workspace
stringclasses 4
values | channel
stringclasses 4
values | text
stringlengths 1
3.93k
| ts
stringlengths 26
26
| user
stringlengths 2
11
|
|---|---|---|---|---|
elmlang | general | In Java class designers are encouraged to provide `equals` and `hashCode` methods if they intend to use objects as Map keys | 2019-01-30T04:34:49.348700 | Lynne |
elmlang | general | It must have been done for a reason given the language's history | 2019-01-30T04:35:15.349100 | Lynne |
elmlang | general | That requires information to be present at runtime which isn't present at runtime, such as the names of the constructors. With `--optimize`, those are stripped so you can have smaller artifacts. Additionally, the compiler does unboxing in `--optimize` mode, meaning something aren't even in a "box" anymore. I.e. `type Foo = Foo String` will result in plain, bare strings being used, rather than objects that contain a string. | 2019-01-30T04:37:12.349700 | Huong |
elmlang | general | <@Lynne> hmm, in other languages, Enums are often stored as Ints. perhaps something similar could be used for custom types? | 2019-01-30T04:37:20.349900 | Nana |
elmlang | general | That is what Elm compiles them to when you don’t add values to them. If they carry values however, they become objects to allow for this. :slightly_smiling_face: | 2019-01-30T04:38:14.351100 | Timika |
elmlang | general | Debug.toString was just an example, it wouldn't have to be the actual name, just some kind of unique thing | 2019-01-30T04:39:45.352700 | Nana |
elmlang | general | So that would only work for _enum-like_ custom types. I can vaguely remember a discussion about this. | 2019-01-30T04:39:56.353100 | Timika |
elmlang | general | Yeah, since we are talking only about custom types converting to ints might be a way to go if only they did not have arguments :slightly_smiling_face: | 2019-01-30T04:39:57.353200 | Lynne |
elmlang | general | but if I had:
```
type Foo = Foo
type Bar a = Bar a
```
then `Bar Foo` could still probably be converted into a more clever thing than an object, maybe some mathematical operation combining the Int representations of `Foo` and `Bar` ? | 2019-01-30T04:42:15.355100 | Nana |
elmlang | general | probably way faster than big String comparisons, though could still get slow on large structures | 2019-01-30T04:43:31.356800 | Nana |
elmlang | general | I was referring to smth like `type Foo = Bar Int | Baz String | Quu Gender`, where the solution is not obvious, at least to me. | 2019-01-30T04:43:34.356900 | Timika |
elmlang | general | And in case of `Bar SomeLongRecord`? | 2019-01-30T04:43:41.357100 | Lynne |
elmlang | general | <@Lynne> yeah it could still get slow | 2019-01-30T04:44:14.357800 | Nana |
elmlang | general | Not even slow, how would you assign a number in this case? | 2019-01-30T04:44:33.358500 | Lynne |
elmlang | general | It is pretty simple when using enum-like types or types which only take enum-like types, but then such a restriction should appear in the language | 2019-01-30T04:45:11.359800 | Lynne |
elmlang | general | I think this is one of the problems that look easy to solve until you dive a bit deeper, then they become super hard, especially when you factor in elegance of design and ergonomics. | 2019-01-30T04:45:33.360400 | Timika |
elmlang | general | So pretty much extend `comparable` to enum-like types and we are good to go | 2019-01-30T04:45:39.360600 | Lynne |
elmlang | general | Still won't be super useful though :slightly_smiling_face: | 2019-01-30T04:45:58.361100 | Lynne |
elmlang | general | <@Lynne> though it'd probably be bad to have this feature for enum-likes only. That'd make other custom types a bit of a second-class citizen, which would encourage worse code | 2019-01-30T04:48:41.362300 | Nana |
elmlang | general | so yeah :pensive: | 2019-01-30T04:49:11.362700 | Nana |
elmlang | general | That's why I said it won't be super useful | 2019-01-30T04:49:16.362900 | Lynne |
elmlang | general | I use `Uuid` and would be happy having them as dict keys | 2019-01-30T04:49:34.363300 | Lynne |
elmlang | general | But `Uuid` is not an enum-like type | 2019-01-30T04:49:52.363700 | Lynne |
elmlang | general | And, generally speaking, it is not possible to come with an integer number for it, only some float (probably with double precision) | 2019-01-30T04:50:51.364600 | Lynne |
elmlang | general | (I mean an integer which would fit into `Int` type) | 2019-01-30T04:51:44.365000 | Lynne |
elmlang | general | <@Lynne> hmm, so for you, `AssocList` could actually be faster than `Dict String Stuff` ? :thinking_face: | 2019-01-30T04:55:22.366400 | Nana |
elmlang | general | because `==` doesn't do a full string comparison to check wether too Uuids are equal, right? | 2019-01-30T04:56:11.367100 | Nana |
elmlang | general | How do you mean it does not do full string comparison? | 2019-01-30T04:59:03.367600 | Lynne |
elmlang | general | or maybe it does? don't know how it works | 2019-01-30T04:59:41.368500 | Nana |
elmlang | general | If custom types are represented as JS objects then Kernel will do structural comparison, i.e. it will compare strings | 2019-01-30T04:59:42.368600 | Lynne |
elmlang | general | But I don't know how they are represented in 0.19. In 0.18 they were objects if I remember correctly | 2019-01-30T05:00:28.369300 | Lynne |
elmlang | general | but with immutable datastructures, aren't you supposed to be able to check equality by just checking the top node? | 2019-01-30T05:01:05.370000 | Nana |
elmlang | general | Could you elaborate on this? | 2019-01-30T05:01:25.370200 | Lynne |
elmlang | general | <@Nana> yes, you can do reference comparisons, but that's only helpful if these data structures are related in some way | 2019-01-30T05:02:18.371200 | Earlean |
elmlang | general | They still are in cases where you have a value with them. Otherwise `--optimize` will make them numbers. | 2019-01-30T05:02:23.371300 | Timika |
elmlang | general | eg. if one is a slightly modified version of the other | 2019-01-30T05:02:43.371700 | Earlean |
elmlang | general | Thanks, good to know :+1: | 2019-01-30T05:02:49.371800 | Lynne |
elmlang | general | Not sure I understood you <@Earlean> | 2019-01-30T05:04:38.374300 | Lynne |
elmlang | general | If you have `a = [1, 2]` and `b = [1, 2]`, they will be two different references but equal in value. | 2019-01-30T05:04:53.374600 | Bert |
elmlang | general | That's why there is structural comparison, right? | 2019-01-30T05:05:25.375000 | Lynne |
elmlang | general | `Basics.eq` is actually doing reference comparison as first step (pretty logical) before it dives deeper in the structure | 2019-01-30T05:05:42.375600 | Lynne |
elmlang | general | <@Lynne> I just remember Immutable.js presenting this as an advantage | 2019-01-30T05:05:56.376200 | Nana |
elmlang | general | I mean JS function which `Basics.eq` is backed by | 2019-01-30T05:05:57.376500 | Lynne |
elmlang | general | <@Bert> yeah that makes sense, so I guess you can quickly prove a positive, but not a negative | 2019-01-30T05:06:19.377400 | Nana |
elmlang | general | <@Nana> It is very possible I am just not sure I understood what "top node" is | 2019-01-30T05:06:29.377600 | Lynne |
elmlang | general | Right. Think of an immutable linked list. | 2019-01-30T05:06:59.378100 | Bert |
elmlang | general | Prepend an item and you have a longer linked list, but the "top node" is now different | 2019-01-30T05:07:56.379100 | Bert |
elmlang | general | If it is a graph node, then, yes, in case of immutable data structures simple reference comparison of the node says if you need to dive deeper to the children | 2019-01-30T05:08:26.379900 | Lynne |
elmlang | general | right, so I guess when you use `==` on two `Uuid`s, would it first check reference equality, but if that returns false, it'd to a full structural comparison? so when going through a list to find a match, you wouldn't get any benefit from it | 2019-01-30T05:12:15.381700 | Nana |
elmlang | general | That's exactly what `Basics.eq` is doing | 2019-01-30T05:12:38.382100 | Lynne |
elmlang | general | <https://github.com/elm/core/blob/22b2d7a235003a88af4e4af58c3d0edf43684bc9/src/Elm/Kernel/Utils.js#L35> | 2019-01-30T05:13:14.382400 | Lynne |
elmlang | general | thanks, this was an interesting discussion :slightly_smiling_face: | 2019-01-30T05:17:25.383100 | Nana |
elmlang | general | hello everyone!
is it possible to pattern match on a string in elm?
say I have a string that has a well defined structure and would like to extract the separate info from it? | 2019-01-30T05:57:37.384400 | Karrie |
elmlang | general | <@Karrie> ah, not like that, you'll have to parse it into a record or something first | 2019-01-30T05:59:12.385400 | Nana |
elmlang | general | or just use Regex, that might be the simplest | 2019-01-30T06:01:16.385700 | Nana |
elmlang | general | (by pattern match I assume you meant `case .. of`) | 2019-01-30T06:02:07.386100 | Nana |
elmlang | general | Sadly, pattern matching in Elm is not as powerful as in Elixir yet | 2019-01-30T06:04:24.386600 | Lynne |
elmlang | general | <@Nana> in elixir you can do `def my_func("say:" <> <<digit::bytes-size(1)>> <> ":" <> thing) do ... end` | 2019-01-30T06:17:02.388200 | Karrie |
elmlang | general | but I guess it’s not possible in elm yet
thanks <@Lynne> and <@Nana>! | 2019-01-30T06:17:30.388800 | Karrie |
elmlang | general | <@Karrie> You could write a parser using <https://package.elm-lang.org/packages/elm/parser/latest/> and pattern match on the result of `Parser.run` | 2019-01-30T06:21:57.389600 | Jin |
elmlang | general | <@Jin> amazing! how did I know about this already?! thank you very much!! | 2019-01-30T06:22:51.390600 | Karrie |
elmlang | general | is there a way to trigger the `scrollIntoView()`-method of a DOM element rendered by Elm _without_ using ports (because ports need a full update -> view circle with a model and stuff but I want to trigger the `scrollIntoView()` on a simple view-component without state)? | 2019-01-30T07:37:06.393100 | Daysi |
elmlang | general | help would be greatly appreciated | 2019-01-30T07:40:28.393600 | Daysi |
elmlang | general | That's not how Elm works, so not possible | 2019-01-30T07:41:00.393900 | Lynne |
elmlang | general | wow | 2019-01-30T07:41:19.394100 | Daysi |
elmlang | general | hmm I think you can use `Browser.Dom.setViewPort` ? | 2019-01-30T07:42:12.396000 | Nana |
elmlang | general | You don't need model and stuff (can simply use `()` as your model) but you have to send some command to trigger side effects | 2019-01-30T07:42:18.396200 | Lynne |
elmlang | general | <@Nana>, it is a `Task`, still not possible with just a view | 2019-01-30T07:43:11.397100 | Lynne |
elmlang | general | yeah not just with view, that'd be weird anyway :stuck_out_tongue: | 2019-01-30T07:43:50.398100 | Nana |
elmlang | general | the problem is I want to have a reusable state-less component, which handles the "onClick" independently from the client-code which uses it. Could I implement a "private" update-function on this component? | 2019-01-30T07:44:32.398500 | Daysi |
elmlang | general | e.g. look at the scrollable TabBar example from material design here <https://material-components.github.io/material-components-web-catalog/#/component/tabs> | 2019-01-30T07:45:06.399000 | Daysi |
elmlang | general | I am implementing this right now and need the tabs to `scrollIntoView()` on click | 2019-01-30T07:45:29.399800 | Daysi |
elmlang | general | but I don't want to delegate this to the client-code every time | 2019-01-30T07:45:45.400500 | Daysi |
elmlang | general | <@Daysi> yes, your view component can take a "message handler" as an argument, which lets you do stuff like that | 2019-01-30T07:46:33.401400 | Nana |
elmlang | general | how so? | 2019-01-30T07:47:05.402000 | Daysi |
elmlang | general | Shortly, there is no "private" update functions | 2019-01-30T07:47:12.402200 | Lynne |
elmlang | general | <@Lynne> no _exposed_ update function. You get what I mean... no? | 2019-01-30T07:47:56.403500 | Daysi |
elmlang | general | All messages you send from your "subcomponents" (in Elm world there is a long-running discussion if "component" is an appropriate term at all) have to go via top-level program (client code in your terms if I got you right) | 2019-01-30T07:48:24.404100 | Lynne |
elmlang | general | <@Daysi> I think I get, but, again, that's not how Elm works :slightly_smiling_face: | 2019-01-30T07:48:58.404900 | Lynne |
elmlang | general | but it wouldn't make sense for the top-level program to handle stuff which belongs into the responsibilities of the view layer | 2019-01-30T07:49:17.405400 | Daysi |
elmlang | general | Who would be triggering this private function if it was there? | 2019-01-30T07:49:25.405800 | Lynne |
elmlang | general | the user | 2019-01-30T07:49:37.406100 | Daysi |
elmlang | general | Programmaticaly | 2019-01-30T07:49:48.406500 | Lynne |
elmlang | general | Given the Elm runtime and stuff | 2019-01-30T07:49:54.406800 | Lynne |
elmlang | general | Well, the only thing top-level program needs to do in this case is to delegate certain messages to another module's `update` function | 2019-01-30T07:50:19.407800 | Lynne |
elmlang | general | So you still have separation of concerns, it is just the message flow which always starts from top | 2019-01-30T07:50:53.408600 | Lynne |
elmlang | general | but the top-level program shouldn't be aware that there are any messages which don't concern it, i.e. `scrollIntoView()` | 2019-01-30T07:51:39.409400 | Daysi |
elmlang | general | it should only care about business-related logic, not view-related logic | 2019-01-30T07:52:40.410800 | Daysi |
elmlang | general | Well, it does not have be aware of exact messages (is it `scrollIntoView` or `scrollOutOfView` should not concern it), but it has to be aware of the very presence of messages themselves | 2019-01-30T07:52:51.411100 | Lynne |
elmlang | general | I hear OO in what you are saying :slightly_smiling_face: | 2019-01-30T07:53:12.411400 | Lynne |
elmlang | general | <@Daysi> here's the most basic version of such a "component" <https://ellie-app.com/4BfGmGPKKdva1> | 2019-01-30T07:56:30.412500 | Nana |
elmlang | general | in the example it only returns it's new state, however it could also return other things like `(model, Cmd.msg)` which is how you can use Tasks like setting the viewport | 2019-01-30T07:58:26.414200 | Nana |
elmlang | general | <@Lynne> *triggered* | 2019-01-30T08:00:47.414500 | Daysi |
elmlang | general | :smile: | 2019-01-30T08:00:48.414700 | Daysi |
elmlang | general | no seriously, I don't know why you hear OO | 2019-01-30T08:01:19.415400 | Daysi |
elmlang | general | abstraction and encapsultion of reponsibility is a universal concept not related to OO or FP | 2019-01-30T08:02:01.416800 | Daysi |
elmlang | general | <@Nana> thanks | 2019-01-30T08:02:16.417500 | Daysi |
elmlang | general | "business-related logic", "view-related logic" are terms I've never heard since I started working with Elm | 2019-01-30T08:02:41.418300 | Lynne |
elmlang | general | People just tend to call things differently here | 2019-01-30T08:02:50.418800 | Lynne |
elmlang | general | with this type of "mapping" you can do encapsulation in Elm, and it lets you create "components" that can have private state and do effects. it's just that they always have to "go through the parent" | 2019-01-30T08:03:50.420700 | Nana |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.