#jruby on 2021-03-22 — irc logs at freenode.irclog.whitequark.org

06:23 <kares[m]> I realized we can work-around the issue (for now) by initializing the java.lang.Object proxy class early

06:23 <kares[m]> so I will finish the PR with that

11:17 <kares[m]> JavaClass PR 💚 - there's a ton of room for cleanup (as well as more "wrapping" to be dropped) ... but should be working fine.

11:17 <kares[m]> compatibility with JavaClass/JavaField/JavaMethod/JavaConstructor is very high except a few gotchas that are expected

11:17 <kares[m]> (e.g. JavaClass.for_name or raising wrapped exceptions from JavaMethod/JavaConstructor invocations)

11:17 <kares[m]> compatibility with JavaClass/JavaField/JavaMethod/JavaConstructor is very high except a few gotchas that are expected

11:17 <kares[m]> (e.g. JavaClass.for_name or raising wrapped exceptions from JavaMethod/JavaConstructor invocations)

12:46 <headius[m]> kares nice work, I wasn't sure it would be possible to get this much compatibility with the old API

13:07 <kares[m]> hopefully smt that is fine for a major release ...

13:07 <kares[m]> I think you're original assumption was correct to just replace them with Java proxies but as noted it isn't perfect.

13:07 <kares[m]> hopefully smt that is fine for a major release ...

13:08 <kares[m]> the compatibility patches aren't externalized in modules right now as you suggested

13:09 <headius[m]> We have steered people away from that API for years so there may not be many that would even notice

13:09 <headius[m]> I have not done a search to find .java_class callers though

14:10 <kares[m]> .java_class on its own is definitely used (how wild everyone gets is anyone's guess)

14:11 <headius[m]> yeah

15:06 <headius[m]> kares: any reason not to merge now? The only failure appears to be a sorting issue in expected packages

15:07 <headius[m]> test_does_not_load_ji_on_boot should just sort expected and actual package lists

15:26 <byteit101[m]> A quick glance makes it look like there will be at most only minor merge conflicts with my PR

15:26 <byteit101[m]> (vs kares PR)

15:32 <headius[m]> byteit101 Ahh good! We will out out 9.2.17 this week and turn attention back to 9.3 wrap-up

15:34 <headius[m]> put out

15:34 <headius[m]> enebo: on that note, I see nothing new we would need to worry about in .17

16:42 <kares[m]> oh there's still a test failure? I thought I handled everything ... let me check

16:43 <headius[m]> go ahead and merge after you are satisfied, I don't see any other issues

16:48 <kares[m]> okay will do in a few - would be great to have this out as a snapshot

16:48 <headius[m]> yeah we need to get stuff landed on master so folks can try out these larger PRs all together

17:41 <enebo[m]> ok I think we can release tomorrow morning?

18:16 <headius[m]> enebo: yeah sounds good to me

18:47 <headius[m]> enebo: today working on some peripheral stuff: jffi support for macos arm64

18:48 <headius[m]> FFI gem appears to have merged in updated libffi and other patches to make it work so it should be possible for us too.. then we will be able to say 9.3 supports apple silicon natively

18:52 <headius[m]> time to fire up the old apple silicon dev kit again

18:54 <enebo[m]> they need a GHA for that hardware