Update scala 3 version, drop support for scala 2

[hughsimpson]

Pull Request Checklist

• I agree to licence my contributions under the MIT licence
• I have added copyright headers to new files -- No new files
• I have added tests for any changed functionality -- No added functionality

Removes scala 2, updates scala 3 to 3.4.2, adds -experimental compiler flag to negate need for bug/workaround

[hughsimpson]

Hadn't updated the scala 3 version in the GitHub workflow so that test failed. Pushed an update there

[hughsimpson]

So anyway this whole pr was way off base and I do apologise for that. I guess I figured it might still be worth leaving open on the off-chance, and because it had already been tagged (presumably because of my initial false claim about the downstream impact 😬). I was tempted to close it or mark as draft but... idk. Potentially scala 3.4 warrants discussing even though I originally introduced the idea for bad reasons

[barkhorn]

Hey, i don't know about that. It will need a review for sure if there's a potential impact. But we should be able to follow new Scala versions in principle unless there's some breaking changes. After all, if someone wants a ScalaMock for an older versions, Sonatype (to my knowledge) doesn't purge versions so there's always a fallback

[martijnhoekstra]

As a library, I'd suggest staying on LTS versions, as not to force downstreams to update

[goshacodes]

I think there is no problem of bumping version up to latest 3.4 or even 3.5.0 if we can get rid of annotating everything as @experimental, this is crucial, essentially for library users. Scala 3 offers backward compatibility for all versions.

Anyway we are exploiting a bug in the compiler to make it work.

If we decide to go experimental way, I would also consider using TupledFunction to make scalamock simpler

[goshacodes]

Also, in my opinion, we should drop support for Scala 2.12 and Scala 2.13, since library (scala 2) is stable enough and nobody would fix anything for scala 2 anyway. This would unlock futher library development.
I have already made the migration as smooth as possible and now we should really move on or the library becomes dead soon

[hughsimpson]

@goshacodes I love that energy. Happy to drop scala 2 support in this pr. Will look into your suggestions on this after I've addressed the ones on the other two.

[barkhorn]

Agree on Scala 2.x support, we have a maintenance-5.x branch that could be used to patch bugs there, but it's been pretty hard to work out some of the macro-related problems in the past, I certainly tried :D

[goshacodes]

Let's proceed then. @hughsimpson please drop scala 2 in this PR

[hughsimpson]

@goshacodes Dropped scala 2, haven't looked into TupledFunction yet - probably won't have the bandwidth for a few days now - but that can probably come later anyway?

[goshacodes]

@goshacodes Dropped scala 2, haven't looked into TupledFunction yet - probably won't have the bandwidth for a few days now - but that can probably come later anyway?

At first lets rewrite everything to scala 3 with scalafix (or similar utility, not sure which one does it)

With tupled functions research is needed anyway, If you want to take it, it will be great

[goshacodes]

@barkhorn Could you please close all scala-2 related issues with wont-fix to cleanup everything, after this MR is merged

Also this is not related, but If you know, could you please tell us, what does proxy package meant for?
Not found anything related in the docs or in the issues

[goshacodes]

@hughsimpson opened an issue about TupledFunction #533

[hughsimpson]

Not sure if I'll get to it or not... Once the next release is cut with the bugfix prs I'll probably be snowed under trying to finalise some chunky migrations, so will probably vanish off the reservation for a while tbqh. But if it's still waiting for me when that work is done I'll be diving straight in

[barkhorn]

@barkhorn Could you please close all scala-2 related issues with wont-fix to cleanup everything, after this MR is merged

Also this is not related, but If you know, could you please tell us, what does proxy package meant for? Not found anything related in the docs or in the issues

The proxy package is using Java reflection, specifically proxies, to dynamically instantiate a mock object. It's very limited compared to macro mocking but the implementation is very concise (https://github.com/paulbutcher/ScalaMock/blob/master/shared/src/main/scala-2/org/scalamock/proxy/ProxyMockFactory.scala), and there may be corner cases where it was/is used to work around some of the macro problems with the Scala 2 macro implementation.
https://docs.oracle.com/javase/8/docs/api/java/lang/reflect/Proxy.html

With a Scala 3 focus, there is probably not a great rationale anymore for this package

[hughsimpson]

I don't mind ripping out a bit more whilst I'm in here. @barkhorn @goshacodes you think this pr should go further? Or merge as it stands and perform the rest of the actions in follow-up?

[goshacodes]

I think we should merge as it is now and handle it later. Just asked)

[barkhorn]

sorry, some recent changes have caused merge conflicts, would you be able to look again please?

[goshacodes]

I've recently found one issue, which IMO should be fixed before we merge this.

Current scalamock internal implementation is not thread safe and may cause throw of ConcurrentModificationException.
This can and should be fixed for scala 2 also

[hughsimpson]

I'll rebase this tomorrow. No worries. As far as the concurrency issue goes, no idea yet. Will look then.

EDIT: I couldn't write a test for that error yet.

[hughsimpson]

I've had a crack at that bug here. Pushed the test before the fix so you can see that it at least failed in scala 2.13 and 3 with the expected error. Probably the best I can do, took way too long to write a test that failed consistently without the fix. Not sure why it passes on 2.12, maybe differences in Future implementation?

[goshacodes]

This should go as 7.0 series