Benoit Daloze

https://github.com/ruby/rdoc/pull/1032 so rdoc uses the more portable pattern `eval("BEGIN { throw :valid, true }\n#{code}")`

Thank you, we should indeed look at this again, especially since the migration to TruffleString which should help with this. Could you share your new benchmark file? In the gist...

Thank you, I can reproduce it with `truffleruby-dev`. And adding: ```ruby if IS_TRUFFLE FILEDATA["gzdata"] = Truffle::Debug.flatten_string(FILEDATA["gzdata"]) end ``` makes them both the same speed.

The plan here is to not use the C extensions lock if `RB_EXT_RACTOR_SAFE` is used for an extension (already used in some extensions), or if `RB_EXT_THREAD_SAFE` is used (would be...

With Truffle Node inlining there is less need to reduce the number of nodes, and we did some of this so closing.

When replacing `Activate.call` by `using(Rf)` it works fine on both CRuby and TruffleRuby: ```ruby puts RUBY_DESCRIPTION class A end module Rf refine(A) do def foo; :rf; end end end class...

Thanks, I forgot about that one. I replied on the CRuby issue. `for ... in` is less clear than `.each` because IIRC CRuby uses the same frame outside and inside...

Right, but I think we could change semantics for such edge cases in CRuby as needed because they're extremely unlikely used in practice, and it would make sense from a...

This is now properly fixed in childprocess 5 :tada: https://github.com/enkessler/childprocess/pull/175 https://github.com/enkessler/childprocess/releases/tag/v5.0.0

Yes, good catch, https://github.com/graalvm/mx/commit/ec2696fabfd92f200a3845c7f7f04c5c805528e8 fixes this (@horakivo ran into this issue today). I'll close this already so I don't forget, but we'll update the minimum mx version in truffleruby so...