Scala Native 0.4.0

[larsrh]

If this PR goes through, I'll back-publish for the last ScalaCheck release.

Should we drop 0.3.9 support then, too?

[SethTisue]

I vote for dropping 0.3.9. Additional targets are enough of a maintenance burden as-is, without also having to worry about multiple versions of those targets.

Even if you don't buy that general argument, I would argue re 0.3.9 specifically that 0.3.9 is so old and remained 2.11-only for so long that it really isn't worth supporting.

[larsrh]

Turns out they aren't even source compatible:

[error] /home/runner/work/scalacheck/scalacheck/native/src/main/scala/org/scalacheck/Platform.scala:14:8: object PreloadedClassLoader is not a member of package scala.scalanative.testinterface
[error] import scala.scalanative.testinterface.PreloadedClassLoader
[error]        ^
[error] /home/runner/work/scalacheck/scalacheck/native/src/main/scala/org/scalacheck/Platform.scala:27:25: not found: type PreloadedClassLoader
[error]     loader.asInstanceOf[PreloadedClassLoader].loadPreloaded(name)
[error]                         ^
[error] /home/runner/work/scalacheck/scalacheck/native/src/main/scala/org/scalacheck/Platform.scala:14:40: Unused import
[error] import scala.scalanative.testinterface.PreloadedClassLoader
[error]                                        ^

So either I'll have to cheat when back-publishing, or cry.

[gabro]

Thanks @larsrh for working on this, this should also unblock support for 0.4.0 support in MUnit (across 2.11, 2.12 and 2.13)!

About dropping 0.3.9, I think it would be great to nudge people out of the 2.11 ecosystem, so that we can finally move on (like in Metals, for example, where 2.11 support is largely there for Scala Native projects)

[larsrh]

@WojciechMazur if you could give us some pointers how this code should look like nowadays, it would be much appreciated!

[gabro]

@larsrh I'm looking at similar errors in MUnit and probably this helps scala-native/scala-native@3077e74

[gabro]

For reference, here's the equivalent fix in MUnit (just tested, it runs correctly) https://github.com/scalameta/munit/pull/298/files#diff-16de51ca52ec70a26d1950ea7aa6c11432074b200e28d9316cba1077dbaadf3a

[WojciechMazur]

I also personally would suggest dropping 0.3.x. Instead of using PreloadedClassLoader directly in 0.3.9 you could use org.scalajs.testinterface.TestUtils (deprecated since 0.4.0), it's boundled in test-interface
You can check TestUtils API in this link

[larsrh]

I've dropped Scala Native 0.3.9, Scala 2.11, changed the build accordingly, but left the broken code intact. To be honest, I don't understand what I need to change in there. If anyone could give me a patch, that'd be much appreciated.

[gabro]

I can take a look.

About 2.11, why the drop?

[larsrh]

Why not? Scala Native is the only reason why anyone still supports it, and that reason just got void.

[gabro]

I agree in principle, but I wouldn’t conflate the two. It would mean to drop 2.11 downstream as well (e.g MUnit-scalacheck) and I would prefer to do it separately than this update

On Thu, 21 Jan 2021 at 20:44 Lars Hupel ***@***.***> wrote: Why not? Scala Native is the only reason why anyone still supports it, and that reason just got void. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#751 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFI5ZGGWCMRERD5O2M7QWTS3B73FANCNFSM4WJQEOQA> .

-- <https://buildo.io/> gabriele petronella / co-founder gabriele@buildo. <gabriele@buildo.io> buildo https://buildo.io

[larsrh]

Sounds plausible, but I was hoping to sneakily back-publish Scala Native support for the latest ScalaCheck release, so you won't even notice downstream 😉

[SethTisue]

About 2.11, why the drop?

my take on this is: because supporting old versions of things is death by a thousand papercuts for unpaid OSS maintainers

[gabro]

@SethTisue 100% agree, I was just wondering on the opportunity of doing so while also dropping a target version at the same time.

But back-publishing resolves this, so no objections on my end 🚀

@larsrh in the meantime, I made it compile, but I'm hitting a runtime crash when running the tests due to how the new reflective instantiation works.

I'll dig a bit more into it later today

[gabro]

All green ✅ 🎉

[larsrh]

Thanks a ton @gabro!

[larsrh]

Seems like it also builds for 1.15.2. Running the release right now.

[larsrh]

Finally the artifacts appear to have been released. Sonatype was exceptionally slow today.

[gabro]

Yep, I can confirm I've successfully downloaded them in MUnit, thanks!