I'm on an Athlon machine and I'm constantly running into this problem with all of the xine ebuilds, I hadn't really looked into this too deeply but the fix is trivial and generic for the affected ebuilds. The problems lies in the CFLAGS getting mangled by the configure script for xine resulting in something like: "-mcpu=k6 -march=k6 -O3 [...] -mcpu=i686", which makes the compiler bomb out with: "cc1: -mcpu=i686 does not support -march=k6 make[2]: *** [readline.lo] Error 1 make[2]: Leaving directory `/var/tmp/portage/xine-ui-0.9.10-r1/work/xine-ui-0.9.10/readline'" I'm attaching two patches that should be applied straigth after the source unpack, one for xine-lib and one for xine-ui Let me know if you feel these can be comitted and I'll put them into the ebuilds... Bart
Created attachment 1586 [details, diff] patch for xine-lib
Created attachment 1587 [details, diff] patch for xine-ui
Your CFLAGS are wrong... xine is trying to fix them for you... The xine configure scripts are correctly detecting that you are on a -mcpu=i686 machine... you are incorrectly telling it that you athlon should be treated as a k6... in gentoo, your patch may be applicable, since we expect all users to specify optimization flags of some sort, however my feeling is that we should let xine do it's thing in case some users are not specifying pretty optimization flags for it.
Ok, just argued with carpaski about this one... concensus: he isn't happy about it, but acknowledges that if CHOST and CFLAGS are set consistantly then xine will build fine, my statement about this is that if a user has inconsistenc CFLAGS and CHOST settings, they should be notified as this will cause at the least improper optimization and at worst strange random breakages...
*** Bug 3321 has been marked as a duplicate of this bug. ***
I actually switched to gcc 3.1 now which supports '-march=athlon' which seems to be accepted by xine and gives better results for my Athlon. On the other hand the previous optimization to march=k6 yielded better results for gcc 2.x Since we'll be switching to gcc 3 relatively soon so I guess we could forget about this one. I'll take it as a xine bug to do something I didn't ask it to do.
I'm marking later, maybe later I'll bring it up again and discuss it with people.
well we've just discussed this and while for athlons it is fine, for K6 it b0rks like mad, so I'm going to be adding these patches in.
*** Bug 4391 has been marked as a duplicate of this bug. ***
marf, these patches were all g00fed up, so I'm doing it better... or worse, but more functional... spect new revs of these soon...
fixed in the latest versions of xine-lib and xine-ui in portage, they are masked, please test and let me know what you get.
solves it for me... thx!
