Simon Cruanes

Ok, this now uses common code for initialization in both domain.ml and thread.ml, thanks to a new `camlinternalTLS` module. I haven't worked on documentation yet (I assume it's mostly updating...

Done, except for step 2. I feel like it's overkill to require a bootstrap just for an internal name like this. This is however just my opinion, possibly biased by...

I bisected a segfault that occurs in tests/parallel/catch_break.ml, back to commit 57e5926b1a07b0483fb719fe003d4b2e02b0a70a . In that test, the GC wakes up in a signal handler and reaches a corrupted state. It...

Broadly what I mean is that tests would pass before it, and fail after it, even with later corrections. Something went wrong in initialization code somewhere and I'm still trying...

Should be fixed on main, could you give it a try before I make a release? Edit: ah nvm it's not just about bounds, I need to use the new...

I'm kind of dreading the move to Exp_function, this new AST is more painful! 😅

Ahah no worries, I'll just need to look at what other people have done :)

@polytypic proposed a nice solution for fast TLS here: https://discuss.ocaml.org/t/a-hack-to-implement-efficient-tls-thread-local-storage/13264 and suggested the discussion continue here. To reply to @gasche from earlier: > Indeed, Thread programming is pretty much unchanged...

It's still critical and I still miss it a lot 🙃

I'm not sure what the problem is. I tried locally with: - server side: ```ocaml Server.add_route_handler ~meth:`HEAD server Route.(exact "head" @/ return) (fun _req -> Response.make_void ~code:200 ~headers:[ "x-hello", "world"...