Lasse Collin wrote: > I have a hunch that only GCC < 15 and glibc's headers are problematic, > so excluding those might keep "make check" happy everywhere. ... > > I attached a draft patch to demonstrate the idea.
That looks like something we could apply, yes. > It is optimistic and > assumes that glibc 2.45 will have updated headers. For now, therefore replace 2.45 with 2.99, and when the glibc issue is fixed, we will now which glibc version to put in instead. > The tests pass on > GNU/Linux (glibc 2.24, 2.41, and 2.43, and musl 1.2.6; a few GCC > versions from 4.9 to 16.1, and Clang 22) and on FreeBSD (Clang), > OpenBSD (Clang), NetBSD (Clang, GCC), AIX (ibm-clang, GCC), and Solaris > (Developer Studio). Good! This looks promising. I would also give it a try in my multi-platform CI, when time has come. Bruno
