Abigael

> Rather than that, how about introducing " `cores` ocamltest variable? Its default value would be 1 but the tests which require more than one core could override it and...

> This is not quite right. I introduced the `OCAML_TEST_SIZE` environment variable to give tests a hint of how many cores they can expect: https://github.com/ocaml/ocaml/blob/trunk/testsuite/HACKING.adoc#creating-a-new-test . It's effective to avoid...

I updated the list following the merge of #11183. @xavierleroy I do have an ARM 32 bits runner PoC on the grill, I need to make a few adjustments and...

Merging #600 would fix this. On Wed 14 Jul 2021 at 18:31, Thomas Leonard ***@***.***> wrote: > How do you compile this? I get: > > In file included from...

Thank you for the test case, I think I managed to nail it down. What happens is that core links `systhreads` in, so the runtime will go through the process...

@kayceesrk I will redo a full run and confirm the core testsuite passes (outside of this specific test that indeed #629 fixed). Once everything is green I'll close this issue,...

Thanks for the testcase! The loop happens when the file descriptor is being closed, triggering a flush: https://github.com/ocaml-multicore/ocaml-multicore/blob/4.12%2Bdomains%2Beffects/runtime/io.c#L211 This loop will never complete because as it turns out, the assertion...

I will do another pass of review this week, and I'd like to ping people who took a look at the C side in the original eventring PR at #10964,...

Agreeing this would be a very nice improvement.

> I get around 50ms (the tick thread timeout?) I think this intuition may be reasonable? https://github.com/ocaml/ocaml/blob/trunk/otherlibs/systhreads/st_stubs.c#L467 When multiple domains are running, if a domain exit and it is running...