Thanks for looking into this. That's a peculiar issue.

The sequel and concurrent-ruby job fails were due to a bug in #9198 that caused the build to generate broken binstubs. Fixed in f21870a.

@headius thanks. I think everything is green now. The single failing test looks like an unrelated time-sensitive issue.

@evaniainbrooks I think those last failures are due to a similar change being needed in some JIT-related path. That spec run runs the language specs 4 times in the same runtime, so that the JIT can fire and fully bind all calls.

I'll see if I can suss out what's wrong.

I think I found the right spot in JIT's argument-handling logic to patch. It has diverged for various reasons and needs to be reined in, but if this passes we can proceed for now.

@headius thank you for the explanation. That certainly explains some other things that perplexed me while I was stepping through this.