It seems that gcc-3.4.2-r1 have a bad slot, -3.4 instead of 3.4. Output from etcat -v gcc: [ I] 3.3.4-r1 (3.2) [ ~ ] 3.4.1 (3.4) [ ~I] 3.4.1-r2 (3.4) [ ~ ] 3.4.2-r1 (-3.4) Infortunately i can't figure how the slot is set in the toolchain eclass. Reproducible: Always Steps to Reproduce: 1. 2. 3.
well that's embarrassing...
I have a patch to the toolchain eclass to solve the problem (a better solution than hardcoding the slot).
Created attachment 39632 [details, diff] patch to correct the calculation of the slot in the eclass This correct the calculation of the slot if $CCHOST isn't defined.
indeed
You forgot to remove the slot in the ebuild. I don't know if this is important since the ebuild is masked.
i'm not touching that ebuild again until: 1) we can figure out those sse2 problems which are screwing things up on x86 and amd64 with multilib enabled 2) we can figure out why 3.4.2 HATES ppc 3) fix some missing and incorrectly placed .so files 4) do yet -another- rewrite for the needs of gentoo-embedded who want to split out libgcc into it's own ebuild 5) probably get rid of chucking files into a versioned directory since we're splitting out the libraries anyways. 6) make sure i can do all of the above without breaking gcc-config in summary, gcc hates me with the burning passion of a million dying stars. but yes, i was going to eventually get to the ebuild. i only committed the toolchain.eclass change right now because i realised if i didnt i'd forget, and it would never get done. ^^;
closed
