workspace
stringclasses 4
values | channel
stringclasses 4
values | text
stringlengths 1
3.93k
| ts
stringlengths 26
26
| user
stringlengths 2
11
|
|---|---|---|---|---|
elmlang | general | try `elm install elm/svg` | 2019-04-27T18:29:45.095000 | Wenona |
elmlang | general | Derp | 2019-04-27T18:29:49.095200 | Isaias |
elmlang | general | :thumbsup: | 2019-04-27T18:29:54.095400 | Wenona |
elmlang | general | Every example I come across for using ports via commands and subscriptions the signature for subscriptions is `Model -> Sub msg` but there's no examples of anyone actually using the model in their subscriptions...
Anyone here have a use case where they've done something with the model in a subscription function? | 2019-04-27T18:35:16.096900 | Buffy |
elmlang | general | I don’t have code-examples handy, but one example I can offer is that we add a `keydown` subscription when a modal is open, so we can close it when `escape` is pressed | 2019-04-27T18:40:04.097900 | Huong |
elmlang | general | we could have the subscription always be there and filter in the model, too | 2019-04-27T18:40:16.098600 | Huong |
elmlang | general | but then that’s always the case, I suppose - ignoring the message based on the state, or not subscribing to the event based on the state, it’s just a matter of how much work needs to happen | 2019-04-27T18:41:05.100100 | Huong |
elmlang | general | I wish I followed that but maybe YAGNI till I do for now? | 2019-04-27T18:50:34.101100 | Buffy |
elmlang | general | The other thing I'm curious about is if subscriptions can be attached if you're expecting a response of some sort from them and otherwise don't bother the run time? Not sure if that makes any sense. | 2019-04-27T18:51:46.102300 | Buffy |
elmlang | general | Yeah, that’s basically what the `Model ->` part is for - based on your model, you can decide whether you are interested in receiving certain events, and hence, subscribe to those events | 2019-04-27T18:56:43.103500 | Huong |
elmlang | general | An example of why you might need `model` is if you have a specific page in your app where you want to track time passing, but you don’t care about it otherwise. One solution is you could have a value in `model` that lets `subscriptions` know whether it should subscribe to `Time.every` or not. While you could just run the event all the time regardless and filter out what you don’t need in `update`, I feel `subscriptions` is a better place for it because: 1. you’re only running `update` constantly when necessary; and 2. it separates the logic of “do I care about this subscription?” from “how should I update my model and what effects need to happen?” | 2019-04-27T18:57:45.103700 | Naoma |
elmlang | general | Now it makes a lot of sense! Thanks y'all! Now if only I wasn't using all this just to wrap aws-amplify lib's js calls... :confused:
Not sure if interacting directly with the Http API's would be easier starting to feel like it would but the Cognito docs are dense. :man-shrugging: | 2019-04-27T19:19:43.105600 | Buffy |
elmlang | general | Sorry <@Myong> :slightly_smiling_face: | 2019-04-27T19:20:06.106100 | Buffy |
elmlang | general | Anybody encountered an issue with `elm publish`? I published new version of package yesterday but it's not visible on <http://package.elm-lang.org|package.elm-lang.org>
package <https://package.elm-lang.org/packages/turboMaCk/lazy-tree-with-zipper/>
tag <https://github.com/turboMaCk/lazy-tree-with-zipper/releases/tag/3.1.0>
```
❯❯❯ elm publish
Verifying turboMaCk/lazy-tree-with-zipper 3.1.0 ...
● Found README.md
● Found LICENSE
✗ Version 3.1.0 is not correct!
-- ALREADY PUBLISHED -----------------------------------------------------------
Version 3.1.0 has already been published. You cannot publish it again!
Try using the `bump` command:
elm bump
It computes the version number based on API changes, ensuring that no breaking
changes end up in PATCH releases!
``` | 2019-04-28T06:48:52.107600 | Zona |
elmlang | general | <@Zona> I can see that version | 2019-04-28T06:53:36.107800 | Huong |
elmlang | general | Caching issue? | 2019-04-28T06:53:40.108000 | Huong |
elmlang | general | let me check | 2019-04-28T06:54:03.108200 | Zona |
elmlang | general | righ... seems like aggressive client side caching I didin't expect that :smile: | 2019-04-28T06:54:46.108800 | Zona |
elmlang | general | maybe it would be good idea to tweak headers to take that cache down a bit | 2019-04-28T06:55:06.109300 | Zona |
elmlang | general | thx <@Huong> | 2019-04-28T06:56:13.109600 | Zona |
elmlang | general | Are you aware about a Elm repl online "fully in the browser"? There is : <http://elmrepl.cuberoot.in/> but it seems each keystroke is sent to the server, which is a bit slow... <@Florencia> is your elm compiler-in-elm ready enough to support this kind of thing? | 2019-04-28T09:39:49.111500 | Loralee |
elmlang | general | is there a way to decorate pointer events w the time at which it happened? something like Task.pair Time.now (pointer event)..though i know mpizenberg doesnt expose subscriptions. | 2019-04-28T11:58:55.116000 | Rozanne |
elmlang | general | You could use the advanced section of the pointer events package and decode this standard field from the event <https://developer.mozilla.org/en-US/docs/Web/API/Event/timeStamp> | 2019-04-28T12:19:18.117500 | Lashawnda |
elmlang | general | <https://package.elm-lang.org/packages/mpizenberg/elm-pointer-events/latest/Html-Events-Extra-Pointer#eventDecoder> | 2019-04-28T12:20:59.118000 | Lashawnda |
elmlang | general | Depending on the browser this value will have a decimal component, so if you want a `Posix` you’ll have to round it first | 2019-04-28T12:25:31.119300 | Lashawnda |
elmlang | general | Note that this is a timestamp relative to when the document was loaded, not an absolute point in time | 2019-04-28T13:01:33.119600 | Huong |
elmlang | general | Definitely not yet :) | 2019-04-28T13:22:14.119900 | Florencia |
elmlang | general | but it's one of the planned usages | 2019-04-28T13:23:10.120100 | Florencia |
elmlang | general | hi | 2019-04-28T14:02:08.120500 | Socorro |
elmlang | general | Will union types be comparable some day? | 2019-04-28T14:38:27.121600 | Isaias |
elmlang | general | I don’t think a decision has been made yet on that particular issue | 2019-04-28T14:43:33.121700 | Huong |
elmlang | general | Is there a discussion thread somewhere <@Huong>? | 2019-04-28T15:49:03.122000 | Isaias |
elmlang | general | There is a meta-issue that tracks a number of tickets related to more flexibility for `comparable` types, of which custom types is only a small part. As per usual, the goal is to (eventually) find a holistic design rather than solving just the immediate issue, thereby possibly excluding other options in the future (or making them harder to add) - <https://github.com/elm/compiler/issues/1008> | 2019-04-28T15:56:57.122200 | Huong |
elmlang | general | Various designs have been proposed and discussed, like “use the definition order” - which has significant downsides, or a bespoke `data module where { type = .., compare = .., eq = .. }` syntax for making data-type modules that would allow making them comparable and allow overriding structural equality, but that also has significant downsides. I’m sure there have been others, though. The point is mostly this:
- this is not an easy issue
- making a decision means committing to a particular design, to the exclusion of other options
- this does not have the highest priority right now, as there are workaround for most immediate issues stemming from the limited `comparable` type variable | 2019-04-28T16:03:37.122400 | Huong |
elmlang | general | Ok thanks for the detailed response | 2019-04-28T16:28:22.122700 | Isaias |
elmlang | general | Ontological question: Suppose you have types `A` and `B`, and `A` has an `A.fromB` constructor.
Say you want to also make `B`'s from `A`'s. Do you make a `B.fromA` constructor or an `A.toB` deconstructor? | 2019-04-28T16:37:09.129800 | Isaias |
elmlang | general | (I realize it might not be obvious what downsides there are to using definition order, so just to give an example - this means that if I write a library that exposes a datatype that happens to be implemented as a custom type, people might rely on the sorting order. Changing the order would then suddenly affect consumers of my library. That would be surprising and definitely lead to issues. Things would be even worst if my type is opaque - the user can’t even see what order things will be sorted in, but that implementation detail now leaks. There are arbitrary ways to limit the impact - only make simple enumerations `comparable` and/or require that the constructors be exposed - but that obviously complicates the design and means people will still ask for `comparable` custom types :smile: TL;DR: it means leaking implementation details, which is tricky business) | 2019-04-28T16:37:22.129900 | Huong |
elmlang | general | I usually prefer `toFoo` functions, and always feel a little bad writing `String.fromInt` or `String.fromFloat` :smile: (this was for me also a point of confusion while learning Rust. While using `.into()` is a common pattern there, this is usually done by implementing a `From<T>` trait on the target type, rather than an `Into<T>` trait on the originating type) | 2019-04-28T16:40:28.132500 | Huong |
elmlang | general | why do you prefer that? | 2019-04-28T16:51:28.132900 | Isaias |
elmlang | general | Consistency, mostly. For whatever reason, it also matches my mental model better - I’m more likely to think “I want to turn X into Y” rather than “I want to create a Y from an X” | 2019-04-28T16:55:08.134300 | Huong |
elmlang | general | Ah, the "source-path-goal" image schema :laughing: | 2019-04-28T17:26:03.135200 | Isaias |
elmlang | general | never thought to check the other fields! thanks much. also great to know about the timestamp. | 2019-04-28T19:47:58.136700 | Rozanne |
elmlang | general | Hi, I know how to decode JS object into Elm alias, but I'm not sure how to decode JS object into Custom type with more values.
```
json = { name: "Joe", unit: "CELSIUS }
// elm
type Unit = Celsius | Fahrenheit
type alias Name = String
type Setting = TemperatureSetting Name Unit
```
I would like to decode that json into `TemperatureSetting`. Could someone give me a hint? Thanks | 2019-04-29T02:30:17.140600 | Bethel |
elmlang | general | <@Bethel> first you need to write a `String -> Maybe Unit` function | 2019-04-29T02:34:58.141400 | Earlean |
elmlang | general | that one I got. | 2019-04-29T02:35:40.141900 | Bethel |
elmlang | general | I'm just not sure how to replace with
```
Json.map2 SettingAlias (field "name" string) (field "unit" decodeTemperature)
``` | 2019-04-29T02:36:25.142900 | Bethel |
elmlang | general | the `SettingAlias` is not alias anymore but just custom type named `Setting` | 2019-04-29T02:36:48.143400 | Bethel |
elmlang | general | `TemperatureSetting` is a constructor for the type `Setting`, it takes a String and a Unit and gives you a `Setting` | 2019-04-29T02:37:26.144000 | Earlean |
elmlang | general | `Json.map2` takes a function as it's first parameter | 2019-04-29T02:37:44.144400 | Earlean |
elmlang | general | ha, I got it. Thank you | 2019-04-29T02:46:32.144700 | Bethel |
elmlang | general | How can I decode an array of objects of various shapes and I need to find just one exact shape in it and get that one.
```
let arr = [ { name: "Jon" }, { type: "Hot" }, { age: 16 } ]
```
and I would like to get the `age` object. | 2019-04-29T02:58:37.146700 | Bethel |
elmlang | general | in brief, you cannot, as Elm lists have to contain all the same type. | 2019-04-29T03:54:39.147300 | Lynn |
elmlang | general | you’ll need to use a CustomType to express the variations that you have | 2019-04-29T03:54:56.147700 | Lynn |
elmlang | general | ah, sorry, read more carefully | 2019-04-29T03:55:19.148100 | Lynn |
elmlang | general | `Decode.field "age" <http://Decode.int|Decode.int> |> Decode.maybe |> Decode.list |> Decode.map (List.filterMap identity)` | 2019-04-29T03:56:11.149100 | Lynn |
elmlang | general | i've been thinking about a possible "solution" to this for a while, but i'm not sure where to put it^^
it's not really polished yet either, as i've never taken the time to really write it down | 2019-04-29T04:28:40.150200 | Emilee |
elmlang | general | To decode all content of the list, maybe <https://package.elm-lang.org/packages/elm/json/latest/Json-Decode#oneOf> could help | 2019-04-29T04:29:05.150700 | Millie |
elmlang | general | If there is the possibility, I'd change the API model instead and go with a dict/object to tackle the root cause. | 2019-04-29T04:31:06.150800 | Millie |
elmlang | general | There’s an additional point, actually: perhaps “custom types being comparable” is an XY-problem! Perhaps the most common reason for this request is people who wish to make a `Dict` or `Set` with custom types - in that case, them being comparable is a prerequisite for that to be possible. However, that’s not the only possible resolution - there is ongoing work for making hashmaps (and hashsets) a thing, and those would not require a `comparable` constraint. | 2019-04-29T04:32:39.151000 | Huong |
elmlang | general | yeah, i also approached the issue differently. in my concept i'd try to move away from implicit typeclasses like `comparable`
what i was thinking is that typeclasses always have the side effect of making things implicit, but elm a lot of the time is about making things explicit.
when we look at a typical problem of "i want to have a sorted datastructure" the explicit method - passing the comparison function - has one of two poblems:
• either we have to pass the function with every operation, which is clumsy and also error prone
• or we have to put the function in the data structure, which makes it hard to work with in the model
*but*: what if you could put the function in the type?
i'm not exactly sure yet how this would be best implemented to maintain readability in elm, but if you could annotate your type with a function, you could have a `SortedDict (sortFunc : a -> a -> Ordering) a` or something.
This would, I think, cover most cases where people with a background in haskell would reach for typeclasses, and still maintain explicitness.
A major issue here is how it interacts with the existing syntax, and how to make it clear when we talk about a function type, and a function *as type* ^^' so yeah...
also, i'm not yet sure how this could be abused, so that's a concern too. | 2019-04-29T04:40:21.151200 | Emilee |
elmlang | general | are there any plans to add memoization to Elm? | 2019-04-29T05:56:06.152300 | Nana |
elmlang | general | Not that I know of. Do you have a *specific* use-case where you feel like you would benefit from this? Seeing a concrete case with code would help (and would allow people to suggest workaround that apply to your case :slightly_smiling_face: ) | 2019-04-29T06:22:40.152400 | Huong |
elmlang | general | well, I have quite a lot of places where I sort/filter a long list of items, before displaying it in the view, for example as options in select boxes
the issue with select boxes is that they also have an index which keeps track of which item the user is hovering over / has selected with up/down keys, so because of that `lazy` doesn't work | 2019-04-29T06:28:44.152600 | Nana |
elmlang | general | I could of course manually cache the result in the model, but it'd be cleaner to have a memoization function I think | 2019-04-29T06:30:43.153600 | Nana |
elmlang | general | If the input to the function changes, memoizing wouldn’t help either, would it? Or, no more than `lazy` would, I’m thinking? | 2019-04-29T06:30:50.153800 | Huong |
elmlang | general | oh, wait, I can see how you’d do that | 2019-04-29T06:31:11.154000 | Huong |
elmlang | general | so I'd memoize just the sorted/filtered list, but not the selection index | 2019-04-29T06:31:48.154200 | Nana |
elmlang | general | so I’m assuming you have a long list of things, and a second variable that identifies a bunch of those things as being hovered or what not. Right, yeah. | 2019-04-29T06:31:55.154400 | Huong |
elmlang | general | gotcha. Yeah, the current recommended workaround, after identifying that this is in fact a performance problem, would be caching it on the model | 2019-04-29T06:32:57.155000 | Huong |
elmlang | general | Why is there no ternary operator in Elm? We have if..then..else, but elm-format insert a whole lot of space and newlines in it.
```
ternary : Bool -> a -> a -> a
ternary condition valueIfTrue valueIfFalse =
if condition then
valueIfTrue
else
valueIfFalse
``` | 2019-04-29T06:34:45.155700 | Una |
elmlang | general | and I think it'd be useful for a lot of similar things | 2019-04-29T06:35:43.155900 | Nana |
elmlang | general | i hade similar problem, but for cache some precalculated data, so i got structure like:
`{data:{..}, cache:{valid:Bool, cachedData:{...}}}`
so then i had some functions that needs that cache, and they return not just result, but also updated element, | 2019-04-29T06:45:17.156500 | Liza |
elmlang | general | but that not a case if You need that inside your view.. | 2019-04-29T06:45:39.156700 | Liza |
elmlang | general | or just before exit update function, you can validate cache and if needed recalculate it.. | 2019-04-29T06:46:14.156900 | Liza |
elmlang | general | I think validating the cache on every update could get expensive though? If it uses structural equality on big lists | 2019-04-29T06:47:42.157100 | Nana |
elmlang | general | that is just `===` in js.. | 2019-04-29T06:48:12.157700 | Liza |
elmlang | general | or maybe not? because it first checks reference equality right? | 2019-04-29T06:48:16.158000 | Nana |
elmlang | general | It first checks reference equality, yes | 2019-04-29T06:48:27.158300 | Huong |
elmlang | general | for example - in my case that was SAT Shape - that had cached normals and projections, so if i change (by some function) position, with, or other component that impacts cache - it was set to “invalid”, and there was (LOT of) functions that not care about that cache - so most of cases that invalid cache can stay for long time, and then if i needed (some one came too near to collide ) cache was recalculated, and that stay like that again for long time.. | 2019-04-29T06:51:20.158500 | Liza |
elmlang | general | and validation of cache was just boolean stored in cache it self.. | 2019-04-29T06:51:37.158700 | Liza |
elmlang | general | so in your example - it would be just, set to false when change filter, and before exit update (if that list will be shown) check - do that prop (`cache.valid`) isn’t True | 2019-04-29T06:53:01.159000 | Liza |
elmlang | general | this is one of the things I like in Vue, most calculations are done in "computed properties", which are memoized, and you can have computed properties which depend on other computed properties, which means that things only get recalculated when they really need to | 2019-04-29T06:55:11.159200 | Nana |
elmlang | general | so - as result you not need to checks whole list, just manual invalidation each time when you make “cache braking” changes | 2019-04-29T06:55:14.159400 | Liza |
elmlang | general | You could either have that `valid` flag, or write a wrapped around `update` which would look something like ```cacheHandler : Msg -> Model -> (Model, Cmd Msg)
cacheHandler msg model =
let
original = model.source
(updatedModel, cmds) = update msg model
in
if updatedModel.source == original then
-- not changed
(updatedModel, cmds)
else
( { updatedModel | cache = calculate updatedModel.source }, cmds )
``` | 2019-04-29T06:56:29.159600 | Huong |
elmlang | general | Since that check has a bypass for referential equality, things that don’t touch the `source` won’t trigger revalidation very very fast. If the reference did change, you still only revalidate the cache if structural equality _also_ says things have actually changed | 2019-04-29T06:58:02.159800 | Huong |
elmlang | general | <@Huong> ah that's pretty neat, saving the original value in a closure. I was thinking I would have to store a duplicate in the model | 2019-04-29T06:58:21.160000 | Nana |
elmlang | general | point with flag - is that you manually set it, and that can be changed only during some updates, that changes that list, so it will return it any way.. | 2019-04-29T06:58:22.160200 | Liza |
elmlang | general | doing the same with `valid : Bool` flag would mean having to recognise that if your code did something equivalent to `List.map identity source`, that should not invalidate the cache | 2019-04-29T06:59:18.160400 | Huong |
elmlang | general | so - you any way will need to create wrappers around own data structure.. | 2019-04-29T07:00:20.160600 | Liza |
elmlang | general | and there will be “set / get / update”.. that will do all that stuff.. | 2019-04-29T07:01:14.160800 | Liza |
elmlang | general | but that depends on how many devs you have.. and how protected / transparent cache you would like to have | 2019-04-29T07:01:44.161000 | Liza |
elmlang | general | we will hopefully be 2 devs soon, atm it's just me :smile:
<@Huong>'s approach looks good, maybe it can be abstracted into a nice lib :thinking_face: | 2019-04-29T07:17:36.161300 | Nana |
elmlang | general | <https://github.com/avh4/elm-format/issues/209> | 2019-04-29T07:31:36.161500 | Velia |
elmlang | general | a "ternary operator" would be odd when `if` already behaves like an expression, rather than a statement
but yeah hopefully elm-format will be updated some day | 2019-04-29T07:40:30.161700 | Nana |
elmlang | general | Hello <@Huong> I tried these solutions, it doesn't work, due to failure in building elm-package-0.18 | 2019-04-29T08:13:25.161900 | Calista |
elmlang | general | ```Please switch to using either the new project style and the new-install
command or the legacy v1-install alias as new-style projects will become the
default in the next version of cabal-install. Please file a bug if you cannot
replicate a working v1- use case with the new-style commands.
For more information, see: <https://wiki.haskell.org/Cabal/NewBuild>
Resolving dependencies...
Notice: installing into a sandbox located at
/usr/local/elm/Elm-Platform/0.18/.cabal-sandbox
Starting elm-package-0.18
Building elm-package-0.18
Failed to install elm-package-0.18
Build log ( /usr/local/elm/Elm-Platform/0.18/.cabal-sandbox/logs/ghc-7.10.3/elm-package-0.18-GAzqzqXmEIs4OwPrTLlD98.log ):
cabal: Entering directory '/usr/local/elm/Elm-Platform/0.18/elm-package'
Configuring elm-package-0.18...
Warning: 'ghc-options: -threaded' has no effect for libraries. It should only
be used for executables.
Preprocessing executable 'elm-package' for elm-package-0.18..
Building executable 'elm-package' for elm-package-0.18..
[ 1 of 27] Compiling Utils.Paths ( src/Utils/Paths.hs, dist/dist-sandbox-90435201/build/elm-package/elm-package-tmp/Utils/Paths.o )
src/Utils/Paths.hs:15:27: Not in scope: ‘N.toFilePath’
src/Utils/Paths.hs:19:26:
Not in scope: ‘Package.versiontoString’
Perhaps you meant one of these:
‘Package.versionToString’ (imported from Elm.Package),
‘Package.versionFromString’ (imported from Elm.Package)
cabal: Leaving directory '/usr/local/elm/Elm-Platform/0.18/elm-package'
cabal: Error: some packages failed to install:
elm-package-0.18-GAzqzqXmEIs4OwPrTLlD98 failed during the building phase. The
exception was:
ExitFailure 1
elm-reactor-0.18-GWP7BE9OpskDnHmLhmsAmZ depends on elm-reactor-0.18 which
failed to install.``` | 2019-04-29T08:13:58.162300 | Calista |
elmlang | general | I was wondering if `Elm.Foo.init({ element })` was also considered when they decided it should be `Elm.Foo.init({ node })` instead and why was that? | 2019-04-29T09:04:26.168900 | Maxwell |
elmlang | general | presumably because they're called DOM nodes | 2019-04-29T09:04:44.169400 | Danika |
elmlang | general | They're also called DOM elements. | 2019-04-29T09:05:25.170000 | Maxwell |
elmlang | general | `<a>` is an element. A DOM node is a specific instantiation of an element | 2019-04-29T09:05:43.170400 | Huong |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.