Antonin Décimo

Earlier, Xavier wrote: > I need to be convinced this is not a regression compared to just using winpthreads, or a more modern implementation of the POSIX API for Win32....

> I started a `precheck` job for this PR: https://ci.inria.fr/ocaml/job/precheck/1066/ Amazing, it's failing on mingw-w64 both 32 and 64 bits. This is fishy. ``` Makefile.common:492: warning: undefined variable 'CYGWIN' Makefile.common:494:...

I don't understand the multicoretest failure, but since they run on Ubuntu, I don't think they would be affected by the switch from winpthreads to the WinAPI ;) The AppVeyor...

Thanks @jmid, multicoretests succeeds on this PR: https://github.com/MisterDA/multicoretests/pull/2!

> You need to carefully regenerate `configure` in commit #.2, there are conflict markers in it (which disappear in #.3). Unless you intend to squash before merge this needs to...

> I simply use `tools/autogen`... This is equivalent, `tools/autogen` calls the system's Autoconf, which might be different (patched, newer, older) than Ubuntu's Autoconf 2.71 that we use in the CI....

Ok, that's not a hill I'm willing to die on, I've removed that commit.

Rebased. @shindere is this good to go?

> Do we really want the first one, though? I suppose you're referring to [configure: say 'the C compiler' instead of printing full path](https://github.com/ocaml/ocaml/pull/13556/commits/8697af489e51080f6c4cd8e045f68380e5750843) instead of [configure: use $warn_error_flag instead...

> > > In one way I do understand that if `$CC` contains command-line options as may be the case for, say, `-m32`, including it in all the messages makes...