Skip to content

Charles Oliver Nutter
headius

Organizations

@jruby @jenkinsci @ruboto @redcar @mirah @jnr @openjdk-mirror
@headius
headius commented on issue jruby/jruby#3656
@headius

Patch for rubygems.rb that makes this work correctly. It should not be necessary: diff --git a/lib/ruby/stdlib/rubygems.rb b/lib/ruby/stdlib/rubyge…

headius opened issue jruby/jruby#3656
@headius
Autoload is not overwritten by subsequent explicit require
headius commented on issue jruby/jruby#3653
@headius

This is fixed, but it's still an issue for other spawn forms. However, I still feel like it's asking for trouble spawning subprocesses with env var…

headius pushed to master at jruby/jruby
@headius
headius pushed to master at jruby/jruby
@headius
headius closed issue jruby/jruby#3653
@headius
`system` method is different behavior from JRuby 1.7.x and CRuby 2.3.x
@headius
headius commented on issue jruby/jruby#3653
@headius

It's worth pointing out that MRI will behave the same as us even without chdir if the command given looks like a command line. For example, if I ch…

headius commented on issue jruby/jruby#3653
@headius

Ok, the problem here is that the new logic for process spawning does not support chdir in the same way as MRI. In MRI, if you specify chdir when ca…

headius commented on issue jruby/jruby#3653
@headius

Ok, interesting. I'll have a look.

headius commented on issue jruby/jruby#3649
@headius

@enebo @subbuss Any thoughts on triggering JIT sooner than 50 calls when we have a hot loop? I know this falls into the "let's add OSR because benc…

@headius
@headius
@headius
@headius
headius commented on issue jruby/jruby#3649
@headius

0.9s versus 1.1s is not bad. And for me, the difference is greater. JRuby wins for both your prime? implementation and for the stdlib version (I se…

headius pushed to master at jruby/jruby
@headius
  • @headius 35d7e5b
    Fix yield and yieldSpecific paths to use helpers.
headius pushed to master at jruby/jruby
@headius
  • @headius 7243b99
    Fix improper yield self and implement new yield path in JIT.
headius commented on issue jruby/jruby#3649
@headius

Ok, so other than the 2.2/2.3 stdlib issue, the remaining problem here is that a single long-running method does not trigger JIT. We don't have a g…

headius commented on issue jruby/jruby#3649
@headius

Ok, it looks like Ruby 2.3 improved the performance of prime.rb. JRuby 9.0.5.0 roughly matches MRI 2.2 performance, and JRuby with 2.3 stdlib rough…

headius commented on issue jruby/jruby#3649
@headius

It appears that at least part of this bad benchmark result is due to the fact that most of the hot call path is not compiling before the benchmark …

headius commented on issue jruby/jruby#3649
@headius

It looks like this is never getting out of the (simple) interpreter: Compiled + native Method 28.9% 1649 + 8 org.jruby.ir.interpreter.StartupInter…

headius commented on issue jruby/jruby#3649
@headius

Wow. Ok, I'll look into this. Definitely something wrong.

headius commented on pull request jruby/jruby#3637
@headius

:+1: Nice work Chris. We do use preallocated exceptions for a few that never have backtracked (like for "expected" exception like EAGAIN) but I think …

headius pushed to master at jruby/jruby
@headius
headius commented on issue rubygems/rubygems#677
@headius

Oh, and for the @copiousfreetime: I'd like to see the feature be more useful, either by making "not-uninstallable normal gems" or by doing the rest…

headius commented on issue rubygems/rubygems#677
@headius

There are a few issues with making things "just gems": Startup time is impacted when RubyGems has to load. Normally on MRI (before 2.3) unless you…

Something went wrong with that request. Please try again.