Gabriel Scherer

Why is the test failing? Is this a bug with the Windows implementation, or more generally an OCaml bug? The test code is cryptic, but it is here: https://github.com/ocaml/ocaml/blob/23bf9056b9ec79a1675ab57cb24dcb64e6c9c646/testsuite/tests/memory-model/publish.ml#L125-L137 My...

(cc @maranget I guess, who wrote the memory-model tests in the first place.)

For the record: - the faulty code is using `Printf.sprintf "%d"` which does not use Bytes.unsafe_to_string, it uses `Bytes.content` which performs a copy of the bytes into a string using...

@maranget is currently jumping up and down in my temporary office, excited at the idea that `memmove` (used in `String.sub`) may use "non-temporal writes". His [cavern of x86 memory behaviors](http://diy.inria.fr/x86-memtype/)...

Can you find the implementation of `memmove` on the CRT versions that you are using?

My understanding is that the bug is still failing randomly (see #12425 for the list of known CI failure, the last one was three weeks ago), but we have no...

The quote describes the intended semantics of the OCaml memory model. If you manage to contradict it with some sensible OCaml code, then you have found a bug; but in...

... or maybe refocus the current issue with some light editing, if you prefer. Thanks!

Related issues: - https://github.com/ocaml/ocaml/pull/11649#issuecomment-1288053810 - https://github.com/ocaml-multicore/domainslib/issues/77#issuecomment-1290398870 The fundamental issue is that when you have more domains than hardware threads, then one of the domain will be paused by the OS...

> Perhaps for 5.0 might have to put a warning in the docs about this, that it is not just a little slower when you exceed number of cores, but...