after emerging gtk+-2.0.6-r1.ebuild, every app that uses the multiline text entry widget coredumps whenever i click on the widget. this includes gedit, nautilus, and a host of others. unmerging gtk+-2.0.6-r1.ebuild and emerging the older gtk+-2.0.5-r2.ebuild fixes the problem. is this a gtk bug, or a bug in the patches gentoo applies? i can't test the original 2.0.6 because the ebuild for it has been removed.
I'm having the same problem too, on the following: gtk+-2.0.6-r1 gcc-3.2 (Gentoo v1.4) Celeron (Mendocino) CFLAGS="-march=pentium2 -O3 -pipe -fomit-frame-pointer" The programs just segfault for no reason; no error message at all. I didn't use gdb to get a more detailed output though, since I don't know how. This bug has been haunting me for months now. Please fix it soon! >_< I've tried and it's beyond me as to what's going on...
gtk+ 2.0.6 never existed officially. we had a 2.0.6-r1 was the first to stay in portage for more than a few hours. This does not happen on my gcc 3.2 -march=athlon-tbird -O2 compiled system please file this bug in http://bugzilla.gnome.org/ in the gtk+ part.
i should add a few more details. i have an athlon 1800+ system, with GCC 3.2, compiling with the following flags: CHOST="i686-pc-linux-gnu" CFLAGS="-march=i686 -fomit-frame-pointer -funroll-loops -O3 -pipe" CXXFLAGS="-march=i686 -fomit-frame-pointer -funroll-loops -O3 -pipe" i'm going to try re-merging this, and if it still doesn't work maybe i'll change it to -O2, or try building GTK from the stock .tar.gz packages.
interesting, i recompiled with -O2 instead of -O3 and now it works fine. i wonder if this a bug in GTK, or in the patches gentoo applies to it before compilation? or, heck, maybe a GCC 3.2 bug?
considering we aren't patching it.... Well. I'd say overoptimization. Closing this now: resolution: Too high optimization options cause odd bugs.
is this true? if gentoo doesn't support -O3, that really should be in the user documentation somewhere. the make.conf defaults have -O3, that's a pretty strong endorsement. of course, i realize that you can't fix this bug on your own, but 2.0.6 should be masked if it breaks with the default optimizations... anyways, i've reposted the bug into the gnome bugzilla here: http://bugzilla.gnome.org/show_bug.cgi?id=94293 so hopefully they'll fix it for the next version.