Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Jan 24 05:51

    mergify[bot] on master

    Update scalatest to 3.2.11 (#57… (compare)

  • Jan 24 05:51
    mergify[bot] closed #572
  • Jan 24 05:49

    mergify[bot] on master

    Update scalatest to 3.2.11 (#35) (compare)

  • Jan 24 05:49
    mergify[bot] closed #35
  • Jan 24 05:44
    mergify[bot] labeled #35
  • Jan 24 05:44
    mergify[bot] labeled #572
  • Jan 24 05:44
    scala-steward opened #35
  • Jan 24 05:44
    scala-steward opened #572
  • Jan 21 14:58
    busti commented #571
  • Jan 21 09:37
    fdietze commented #571
  • Jan 21 09:31
    fdietze commented #570
  • Jan 20 23:08
    busti commented #571
  • Jan 20 22:16
    busti commented #571
  • Jan 20 16:41
    fdietze commented #571
  • Jan 20 12:41

    cornerman on master

    add inner html (compare)

  • Jan 20 01:11
    busti edited #571
  • Jan 20 01:11
    busti commented #571
  • Jan 20 00:38
    busti edited #571
  • Jan 20 00:37
    busti edited #571
  • Jan 20 00:37
    busti edited #571
bblfish
@bblfish:matrix.org
[m]
I am trying to upgrade the demo example to Scala 3. I tried this https://github.com/bblfish/SolidCtrlApp but without got the following error
[success] Total time: 1 s, completed 7 Nov 2021, 21:43:34
[error] [webpack-cli] Unable to load '@webpack-cli/serve' command
[error] [webpack-cli] TypeError: options.forEach is not a function
[error]     at WebpackCLI.makeCommand (/Volumes/Dev/Programming/Scala3/solidapp/target/scala-3.1.0/scalajs-bundler/main/node_modules/webpack-cli/lib/webpack-cli.js:108:21)
[error]     at ServeCommand.apply (/Volumes/Dev/Programming/Scala3/solidapp/target/scala-3.1.0/scalajs-bundler/main/node_modules/@webpack-cli/serve/lib/index.js:41:19)
bblfish
@bblfish:matrix.org
[m]
I guess it is not a problem with scala 3.x because going down to 2.13 I get the same problem. Perhaps the version of webpack...
johannes karoff
@cornerman
@bblfish:matrix.org Hmm, I also do not get it to work with webpack5 and the scalajs-bundler rc1. Somehow there is some incompatible version of webpack-cli and webpack-cli/serve, with that version configuration.
Do you need webpack 5 for a specific reason?
bblfish
@bblfish:matrix.org
[m]
No, I was just trying to get the latest.
I found this https://youtrack.jetbrains.com/issue/KT-49124
But I think I am using 4.9.1
I am a bit slow because I don't really know the JS ecosystem tools
johannes karoff
@cornerman
If I force the deps webpack-cli -> 4.9.1 and @webpack-cli/serve -> 1.6.0, I get another runtime error:
[error] [webpack-cli] Invalid options object. Dev Server has been initialized using an options object that does not match the API schema.
[error]  - options has an unknown property 'inline'. These properties are valid:
[error]    object { allowedHosts?, bonjour?, client?, compress?, devMiddleware?, headers?, historyApiFallback?, host?, hot?, http2?, https?, ipc?, liveReload?, magicHtml?, onAfterSetupMiddleware?, onBeforeSetupMiddleware?, onListening?, open?, port?, proxy?, server?, setupExitSignals?, static?, watchFiles?, webSocketServer? }
Something is incompatible there. Hmm.
I am still using webpack 4 in my projects without a problem. But might be worth asking in the scalajs-bundler channels.
bblfish
@bblfish:matrix.org
[m]
do you have a public git repo with a config that is working? (idally for scala3)
I don't really need the latest right now, I just need to get going :-)
bblfish
@bblfish:matrix.org
[m]
I'll got back to the seed.g8
johannes karoff
@cornerman

@bblfish:matrix.org Regarding the question about subscriptions and comparison to laminar.

Laminar uses their own streaming library called airstream, which has a really nice concept about reactive state management. There each operation that might lead to a subscription needs an implicit Owner. This owner then handles the subscription - i.e. is responsible for killing it. Using this in laminar, the owner is the enclosing html element. Airstream is in some ways similar to scala-rx, both fixing frp-glitches and using an Owner concept.

In outwatch we try to abstract over different streaming libraries, and not all have this owner concept. The subscription methods there, do not require an Owner, but instead return a Cancelable - a token to cancel the subscription. As long as it is handled, everything is fine.
Now as long as you do not call subscribe on Observables yourself (in colibri or monix), but just use them in outwatch components, everything is fine and subscriptions will be automatically cleaned up.
The only way to leak here, is to call subscribe yourself and actively ignore the returned subscription, like val _ = observable.subscribe(observer). You can still let outwatch handle the subscription manually with managed(SyncIO(observable.subscribe(observer)) inside a component - which imho is rarely needed. Normally, you just use an observable in your component and get everything handled automatically.

Also, if you use scala-rx directly - even with outwatch - the mentioned danger does not apply. Additionally, we should be able to support airstream with outwatch as well.
The seed should work. I have not used scala-3 with outwatch yet, so please let me know of any blockers you might find!
bblfish
@bblfish:matrix.org
[m]

Things start working when I move from

addSbtPlugin("ch.epfl.scala" % "sbt-scalajs-bundler" % "0.21.0-RC1")

down to

addSbtPlugin("ch.epfl.scala" % "sbt-scalajs-bundler"  % "0.20.0")

in the plugins

Indeed the little app now works.
bblfish
@bblfish:matrix.org
[m]
It even works with scala 3.1.0 then . I pushed those changes to the https://github.com/bblfish/SolidCtrlApp
johannes karoff
@cornerman
Awesome!
We should start cross-compiling for scala 3 i guess :)
1 reply
bblfish
@bblfish:matrix.org
[m]
I found a nice intro to fs2 and cats effects for UI development. https://twitter.com/bblfish/status/1457992937804124166
That one writes directly to the dom interestingly enough, but is otherwise very FP. Is there any larger project one can look at with a lot of IO that uses OutWatch?
bblfish
@bblfish:matrix.org
[m]
Actually I went off to study cats-effect and co, but I am not sure if you work with cats-effect, and so http4s.
At least I could not find any references in the build file.
bblfish
@bblfish:matrix.org
[m]
(well it says you use cats-effect on the front page, but I guess it must be imported into the project in some indirect way)
bblfish
@bblfish:matrix.org
[m]
Ah cool the sbt dependencyTree tells me it is colibri that colibri that imports it.... (nice tool)
bblfish
@bblfish:matrix.org
[m]
ok, so I guess the latest http4s-dom uses cats-effects 3.x whereas colibri is using 2.6.1
moritz bust
@busti

@bblfish:matrix.org afaik technically projects that directly use a library should define it in their sbt file rather than relying on it through other dependencies. We should probably do that.

http4s-dom looks nice. I was wondering when they would ship their http client to js.

bblfish
@bblfish:matrix.org
[m]
they just release http4s-dom 1.0.0-M29 btw.
moritz bust
@busti
Studying cat's effect is a nice idea in general. The typelevel style in general is used in a big portion of scala libs and it's good to know what you're looking at when encountering libraries or code using the style.
I personally only use a single websocket connection in my project, but if I ever need to talk to a rest api it's probably a good choice
bblfish
@bblfish:matrix.org
[m]
For The Solid Project we write apps for the web thought of as one giant hyper-data RESTful system. Clients follow links from one server (Pod) to another to build up UI elements. So even though websockets are also used, we really need streaming http libraries.
bblfish
@bblfish:matrix.org
[m]
I have written my server in Akka but I think having akka on the client would be a bit heavy on the client. So that means I need to learn another set of libraries on this. Cats-effect seemed indeed like a good thing to try out, and as you say it is widely used...
elyphas
@elyphas
I have some kind of problem when I use onInput the focus on element input is loss more or less, because the element is showing is focused but the cursor is loss.
this happens on chrome but not on firefox.
I was searching for some solutions on google but I have some Observer connecting on this element so I don't quite get the solution.
Any idea please. :-)
this my code:
  val onChangingValueTxt =
        SyncIO(
          onInput
             .value
             .transformLifted( (e: Observable[String]) =>
               e
               .distinctUntilChanged
               .withLatestFrom(hdlFieldActive){ case (txt: String, (field: String, currentValue: JsonValue)) =>
                    val currentValue2: JsonValue = Try {
                            wrapValue(txt, currentValue)
                    } match {
                            case Success(value) => value
                            case Failure(exception) => JsonNull
                    }

                  (field, currentValue2)
                }
             ) --> hdlChangingValue
        )
elyphas
@elyphas
making a filter( x => x.length > 5) loses the cursor after 5 keys
elyphas
@elyphas
I found the problem all of that is inside the Try block:
    val mountCellFocusSelectContent = Observer.create[(dom.Element, String, String)] { case (elem, inputType, idTxt) =>
        if (idTxt == elem.id) {
                val txt = if ( inputType == "input" ) elem.asInstanceOf[html.Input] else elem.asInstanceOf[html.TextArea]
                Try {
                    val range = dom.document.createRange
                    val sel = scalajs.dom.window.getSelection
                    sel.removeAllRanges
                    range.selectNodeContents(txt)
                    sel.addRange(range)
                } match {
                    case Success(value) =>
                        txt.focus
                    case Failure(exception) => alert("Hubo un error al convertir el valor del componente: line:102 from Grid3 " + exception.getMessage())
                }
        }
    }
the solution seems is this:
$('#Search').focus(function() {
    setTimeout((function(el) {
        var strLength = el.value.length;
        return function() {
            if(el.setSelectionRange !== undefined) {
                el.setSelectionRange(strLength, strLength);
            } else {
                $(el).val(el.value);
            }
    }}(this)), 0);
});
What would be the correct translation to Outwatch ?
the comment about the problem is this:
"It seems like the focus event is fired before the cursor is placed when you focus an input, a hacky workaround would be to use a setTimeout like so:"
elyphas
@elyphas

well I did this:

    val mountCellFocusSelectContent = Observer.create[(dom.Element, String, String)] { case (elem, inputType, idTxt) =>
        if (idTxt == elem.id) {
                val txt = if ( inputType == "input" ) elem.asInstanceOf[html.Input] else elem.asInstanceOf[html.TextArea]
                Try {
                    if (window.navigator.userAgent.indexOf("Chrome") == -1 || window.navigator.userAgent.indexOf("Safari") == -1) {
                        val range = dom.document.createRange
                        val sel = scalajs.dom.window.getSelection
                        sel.removeAllRanges
                        range.selectNodeContents(txt)
                        sel.addRange(range)
                    }
                } match {
                    case Success(value) =>
                        txt.focus
                    case Failure(exception) => alert("Hubo un error al convertir el valor del componente: line:102 from Grid3 " + exception.getMessage())
                }
        }
    }

althought I don't know if is right. :-)

johannes karoff
@cornerman
We finally have a real maven release again. Welcome version 1.0.0-RC3. Thank you to @FloWi for setting this up in our github actions!
moritz bust
@busti
@cornerman Quick question, is there a reason why are the map, filter, collect and mapFilter methods are implemented using transformSinkWithExec and contraMap, contraFilter and so on, while all the others modify the source?
moritz bust
@busti
Also, why do some methods use the Observable.method(source)(args) syntax, while others use the source.method(args) syntax? Feels like some verbosity could be avoided by switching to the latter.
johannes karoff
@cornerman
@busti That is a great question! Actually it is because dealing with an Observable means, we need to handle a subscriptions: adding callbacks onMount and onDismount. Operating on an Observer, is great because you can just push things into an Observer without state. So, it is kind of a performance optimization to prefer Observer operations.
We do not have syntax extensions for Source, Sink generally, because I felt, it would be too intrusive, when just working with Monix or scalarx or airstream natively. Maybe, just hide them under a syntax._-import? It is nicer too write. And that is an advantage of the Sink, Source archtiecture.
Ah, or a we just working on Observable/Observer there?
moritz bust
@busti
Yeah, coding against observer is probably more performant in many cases, I never thought about that, but after familiarizing myself with colibri it makes sense.
EmitterBuilder kinda is like a wrapper in many cases, which makes it difficult to refactor in many cases, but I like the API it produces a lot.
moritz bust
@busti
I agree, syntax extensions for Source and Sink would feel wrong because usually algebraic methods are members these days. I hope that's going to change with scala 3 and the extension keyword. No more .map on everything. Just a single global extension method for all functor instances.