GNOME Bugzilla – Bug 794555
glib meson build fails on MinGW due to misdetected functions
Last modified: 2018-05-04 23:29:22 UTC
Meson incorrectly detects stpcpy() and posix_memalign() functions to be availabe due to the fact that gcc has built-ins that kick in for the simple cases that the normal meson function test source code presents.
Created attachment 369946 [details] [review] Check stpcpy() and posix_memalign() to *not* to be built-in GCC has built-ins for these functions, which might give a compile-only test an impression that the functions are actually present in the C runtime. Use a linked test to be sure. Specifically, both functions are missing on Windows.
Review of attachment 369946 [details] [review]: Okay
Attachment 369946 [details] pushed as dcc452b - Check stpcpy() and posix_memalign() to *not* to be built-in
The stpcpy check doesn't seem to work under MSYS2, any ideas? https://gitlab.gnome.org/creiter/glib/-/jobs/26603
(In reply to Christoph Reiter (lazka) from comment #4) > The stpcpy check doesn't seem to work under MSYS2, any ideas? > https://gitlab.gnome.org/creiter/glib/-/jobs/26603 I think that became attachment #371365 [details] on bug #793729, right?
(In reply to Philip Withnall from comment #5) > (In reply to Christoph Reiter (lazka) from comment #4) > > The stpcpy check doesn't seem to work under MSYS2, any ideas? > > https://gitlab.gnome.org/creiter/glib/-/jobs/26603 > > I think that became attachment #371365 [details] on bug #793729, right? Yes.
I've seen this bug linked in a meson report: https://github.com/mesonbuild/meson/issues/3482 I don't know what's it's about, but does it mean a better fix is possible than the one merged? Nirbeek, can you please comment what glib should do if meson is fixed?