14:18 <JasonLunn[m]> Good morning - has anyone reported changed behavior for the JRubyMethod annotation in 9.4.x?

14:18 <JasonLunn[m]> I didn't see anything in GH Issues about it

14:19 <JasonLunn[m]> I ran the test suite for Protobuf under 9.4.x for the first time today using 9.4.3.0

14:21 <JasonLunn[m]> And one of the failures I'm chasing down suggests that @JRubyMethod isn't blocking calls to our Java code when the ruby invocation of the method doesn't provide enough required arguments

14:21 <JasonLunn[m]> https://github.com/protocolbuffers/protobuf/blob/3a477b19523d5401f1ef1c35f1c459b3c37dea5c/ruby/src/main/java/com/google/protobuf/jruby/RubyRepeatedField.java#L82

14:22 <JasonLunn[m]> Instead of an ArgumentError getting thrown by the JRuby runtime, our method gets called with a zero length args array,

14:23 <JasonLunn[m]> and then we proceed to trigger a Java Array index out of bounds exception by assuming that it is safe to access index 0.

15:00 <enebo[m]> Jason Lunn: Can you open an issue? Also if you can figure out the caller from Ruby side

15:00 <enebo[m]> which could also be **h where h = {}

15:00 <enebo[m]> My only thought is in Ruby 3 **{} will get removed as an explicit argument

15:01 <enebo[m]> with that said I don't think our anno should accept that but I am just thinking stream of consciousness

15:02 <JasonLunn[m]> Will do. I'm trying to create a minimum reproducible fail case now. A trivial example works as expected, so looking into what we're doing in Protobuf to excite the issue.

15:03 <enebo[m]> I will git blame that code and see what else may have changed. All I changed was keyword args stuff but I don't think it should affect arity

16:43 <JasonLunn[m]> Got it down to a minimal reproduction case

16:44 <JasonLunn[m]> Writing the issue up now

16:59 <JasonLunn[m]> https://github.com/jruby/jruby/issues/7851

18:10 <headius> Davinci: Did you find a way to do this? We could possibly up the limit ourselves in the Java code but I exposed this setting to avoid having to do that

18:46 <headius> enebo: woot I think I have mavengem updated and building

18:48 <headius> all base tests in the mojo seem to pass... then it fails on tests related to "virtus"

18:48 <headius> etc

18:49 <headius> Tests in error:... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/44c37a0a34466b5c336ff2a3894f1aec045f12d4>)

18:50 <headius> public void virtusPomWithAuthentication() throws Exception {

18:50 <headius> // this test goes online to rubygems.org

18:50 <headius> these may be testing the old API but perhaps the user agent is not matching the whitelist

19:07 <headius> looks like that's the problem

19:08 <headius> so I'm going to start a PR with my updates and then look at getting this moved to the new API

19:27 <headius> https://github.com/jruby/mavengem/pull/9