https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124641
--- Comment #13 from H.J. Lu <hjl.tools at gmail dot com> --- (In reply to Jakub Jelinek from comment #9) > (In reply to Jakub Jelinek from comment #7) > > Ugh, the port expect to tcl9 mess was even backported to F43? Even in F44 > > it causes a mess: > > https://lists.gnu.org/archive/html/bug-dejagnu/2026-03/msg00000.html > > https://gcc.gnu.org/pipermail/gcc-patches/2026-February/708794.html > > https://gcc.gnu.org/pipermail/gcc-patches/2026-March/711403.html > > I've committed the last one now in r16-8295, the first one is still > unreviewed. > Could you please try the patch in the middle and maybe the dejagnu one as > well, if that fixes it? I certainly don't see it on Fedora 45 with the > fixed up dejagnu and the middle patch applied. > Perhaps tcl9 in Fedora 43 behaves differently from the one in Fedora 45 and > doesn't error on -encoding binary but silently ignores it or something? I applied all 3 patches on Fedora 43. It didn't help. After downgrading expect, the problems went away. If GCC testsuite doesn't work with tcl 9, should "make check" stop when tcl 9 is in use?
