Charles Oliver Nutter

> So there's basically no way to avoid this warning when using bundler and the jruby-launcher gem. Oops, I did not notice it was complaining about the "2.0.0.pre2" version of...

So the short story here is that it should be possible to step-debug Ruby and Java code together by setting JRuby to force JIT compilation of code it runs. That...

I started doing a lot of work on this in #8259: * More readable stack trace, still up for debate on format * Local variables show up as locals in...

Good catch! Must be some slight variation in the JIT logic for Date parsing.

Reusing a whole Hash instance would be weird, but I would not be surprised if it's not dup'ing everywhere it should. I'll look into it.

Seems to be specific to the invokedynamic mode. Disabling that with `-Xcompile.invokedynamic=false` allows it to run to completion. Can be forced to fail immediately with `-Xjit.threshold=0`.

Wow... if I modify `_parse_impl` to print out the incoming hash, it runs to completion. But it does look unique every time. So there's something weird up with that hash.

Must fix for 10.0.1.0. Looking into it more now.

Fixed on 9.4 branch for 9.4.14.0 and merged forward for JRuby 10.0.1.0.

We may want to use the same excludes logic as in JRuby's CI so we can track regressions. Thank you for kicking this off!