be keyboard navigable (because Mozilla can’t set focus on a link that has no href). It’s essentially the same trick as using
href="#", but this method is cleaner: it doesn’t do anything else, so you don’t have to control its return value the same way.
Improving Web Usability
All the examples we’ve looked at so far are designed for specific Websites. But user scripts can be made to run on every Website, and this is perhaps where the idea becomes remarkable: you can write user scripts to change how the whole Web behaves. Effectively, you can customise your browser just by writing some simple JS.
The practical considerations are slightly different here, because, generally speaking, we won’t be looking for specific structures: we’ll be making something new, or modifying widely-used attributes. As such, the scripts in this example are generally more forwards-compatible than previous examples.
Change Link Targets
An obvious application, this script identifies links that have a
"_blank" target attribute, and retargets them to
"_self":
View or install the script target-changer.user.js from the code archive.
This script uses a single document onclick handler to change the link targets on-the-fly, rather than iterating through all links and changing them in advance. We do this partly because it’s more efficient, but mostly so that the script will still work if the target has been set by other scripting (remember that Greasemonkey user scripts run before the
window.onload event).
With a more complex script you could go much further: change or remove onclick events, rewrite javascript: URIs, even override the
open() method completely. But with such aggressive measures comes the risk of blocking out functionality you do want: if the
open() method is overridden, some links might end up doing nothing. That’s the kind of thing you’d probably only want to do for specific sites, not for the Web in general.
Remove Ads
The difficulty with removing ads is that not all of them are unwanted. Some people don’t like any advertising at all, while for many, simple text-based ads are acceptable and only large or animated graphical ads are deemed intrusive. The source of the ads may also be relevant: maybe particular companies or sites are more likely to carry advertising you’re interested in, or are inclined to trust.
Any ad-blocking program will inevitably have limits to the degree of precision with which it can be customised, but with user scripting there are no such limitations: you can tailor the script exactly how you like, and avoid any unwanted interference.
In this example I’ve split the scripting into methods based on the type of advert — whether it’s an image, iframe, or flash ad; each object is further tested to see if it comes from a known ad server:
View or install the script remove-ads.user.js from the code archive.
If you don’t want to remove a certain type of object, simply comment-out its method call at the bottom of the script:
//instantiate and run
var rem = new removers();
rem.banners();
rem.iframes();
rem.flash();
If you want to extend the list of ad servers, you can add as many as you like to the domains array at the top of the object constructor:
//list the domains or subdomains from which ads might be coming
this.domains = ['doubleclick.net','servedby.advertising.com'];
And, if you’d prefer to allow certain sites to show their advertising, whatever it may be, you can add those sites as
@exclude comments:
// @exclude*
The principle could be taken much further, to control to the nth degree whether an element is removed or not — you could differentiate by size, for example, images that are 468 x 60 pixels. Or you could allow Flash only if it’s embedded using
<object>, not if it uses
<embed>. This would filter out most advertising, while leaving compliant Websites alone!
Correct Language and Spelling
A study of commonly misspelled words, by Cornell Kimball, analysed Internet (Usenet) newsgroups to discover the most common misspellings. Using some of that data to construct regular expressions, this script makes text-replacements to correct common errors:
View or install the script language-corrector.user.js from the code archive.
You could extend the idea by adding profanity filters to the list: converting unacceptable language into "***" or any empty string. But do be careful with any words that might also be substrings of other words, such as "ham" is to "gingham". (There’s a story of a local authority in the British town of Scunthorpe that went without email for an entire day because of a crucial omission in their mail filtering rules!) You may have to test for a word, plus a leading and trailing space or punctuation, to ensure you don’t have this kind of problem.
A more sophisticated example could scan individual pages for complicated terminology, or technical phrases which are not properly explained on the site, adding
<abbr> or
<dfn> where required. You could even construct links directly to an online dictionary or other reference source.
Google Site-search
A site that lacks a search facility can be frustrating to use, but Google allows you to search within individual sites simply by appending site:domain.com to the query. This script uses that syntax to create a site-specific search box on every page.
Or, at least … that was the plan:
View or install the script site-search.user.js from the code archive.
The scripting is solid, the idea sound, but it has quite a serious problem: it doesn’t work.
I wouldn’t usually publish a script if I couldn’t get it to work, but I think this is such a good idea it would be a shame not to include it, and anyway, I wanted to look at the questions it brings to light.
As you’ll see if you try it, the core scripting works but you can’t type into the textbox, even though you can copy and paste into it, and then press Enter to submit. Further investigation reveals that the
"-moz-user-modify" property has a value of
"read-only" (although explicitly setting it to
"read-write" doesn’t help).
Now I did actually think of two ugly hacks to get around this, and they both work, but I’m very reluctant to use them, because I think they challenge a security restriction. The script creates a user-interface element, and perhaps Firefox extensions aren’t allowed to initiate user input in the DOM of the page you’re viewing. I’m making guesses here, because I haven’t been able to find the facts, but if that, or something like it, is the case, I suspect it will also be true for any form-based user scripting, making such things basically untenable.
If anyone can shed any light on this, I’m all ears!
Onfocus Tooltips
I’m often frustrated by the fact that title attribute tooltips don’t show up from keyboard navigation — they work only when you use the mouse. This script compensates for that by creating tooltips that are triggered by focus events:
View or install the script onfocus-tooltips.user.js from the code archive.
The scripting is quite straightforward; it creates a single element and writes in the title text (if any exists). The complications lie in the positioning of the tooltip: it has to position itself relative to the triggering element, then compensate for potentially being located outside the window, or below the fold.
It works on links, iframes, objects and form elements — those that can receive the focus — and creates tooltips that are styled using CSS2 System Colors.
Headings-navigation Bar
Most serial browsers (screenreaders like JAWS, and text-only browsers like Lynx) have a "headings mode" or something like it, where they list and link to all the headings on the page. It provides a greater degree of random access (on pages which use proper headings).
This script emulates that functionality, creating a small "H" icon at the top-right of the page, from which a menu drops down containing links to every heading:
View or install the script headings-navigation.user.js from the code archive..
Persist Alternate Stylesheets
The final example in this section begins with a well-meaning script, but opens a can of worms along the way…
One issue with "alternate" stylesheets is that changes made using most browsers’ built-in switcher don’t persist between pages, making it little but a novelty feature unless the author intervenes with cookies. Perhaps, in future, more browsers will implement this behaviour, but until then, we can write a user script to do it for us.
This example looks for stylesheets that are included via
<link> elements, and saves their disabled state to a cookie whenever it changes; the script then re-applies the last state to subsequent page views. This effectively adds domain-specific persistence to sites that use native stylesheet switching:
View or install the script persist-stylesheets.user.js from the code archive.
It works because Firefox updates the stylesheets’ disabled property with changes from the native switching mechanism, so we can implement persistence simply by continually getting and saving that data to a cookie.
This brings us to the first significant point, and the lid of our can of worms: the cookie we’re using is in a domain we don’t control. From a practical perspective, this means that if the site already uses cookies a lot, we run the risk of filling up the data limit (4K) and thereby overriding data that the site needs. We can’t avoid this possibility (short of not using cookies at all), but we can reduce the risk by keeping the data as small as possible. You could also reduce the impact by having a list of specific
@include domains, where you know the feature is needed, rather than running it on every site that uses alternate stylesheets.
But suddenly there are security implications here, as well: it would be very easy to write a user script that steals the cookie and other data from every site it encounters, then sends it all somewhere else.. It certainly seems unfortunate, if, with more people using Greasemonkey, we can no longer predict the DOM of our Websites; but the question is really moot, because that’s already the case. Anyone can make any page appear any way they want in their own browsers; many browser add-ons, screenreaders and other user-agents will change, rewrite or add to the DOM of the page before any client-side scripting is run.
The capabilities Greasemonkey gives us are no different, apart from the larger number of people who might use it, and I suppose that’s really the rub — especially when it comes to things like the impact on advertising revenues. Just like TiVo (a digital TV system that can filter out ad-breaks from recordings), this becomes a threat only if lots of "ordinary" people use it.
But the security concerns are very serious, and the Greasemonkey blog takes it up by suggesting that browsers shouldn’t be able to make http requests outside the current domain without a user prompt. That could certainly reduce the problem of data stealing, though not without unwanted side-effects (what about the impact on legitimate remote syndication?). And it doesn’t eradicate the problem completely, as the author goes on to concede: nothing would stop a user script from rewriting page links via another server, adding query-string data along the way.
I think the implication is this: non-technical users must be able to trust the source of a user script they install. One possible solution might arise if an archive of user scripts was made available from a verifying source, as Firefox extensions are. This wouldn’t preclude developers from offering scripts directly, or users from downloading them, but it would give non-technical users a place where they could be sure that whatever they’re installing is safe. And in fact the ever-active GM development team are looking into ways of doing this, with a userscript.org Website planned to manage a catalogue of available scripts.
Developer Tools
User scripts might also represent a niche for the creation of developer tools — scripts that you can write and customise for specific projects, or keep as a collection of general tools that are enabled or disabled as required.
The obvious comparison with bookmarklets comes to mind again, as does the fact that user scripts run automatically, so we can do things that wouldn’t be possible with a run-once, manually initiated bookmarklet.
Word Count
Word counters are useful tools, but it’s not the kind of thing you’d expect in a browsing toolset, and it wouldn’t be much use for general surfing. But for development — and particularly for writing articles directly in HTML, as I always do — it’s handy to have a counter going as you write.
View or install the script word-count.user.js from the code archive.
This script uses a similar technique to the language corrector we looked at earlier, extracting the text from a page by iterating through elements and looking for their child text-nodes. It then splits and processes the text to count the number of words, and displays the result in the
<title> attribute (adding to, rather than overwriting it).
Viewport Size
This short script simply reads the viewport (inner window) dimensions using Mozilla properties, and displays that data in the
<title>:
View or install the script viewport-size.user.js from the code archive.
Fun with Greasemonkey
I was delighted to realise the comedic angle to user scripting … It reminds of a job I had once, where we used to amuse ourselves by writing bookmarklets that made subtle and not-so-subtle changes to the company Website. It made us laugh … and it’s all ultimately harmless, so if you want to make serious Websites look silly, I’m all for showing you how. Here are a couple of simple scripts that make the Web a more entertaining place!
Name that Celebrity
The names of politicians, industry leaders, and other well-known people are ripe for transformation into comical phrases … I’ll show you the basic script by making a few… er… adjustments to some names you’ll undoubtedly know:
View or install the script name-that-celebrity.user.js from the code archive. from the code archive.
User JS in Opera and Beyond
Opera has been planning user JavaScript for a while, and the feature is now available in Opera 8, having been introduced to the beta version and heralded by Hallvord Steen in a recent Journal entry. (Though, amusingly, this isn’t the first implementation to go public — the Opera 7 "Bork" edition loaded a JS file into MSN pages, to transform them into the language of the Muppets’ Swedish Chef!)
User scripting works similarly to Greasemonkey in Opera. The browser does implement a direct compatibility mode, but it also offers additional functionality in the form of special-purpose events from the
window.opera object, and is really only intended for power-users, as some manual configuration is required. Rijk van Geijtenbeek has the most comprehensive resources for user scripting in Opera, and converting this article’s examples shouldn’t be difficult for the most part, especially since we planned for something like this from the beginning.
The workings and permissions of user scripting may have to change, or cause browsers to change, in light of the ways in which people use it. But, whatever else happens, it’s clear that user scripting is here to stay.. But there are strong notes of dissension in several corners, and perhaps it won’t be long before we’re deluged with anti-greasemonkey coding techniques, followed by anti-techniques to those!
I’ll be watching this develop with great interest. | https://www.sitepoint.com/beat-website-greasemonkey/ | CC-MAIN-2020-10 | refinedweb | 3,995 | 56.18 |
I am just trying to figure out basic classes/objects. I am trying to write a program where it will return the area of a triangle. I realize this can be done easier another way but I am just trying to figure out how these things work....
Code :
import java.util.Scanner; class Triangle { double height; double base; double getArea() { return 1/2 * height * base; } } public class study { public static void main(String[]args) { Triangle Triangle = new Triangle(); Scanner in = new Scanner(System.in); System.out.println("Height?"); Triangle.height = in.nextInt(); System.out.println("Base?"); Triangle.base = in.nextInt(); Triangle.getArea(); System.out.println("The area of a triangle with height " + Triangle.height + " and base " + Triangle.base + " is " + Triangle.getArea()); } }
It keeps printing the null value 0.0 for area of a triangle. | http://www.javaprogrammingforums.com/%20whats-wrong-my-code/29152-objects-classes-printingthethread.html | CC-MAIN-2016-18 | refinedweb | 134 | 52.05 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.