https://kotlinlang.org logo
Join Slack
Powered by
Give me a little reminder please. If I have a `Dic...
# javascript
v
Give me a little reminder please. If I have a 
Dict
subinterface that I need to give as argument, how do I properly construct it? In this case it is the 
external interface OutgoingHttpHeaders : Dict<dynamic /* Number | String | Array<String> */>
from 
kotlinx-nodejs
t
For latest OutgoingHttpHeaders you can use 
Record
factory function
v
Ah, so 
kotlin-node
is kind of successor of 
koltinx-nodejs
I guess? Well keep in mind when I go to newer Node. I guess the 
kotlin-node
version corresponds to the Node.js version? Because currently I'm on Node.js 12 and only have the choice between 12 and 16, as it is about a GitHub Action. For now I solved it using
Copy code
fun Map<String, Any>.asOutgoingHttpHeaders() = asDynamic().unsafeCast<OutgoingHttpHeaders>()
t
Ah, so 
kotlin-node
is kind of successor of 
koltinx-nodejs
I guess?
In some way
I guess the 
kotlin-node
version corresponds to the Node.js version?
@types/node
version used
Because currently I’m on Node.js 12 and only have the choice between 12 and 16, as it is about a GitHub Action.
16 - current LTS 🙂
For now I solved it using
Copy code
fun Map<String, Any>.asOutgoingHttpHeaders() = asDynamic().unsafeCast<OutgoingHttpHeaders>()
Map
!= 
Record
(JSO)
v
16 - current LTS 🙂
When I wrote the action only 12 was available. 🤷 
Map
!= 
Record
(JSO)
Well, it works as expected according to my tests. :-)
t
Map
- Kotlin Map?
v
Yes
t
image.png
v
Well, I derived it from the previous
Copy code
fun Map<String, Any>.asIHeaders() = asDynamic().unsafeCast<IHeaders>()
that I created with the help of you afair 🙂
Ah, no, it was not you but araqnid (https://kotlinlang.slack.com/archives/C0B8L3U69/p1597445606499200). And of course I should also add the 
asDynamic
😄
Copy code
fun Map<String, Any>.asOutgoingHttpHeaders() = asDynamic().unsafeCast<OutgoingHttpHeaders>()

fun <V> Map<String, V>.asDynamic() = jsObject<dynamic> {
    for ((name, value) in this@asDynamic) {
        this[name] = value
    }
}
t
when Node.js 18 enters long-term support (LTS) later this month
https://nodejs.org/en/blog/announcements/v19-release-announce/
In 
kotlin-wrappers
we provide many additional types for strict JS interop. As result - you can avoid 
asDynamic()
and 
unsafeCast()
usages (like in case above 😜)
👌 1
v
Hm, GitHub stops supporting node 12, but does not yet support node 18, so just node 16. But kotlin-node just provides node 17 and node 18. So what should I use for node 16? Still continue using koltinx-nodejs? And btw. will kotlin-node ever leave pre-state? 😄
t
But 
kotlin-node
just provides node 17 and node 18.
We support Node 16 as minimum version
So what should I use for node 16?
In case of wrappers 18 = 16+ You will see some new API, but most of them marked with 
@since
tag
And btw. will kotlin-node ever leave pre-state? 😄
Yes, when we will find better versioning scheme for our use case (NPM version + wrappers build or version). Do you have ideas? 🙂
v
We support Node 16 as minimum version
Oh really? But I only have seen 17 and 18 versions on Maven Central
In case of wrappers 18 = 16+
You will see some new API, but most of them marked with 
@since
tag
Oh, I see, great. I thought you need the version matching your node version. Then I'll probably switch of course. For now it actually worked without any code change and just changing node version to 16 🙂
Do you have ideas?
Probably none you didn't have yet. <npm version>-<your release as increasing number> or something like that. But maybe I just cannot think, it's 4 AM. Just wanted to quickly get rid of the issue ticket. 😄
t
@Vampire FYI - we have 
suspend
adapters for converted promise API in 
kotlin-node
. For example for 
fs
module API
v
Mind elaborating what you mean?
t
v
So like 
mkdirSync
in 
kotlinx-nodejs
, isn't it?
t
In 
mkdirSync
- synchronous API Inside 
mkdir
we use asynchronous API (
mkdirAsync
)
v
Well, then like 
mkdir
, isn't it? Or what do I miss? 🙂
Ah, now I see it. That returns a 
Promise
and you abstract that away 👌
😜 1
What are the 
...-compat
versions?
t
AFAIK It’s the same build, but with
Copy code
kotlin.mpp.enableCompatibilityMetadataVariant=true
option.
v
I see, just wondered because not every version has that variant, thx
Back, kind of, to the original topic. I'm currently finally trying to migrate from 
kotlin-extensions
and 
kotlinx-nodejs
to 
kotlin-js
and 
kotlin-node
, so I might come back with some more question if you don't mind. 🙂
The 
OutgoingHttpHeaders
I was now able to do using
Copy code
additionalHeaders = Record<String, String> {
    set("Content-Type", "application/x-www-form-urlencoded")
}
That's the intended way and what you meant, correct?
👌 1
t
Factory, which we discuss right now:
Copy code
additionalHeaders = recordOf(
    "Content-Type" to "application/x-www-form-urlencoded",
)
v
Argh, I should read properly. Seen it, liked it, changed my code, wondered I don't find it, then seen that you said this is an idea for the future. 😄
Ok, next is, I call 
exec(...)
from 
@actions/exec
. That needs an 
ExecOptions
so I now do 
options = jso { ... }
instead of 
options = jsObject { ... }
, right?
👌 1
Now the 
ExecOptions
need (or I need to set it) 
env
of type 
tsstdlib.T$2
which comes from 
lib.es5.kt
and is defined as
Copy code
external interface `T$2` {
    @nativeGetter
    operator fun get(key: String): String?
    @nativeSetter
    operator fun set(key: String, value: String)
}
What is the correct way here? In approximation of my old code mentioned above, I'd guess
Copy code
env = Record<String, String> {
    set("DEBIAN_FRONTEND", "noninteractive")
    set("WSLENV", "DEBIAN_FRONTEND/u")
}.unsafeCast<`T$2`>()
?
t
T$2
is effectively 
Record
- could you replace it?
v
It's probably coming from Dukat, but I have a "fix the result up" procedure, so I guess I could.
t
ReadonlyRecord
looks like better replacement 😉
recordOf
will return it
v
So I would in the 
ExecOptions
replace
Copy code
var env: `T$2`?
    get() = definedExternally
    set(value) = definedExternally
by
Copy code
var env: js.core.ReadonlyRecord<String, String>?
right?
t
Yes
Analog from JVM world 
Map
-> 
ReadonlyRecord
MutableMap
-> 
Record
v
Great, now I can even do 
env = Record {
without type arguments 🙂
t
recordOf
is ready
Will be available in 
pre.488
❤️ 1
Great, now I can even do 
env = Record {
without type arguments 🙂
recordOf
v
Sure, when 488 is available 🙂
Hm, where is 
node.fs.readdir
?
And is there some way to make Dukat use the new wrappers classes? Now various of the generated classes use 
http.IncomingMessage
, 
http.OutgoingHttpHeaders
, 
url.URL
, and so on which are all in other packages now. Or do I need to add the import-fixing to my post-processing step?
t
Which NPM libraries do you need?
v
Copy code
implementation(npm("@actions/cache"))
    implementation(npm("@actions/core"))
    implementation(npm("@actions/exec"))
    implementation(npm("@actions/http-client"))
    implementation(npm("@actions/io"))
    implementation(npm("@actions/tool-cache"))
    implementation(npm("@types/semver"))
    implementation(npm("semver", generateExternals = false))
    implementation(npm("null-writable")) 

    implementation(npm("@vercel/ncc", generateExternals = false))
t
Hm, where is 
node.fs.readdir
?
As quick fix - PR in 
kotlin-node
(with manual declarations) + issue
v
So it is missing? 😞
t
I did’t see it right now
v
I would even take 
readdirSync
or 
readdirAsync
, but it seems all are missing. 
exists
is also missing, but at least 
existsSync
is available.
t
Issue will be helpful
v
Of course, just thought I might miss something and had you at hand 🙂 Thanks for your great help always. 🙂
😀 1
And is there some way to make Dukat use the new wrappers classes?
Now various of the generated classes use 
http.IncomingMessage
, 
http.OutgoingHttpHeaders
, 
url.URL
, and so on which are all in other packages now.
Or do I need to add the import-fixing to my post-processing step?
Which NPM libraries do you need?
``` implementation(npm("@actions/cache"))
implementation(npm("@actions/core"))
implementation(npm("@actions/exec"))
implementation(npm("@actions/http-client"))
implementation(npm("@actions/io"))
implementation(npm("@actions/tool-cache"))
implementation(npm("@types/semver"))
implementation(npm("semver", generateExternals = false))
implementation(npm("null-writable"))
implementation(npm("@vercel/ncc", generateExternals = false))```
So, ...? 🙂
t
And is there some way to make Dukat use the new wrappers classes?
AFAIK Dukat is freezed right now 😞
@actions/...
libraries look like good candidates for 
kotlin-wrappers
Issue will be fine start point, especially if you are ready to test results 🙂
https://github.com/JetBrains/kotlin-wrappers/issues/1889
Could you list all missed methods, which you found?
v
AFAIK Dukat is freezed right now
So no No problem, I'll simply do it in my post-processing. Actually, it should unfreeze now. At least it was always said that it is freezed until IR is stable which it is now with 1.8. 
@actions/...
libraries look like good candidates for 
kotlin-wrappers
Issue will be fine start point, especially if you are ready to test results 🙂
Nice. :-) Here you have it: https://github.com/JetBrains/kotlin-wrappers/issues/1891 Of course I'm willing to test, as I use almost all of them anyway already and have extensive tests for my action.
> https://github.com/JetBrains/kotlin-wrappers/issues/1889
Could you list all missed methods, which you found?
I did. 
readdirSync
was the only one I used for which I did not find a replacement.
Ah, actually found another hole. 
writeFile
previously allowed to set the mode for the file, but now only the encoding can be set 😞 https://github.com/JetBrains/kotlin-wrappers/issues/1893
t
pre.488
released! 
recordOf
is available 😉
❤️ 1
v
Not yet on MavenCentral?
t
15+ minutes usually required for sync
If MavenCentral is fine :)
👌 1
v
Hm, at least where it takes an 
OutgoingHttpHeaders
I cannot use the 
recordOf
factory as it complains because 
OutgoingHttpHeaders
is an alias for 
Dict<Any /* OutgoingHttpHeader */>
which is an alias for 
Record<String, Any /* OutgoingHttpHeader */>
so it does not accept the 
ReadonlyRecord
🤔 1
t
Fix is ready!
❤️ 1
Requested fixes for Node also expected in 
pre.489
👌 1
Released :)
v
Nice, thanks. Though I question the naming of the new interface. 🙂 properties are 
mode
of type 
Mode
and 
flag
of type 
OpenMode
, but the interface containing them is called 
FlagAndOpenMode
which both just refer to the second property. 🙂
t
Good news - we have single source for all plaforms! 😉
v
Ah, damn, maybe one more. I still have one 
...Sync
method, 
existsSync
which is the only one you have in the wrapper. Should there be an 
existsAsync
and therefor a 
suspend exists
?
👍 1
t
Ah, damn, maybe one more.
Don’t stop! 😉
Fallbacks for 
exists
described in header
v
So if I got it right, • async 
exists
returning a promise does not exist as it is not in 
promises.d.ts
• 
existsSync
does exist and is also supported • 
exists
in an async way only exists with callbacks and these are deprecated and should be replaced by 
stat
or 
access
Correct?
So from what I understood from those docs, I should use 
fs.access
instead. But 
node.fs.access
in the wrapper returns 
Unit
😕
t
Success = exists 🙂
stat
looks like best async variant
v
But the docs of 
stat
tell to use 
access
for pure file-exists check 😄
So I guess I need a
Copy code
suspend fun exists(path: PathLike) = runCatching {
    access(path)
    true
}.getOrDefault(false)
t
Copy code
suspend fun exists(path: PathLike): Boolean =
    accessAsync(path)
        .then { true }
        .catch { false } 
        .await()
v
Is that better than my version and if so, why, or is it just an alternative? :-)
Ah, well, yours does not need exception handling, so better
No exception handling for flow-control if avoidable 🙂
Ok, thanks again for you help, all usages updated. Now let's see whether the tests still run successfully. 🙂
Oh, damn, it does not even run anymore. :'''-(
Copy code
Error: No such built-in module: node:test
    at new NodeError (node:internal/errors:372:5)
    at Function.Module._load (node:internal/modules/cjs/loader:785:13)
    at Module.require (node:internal/modules/cjs/loader:1005:19)
    at require (node:internal/modules/cjs/helpers:102:18)
    at Object.2008 (D:\a\setup-wsl\setup-wsl\build\webpack:\setup-wsl-ncc-packer\external node-commonjs "node:test":1:1)
    at __nccwpck_require__ (D:\a\setup-wsl\setup-wsl\build\webpack:\setup-wsl-ncc-packer\webpack\bootstrap:21:1)
    at Object.6166 (D:\a\setup-wsl\setup-wsl\build\node_modules\kotlin-node\kotlin-node.js:2070:1)
    at __nccwpck_require__ (D:\a\setup-wsl\setup-wsl\build\webpack:\setup-wsl-ncc-packer\webpack\bootstrap:21:1)
    at Object.4624 (D:\a\setup-wsl\setup-wsl\build\webpack:\setup-wsl\kotlin\setup-wsl.js:3364:1)
    at __nccwpck_require__ (D:\a\setup-wsl\setup-wsl\build\webpack:\setup-wsl-ncc-packer\webpack\bootstrap:21:1)
Any idea where that might come from?
t
Do you use Legacy compiler?
AFAIK 
node:test
is from Node 18
v
Yes, that project is still on Kotlin 1.4.
How would node:test come in then? I use Node 16 to build and also to run
And the Node version also did not change
It's also not the 
ncc
version, if I update that separately first, the project still runs as expected.
But I then replace `kotlin-extensions`/`kotlinx-nodejs` by `kotlin-js`/`kotlin-node` it fails with that error
t
It’s also not the 
ncc
version
What does it mean?
Yes, that project is still on Kotlin 1.4.
Magic 🙂
v
I wanted to switch to `kotlin-js`/`kotlin-node` only as a first step. But with those changes 
ncc
failed to run. So I updated 
ncc
along to see whether it helps and then the 
ncc
run worked. But now the execution of the result fails. I now reverted all changes and only updated 
ncc
first to see whether it fails or not, but with that it still works as expected.
t
What is 
ncc
?
v
Input is the result of Kotlin/JS, output is a self-contained runnable JS, that then runs as GitHub action
t
Do you use Webpack?
v
iirc ncc uses webpack internally
t
I see 2 options: 1. Use IR with Kotlin 1.8 to avoid redundant imports (
node:test
in your case) // recommended 2. Configure webpack fallback for 
node:test
if it’s possible
Webpack fallback
3*. Remove/replace 
node:test
import in Gradle task 😉
v
With 3. you mean a post-processing of the ncc result? I wonder how it actually lands in the result just by switching from `kotlin-extensions`/`kotlinx-nodejs` to `kotlin-js`/`kotlin-node`
t
With 3. you mean a post-processing of the ncc result?
I mean post-processing of Kotlin compilation result It’s what you can find in 
build/js/packages/%your-package%
v
find build/js/packages/setup-wsl -type f -exec grep node:test {} +
does not find any result though
Ok, maybe I should take the recommended approach and first stash all my changes and update Kotlin to 1.8. But I guess I will go on bugging you with questions then 😄
Is there docs what the tasks do? e. g. now with 1.8 and IR, there is 
compileProductionExecutableKotlinJs
and 
compileDevelopmentExecutableKotlinJs
. But I don't find info about them in the docs. Or 
productionExecutableCompileSync
and 
devleopmentExecutableCompileSync
.
Or previously 
run
did run 
nodeRun
and the docs still say it does. But now 
run
runs 
nodeRun
, 
nodeProductionRun
, and 
nodeDevelopmentRun
.
Ah, I think I got it. 
compileKotlinJs
- 
*.kt
=> IR 
compileDevelopmentExecutableKotlinJs
- IR => JS without DCE 
compileProductionExecutableKotlinJs
- IR => JS with DCE 
developmentExecutableCompileSync
- JS without DCE + resources => to other folder 
productionExecutableCompileSync
- JS with DCE + resources => to other folder 
nodeRun
- Runs the result of 
developmentExecutableCompileSync
nodeDevelopmentRun
- Also runs the result of 
developmentExecutableCompileSync
nodeProductionRun
- Runs the result of 
productionExecutableCompileSync
While I think it is very suboptimally solved that the 
...CompileSync
tasks are 
Copy
tasks, not 
Sync
tasks, and that they also have the same output directory. Will create an issue about that.
👍 1
Yay, not many questions necessary actually. I successfully moved to Kotlin 1.8 and due to incompatibility Gradle 7.6 and thereby the long overdue switch to included builds instead of 
buildSrc
I long postponed due to other stuff and thereby cleaning up some really ugly stuff did in the past in that build, rewriting half of the build. 😄 And as a side-effect fixed invalid tests that were successful do the wrong reason. :-D Now the end result is immediately just 6.3 MiB instead of 14 MiB. :-)
😉 1
t
FYI - If you use IR - mangling will be applied by default
v
Can you elaborate?
t
v
Yeah, I've read that before. I'm just not sure whether you wanted to imply something. :-)
Perfect, now my stashed changes also work flawlessly 🙂
Hm, not exactly flawlessly. 😞 In production mode it works fine. But if I do 
nodeRun
or 
nodeDevelopmentRun
, it fails with
Copy code
D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:25473
}(module.exports, require('semver'), require('node:fs/promises'), require('null-writable'), require('node:path'), require('node:os'), require('node:process')));
 ^
ReferenceError: AbstractCoroutineContextElement is not defined
    at D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:454:73
    at Object.<anonymous> (D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:25473:2)
    at Module._compile (node:internal/modules/cjs/loader:1152:14)
    at Object.Module._extensions..js (node:internal/modules/cjs/loader:1209:10)
    at Module.load (node:internal/modules/cjs/loader:1033:32)
    at Function.Module._load (node:internal/modules/cjs/loader:868:12)
    at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:81:12)
    at node:internal/main/run_main_module:22:47
🙈 Any idea where this might come from and / or how to fix it?
t
Outdated 
coroutines
version?
v
org.jetbrains.kotlinx:kotlinx-coroutines-core:1.6.4
And in production mode it works fine
t
Which Kotlin version do you use?
v
1.8.10
Same with 1.8.0, just updated to 1.8.10 as it came out
t
Do you use 
whole-program
granularity? 🧐
v
As you advised, yes. But iirc I also tried without and it didn't change
t
It means, that you can simply check result file. 
AbstractCoroutineContextElement
really missed in development build?
v
Ah, yeah, I did check that too:
Copy code
$ grep AbstractCoroutineContextElement build/compileSync/js/main/developmentExecutable/kotlin/setup-wsl.js
  setMetadataFor(CoroutineDispatcher, 'CoroutineDispatcher', classMeta, AbstractCoroutineContextElement, [AbstractCoroutineContextElement, ContinuationInterceptor], undefined, undefined, []);
  setMetadataFor(YieldContext, 'YieldContext', classMeta, AbstractCoroutineContextElement, undefined, undefined, undefined, []);
    AbstractCoroutineContextElement.call(this, Key_getInstance());
    AbstractCoroutineContextElement.call(this, Key_getInstance_3());
Feel free to also have a look if you like, I pushed the current state to branch 
foo
of https://github.com/Vampire/setup-wsl
In production mode it is
Copy code
$ grep AbstractCoroutineContextElement build/compileSync/js/main/productionExecutable/kotlin/setup-wsl.js
  setMetadataFor(AbstractCoroutineContextElement, 'AbstractCoroutineContextElement', classMeta, undefined, [Element], undefined, undefined, []);
  setMetadataFor(CoroutineDispatcher, 'CoroutineDispatcher', classMeta, AbstractCoroutineContextElement, [AbstractCoroutineContextElement, ContinuationInterceptor], undefined, undefined, []);
  function AbstractCoroutineContextElement(key) {
  AbstractCoroutineContextElement.prototype.w = function () {
    AbstractCoroutineContextElement.call(this, Key_getInstance());
  AbstractCoroutineContextElement.prototype.m2 = get;
  AbstractCoroutineContextElement.prototype.s2 = fold;
  AbstractCoroutineContextElement.prototype.r2 = minusKey;
  AbstractCoroutineContextElement.prototype.t2 = plus;
t
Call in 
main
:
Copy code
(AbstractCoroutineContextElement::class.js)
as workaround works for you?
I see, that you use different versions for Kotlin (
1.8.10
) and Kotlinx Serialization plugin (
1.7.10
) . Why?
v
The serialization plugin is used in the build, so I chose the one matching the Kotlin version embedded in Gradle
In the production code no serialization is used
I just use it to read the 
action.yml
file and to write a file in the sub-builds that I read in the root build
The suggested work-around does help in so far, that it then complains about the next class
Copy code
D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:31451
}(module.exports, require('node:fs/promises'), require('@actions/http-client'), require('@actions/core'), require('@actions/exec'), require('semver/classes/semver'), require('null-writable'), require('node:path'), require('node:os'), require('@actions/tool-cache'), require('@actions/cache'), require('node:process'), require('@actions/io')));
 ^
ReferenceError: CancellationException is not defined
    at D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:627:91
    at Object.<anonymous> (D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:31451:2)
    at Module._compile (node:internal/modules/cjs/loader:1155:14)
    at Object.Module._extensions..js (node:internal/modules/cjs/loader:1209:10)
    at Module.load (node:internal/modules/cjs/loader:1033:32)
    at Function.Module._load (node:internal/modules/cjs/loader:868:12)
    at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:81:12)
    at node:internal/main/run_main_module:22:47
And if I also add 
(CancellationException::class.js)
, then it at least starts to run but then fails with
Copy code
> Task :nodeDevelopmentRun FAILED
::group::Verify Windows Environment
::endgroup::
::debug::ReferenceError: intercepted is not defined%0A    at await_0 (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\CancellableContinuation.kt:315:27)%0A    at $verifyWindowsEnvironmentCOROUTINE$4.doResume_5yljmg_k$ (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:345:26)%0A    at verifyWindowsEnvironment (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:341:9)%0A    at verifyWindowsEnvironment$ref.invoke_pbb09x_k$ (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:271:45)%0A    at $groupCOROUTINE$3.l [as fn_1] (D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:30474:16)%0A    at $groupCOROUTINE$3.doResume_5yljmg_k$ (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:335:16)%0A    at group (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:332:9)%0A    at $mainCOROUTINE$0.doResume_5yljmg_k$ (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:271:9)%0A    at main (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:266:9)%0A    at D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:31495:3%0A
::error::intercepted is not defined
Actually, as the production run works, it is not super-blocking for me. I'd just like to understand what the problem is and what it is caused by. Whether I did something wrong I should fix, or whether I hit a bug I should report and where to.
t
Dukat generate types in root package. My expectation is that such type(s) can produce such behaviour. Known “duplication” - 
Symbol
, we have it in browser API and we have it in 
coroutines
.
We expect 
kotlin-actions-toolkit
release tomorrow. Probably it will simplify future checks.
With 
per-module
granularity it also doesn’t work, right? 🤔
v
Not at my computer right now, but I think I tried and it behave exactly the same
t
kotlin-actions-toolkit
released 🚀
v
Nice, thanks
Hm, I almost expected the same pattern like for the Node.js stuff. That the 
Promise
returning methods are suffixed with 
Async
and that there is a 
suspend
one that awaits that promise.
t
Step by step :) For first step it will be fine to use functions with same (“old”) contracts. Especially when you have problems with 
coroutines
in dev mode.
v
I see, I switched to that wrapper, and besides that, it works fine, thanks, basically drop-in replacement, just needed to change the imports. All tests are green. But the dev-mode problem is unchanged.
t
Hm, I almost expected the same pattern like for the Node.js stuff.
Are you talking about top level functions only?
v
Not 100% what you mean
t
In Node we create 
suspend
wrappers only for top level functions (
mkdir
, 
readFile
)
In case of 
@actions/*
looks like you want more
For example wrappers for 
HttpClient
methods
v
Well, I take anything I can get 😄 Those two http client methods (
post
and 
readBody
) are indeed the only non-top-level I use. What is the reasoning to only do it for the top-level methods in node?
t
1. I had use cases only for top level functions 2. Without 
inline
sugar you won’t see it in default suggestion from IDEA
v
I see. Well, whatever you think makes sense. If you think it is best to keep it like it is, fine. If you think it is best to have the suspend sugar for the top-level methods, also fine. Anything else, you guessed it, of course fine. 😄
Any more ideas how I could investigate the dev-run problem? 🙂
t
Kotlin 
1.8.20-beta
?
v
Hard to quickly try, they disabled Dukat
Ah well, thanks to your wrapper work for the actions toolkit I just needed to define three classes, and the dev run worked
Now let's see whehter the same works with 1.8.10 already 😄
No, indeed with the same setup but 1.8.10 it fails with the 
AbstractCoroutineContextElement
error. So it seems to indeed be a Kotlin/JS bug that will be fixed with 1.8.20. Thanks for the help in figuring out.
Hm, what did I wrong with
Copy code
import node.stream.Writable

@JsModule("null-writable")
@JsNonModule
external class NullWritable : Writable
?
Copy code
::debug::TypeError: NullWritable is not a constructor%0A    at wslHelp$slambda.gc (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:77:25)%0A    at wslHelp$slambda.n11 (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:67:47)%0A    at l (D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:13007:16)%0A    at _no_name_provided__qut3iv_2.gc (D:\Sourcecode\other\setup-wsl\build\compileSync\js\main\productionExecutable\kotlin\src\kotlin\coroutines_13\IntrinsicsJs.kt:163:40)%0A    at _no_name_provided__qut3iv_2.CoroutineImpl.r7 (D:\Sourcecode\other\setup-wsl\build\compileSync\js\main\productionExecutable\kotlin\commonMainSources\libraries\stdlib\src\kotlin\util\Standard.kt:55:40)%0A    at _no_name_provided__qut3iv_2.CoroutineImpl.j2 (D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:7214:17)%0A    at <http://DispatchedContinuation.DispatchedTask.gh|DispatchedContinuation.DispatchedTask.gh> (D:\Sourcecode\other\setup-wsl\build\compileSync\js\main\productionExecutable\kotlin\commonMainSources\libraries\stdlib\src\kotlin\coroutines\Continuation.kt:45:5)%0A    at <http://ScheduledMessageQueue.MessageQueue.no|ScheduledMessageQueue.MessageQueue.no> (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\js\src\JSDispatcher.kt:153:25)%0A    at D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\js\src\JSDispatcher.kt:19:48%0A    at processTicksAndRejections (node:internal/process/task_queues:78:11)%0A
::error::NullWritable is not a constructor
😕
t
File annotations must be used instead
v
Oh, ok, it seems for semver it worked properly with
Copy code
@JsModule("semver/classes/semver")
external class SemVer(version: String, options: RangeOptions)
external interface RangeOptions
though, or do I need to use file annotation there too?
No, with semver it is exactly the other way around. If I do it as file annotation, I get "TypeError: SemVer is not a constructor", but with the class annotation it works fine
One day I should learn to understand that 😄
Thanks again for your help
Yay
9 files changed, 45 insertions(+), 575 deletions(-)
Now Dukat is eliminated from my build, many boilerplate removed (and most additions are copyright), no time-waste for running Dukat or compiling the Kotlin classes and it works with 1.8.20-Beta. 🙂 So yeah, if I in the current state just change 1.8.10 to 1.8.20-Beta, the dev-run suddenly works. 🙂
🎉 1
t
Looks like you are ready for more 
suspend
wrappers 🙂
One day I should learn to understand that 😄
There 2 main variants:
Copy code
// #1 Named import
// JS
import { Vampire } from "vampire"

// Kotlin
@file:JsModule("vampire")

external val Vampire

------------------------------

// #2 Default import
// JS
import Vampire from "vampire"

// Kotlin
@file:JsModule("vampire")

@JsName("default")
external val Vampire
FYI - 
@JsNonModule
is redundant for Node, because you need modular format in any case
More transparent 
JsImport
already requested - feel free to vote 🙂
v
Looks like you are ready for more suspend wrappers 🙂
Always :-)
There 2 main variants:
Thanks
FYI - @JsNonModule is redundant for Node, because you need modular format in any case
Ah, ok, will remove it, thanks. Basically looked at what Dukat had produced. :-)
feel free to vote 🙂
Done :-)
Looks like you are ready for more suspend wrappers
Should I find them already and am just blind? I updated to the new version and see that now there is 
execAsync
and alike, but I don't find the suspend versions. Or is that yet again another step? 🙂
The 
...Async
 methods work fine too, anyway 🙂
The suspend methods now are the remaining piece for my current update streak. :-)
t
The 
...Async
 methods work fine too, anyway 🙂
Fine On next step I will generate 
suspend
adapters
I will ping you here or in issue (if you will create it), when it will be ready
👌 1
v
```suspend fun exists(path: PathLike): Boolean =
accessAsync(path)
.then { true }
.catch { false }
.await()```
Actually I'm just blind, there is 
actions.io.exists
ready to use 🙈
t
Yes
And release is ready
v
Yep, seen it, thanks. 🙂 It is already integrated and tests running.
Argh, why do I now get "TypeError: exists is not a function" 😕
Might that indeed be a problem in the wrapper?
Copy code
TypeError: exists is not a function%0A    at exists_0 (D:\Users\leonid.khachaturov\code\kotlin-wrappers-new\kotlin-actions-toolkit\src\jsMain\generated\actions\io\exists.kt:10:5)%0A
  at $mainCOROUTINE$<http://11.ec|11.ec> (D:\Sourcecode\other\setup-wsl\src\main\kotlin\net\kautler\github\action\setup_wsl\SetupWsl.kt:287:17)%0A    at $verifyWindowsEnvironmentCOROUTINE$15.CoroutineImpl.r7 (D:\Sourcecode\other\setup-wsl\build\compileSync\js\main\productionExecutable\kotlin\commonMainSources\libraries\stdlib\src\kotlin\util\Standard.kt:55:40)%0A    at $verifyWindowsEnvironmentCOROUTINE$15.CoroutineImpl.j2 (D:\Sourcecode\other\setup-wsl\build\js\packages\setup-wsl\kotlin\setup-wsl.js:6918:17)%0A    at resume (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\internal\DispatchedTask.kt:178:26)%0A    at dispatch (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\internal\DispatchedTask.kt:166:9)%0A
 at dispatchResume (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\CancellableContinuationImpl.kt:397:9)%0A    at resumeImpl (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\CancellableContinuationImpl.kt:431:21)%0A    at resumeImpl$default (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\CancellableContinuationImpl.kt:420:13)%0A    at CancellableContinuationImpl.j2 (D:\opt\teamcity-agent\work\44ec6e850d5c63f0\kotlinx-coroutines-core\common\src\CancellableContinuationImpl.kt:328:9)%0A
exists is not a function
Yeah, according to the readme it does not exist: https://github.com/actions/toolkit/tree/main/packages/io
It is in 
io-utils.ts
, not in 
io.ts
like the others
That also explains why I didn't use it when not having the wrappers version 😄
t
It can be fixed. Could you please report an issue?
v
Sure, will the fix be that it will be not available, or that it will work? I guess the former as it is not an officially supported function?
t
Sure, will the fix be that it will be not available, or that it will work?
Looks like it will be better to delete it
👌 1
For now we have only 
generated
sources for Node
If you will create PR with 
existsAsync
and suspend 
exists
I will merge it
v
I guess those could easily confused with the deprecated 
exists
from Node itself?
Especially the existing 
existsSync
representing that deprecated 
exists
t
Deprecated 
exists
- with callback. Your version will be without callback - no problem detected.
v
I'm talking about
Copy code
package node.fs
external fun existsSync(
    path: PathLike,
): Boolean
For the deprecated 
exists
with callback you don't even provide a wrapper, do you?
Or actually I don't know what I'm saying, I mixed those up. 😄
t
Yes, we don’t convert functions with callback
v
t
Return type missed 😞
v
whoops, bad copy & paste, sorry
fixed
t
pre.498
is waiting for checks 😉
v
Yep, running 🙂
Yay, everything green. I think I'm finally finished for now. :-) Thanks for all your help again.
🚀 1
😃 1
24 Views
Open in Slack
PreviousNext
Powered by