With a current (todays) snapshot, hwsetup fails on livecd-building: >>> Emerging (59 of 85) sys-apps/hwsetup-1.1 to / * Applying hwsetup-1.1-dyn_blacklist.patch ... [ ok ] * Applying hwsetup-1.1-fastprobe.patch ... [ ok ] * Applying hwsetup-1.1-gentoo.patch ... [ ok ] gcc -I/usr/include/kudzu -DBLACKLIST -Wall -fPIC -O2 -s -o hwsetup hwsetup.c -lkudzu -lpci /usr/lib/gcc/powerpc-unknown-linux-gnu/4.1.1/../../../libkudzu.a(macio.o): In function `macioProbe': : undefined reference to `minifind' /usr/lib/gcc/powerpc-unknown-linux-gnu/4.1.1/../../../libkudzu.a(macio.o): In function `macioProbe': : undefined reference to `minifind' collect2: ld returned 1 exit status make: *** [hwsetup] Error 1 !!! ERROR: sys-apps/hwsetup-1.1 failed. Call stack: ebuild.sh, line 1539: Called dyn_compile ebuild.sh, line 939: Called src_compile hwsetup-1.1.ebuild, line 32: Called die wolf31o2 said, that dberkholz played with that stuff...
I did not touch libkudzu in any relevant way. If you have sys-apps/kudzu installed rather than sys-libs/libkudzu, feel free to reassign back to me. (Although I'm not sure how you could, as hwsetup directly depends on libkudzu.)
Nope, sys-libs/libkudzu-1.1.62-r1 has been installed.
Donnie, you're right, you didn't change anything. :) The problem is that this line: emake libkudzu.a ARCH=$(tc-arch-kernel) RPM_OPT_FLAGS="${CFLAGS}" || die Will set ARCH=powerpc on all kernels >= 2.6.16 on ppc, instead of the old behavior returning ppc/ppc64. To fix this, we can either force it to ppc when applicable, or fix the Makefile so that it substitutes ppc for powerpc. What do you want to do?
Created attachment 102507 [details, diff] Fixes the above bug This patch is against the ebuild and fixes the bug. I'll apply it if nobody has any objections...
FIXED... =]
