=dev-lang/polyml-5.4.1 appears to work with sys-libs/glibc-2.11.3. I suspect that it works with earlier versions too. Also, I can build, run and use on Gentoo Prefix on Mac OS X, so the true dependency seems to be on virtual/libc, which is something that is guaranteed by the @system set. Is there any particular reason why 'threads? ( >=sys-libs/glibc-2.13 )' was chosen as a dependency?
> that it works with earlier versions too. Also, I can build, run and use on > Gentoo Prefix on Mac OS X, so the true dependency seems to be on virtual/libc, That should have been "... I can build, run and use PolyML 5.4.1 on Gentoo Prefix on Mac OS X ...".
(In reply to comment #0) > =dev-lang/polyml-5.4.1 appears to work with sys-libs/glibc-2.11.3. I suspect > that it works with earlier versions too. Also, I can build, run and use on > Gentoo Prefix on Mac OS X, so the true dependency seems to be on > virtual/libc, which is something that is guaranteed by the @system set. > > Is there any particular reason why 'threads? ( >=sys-libs/glibc-2.13 )' was > chosen as a dependency? There were 2 reasons why threads? ( >=sys-libs/glibc-2.13 ) was chosen as a dependency: (1) when built with USE=threads on arch amd64 /lib64/libpthread.so.0 is listed as a NEEDED library and it belongs to the sys-libs/glibc package $ scanelf -n /usr/lib64/libpolyml.so.3.0.1 TYPE NEEDED FILE ET_DYN libpthread.so.0,libgmp.so.10,libXm.so.4,libXt.so.6,libX11.so.6,libdl.so.2,libstdc++.so.6,libm.so.6,libc.so.6,libgcc_s.so.1 /usr/lib64/libpolyml.so.3.0.1 $ ldd /usr/lib64/libpolyml.so.3.0.1 | grep libpthread.so.0 libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f7d1a39a000) argus (ghclive) ~ $ fquery b /lib64/libpthread.so.0 sys-libs/glibc-2.15-r2 (/lib64/libpthread.so.0 -> libpthread-2.15.so) $ (2) version >=sys-libs/glibc-2.13 was chosen as I tested polyml with that version, and have since tested it with later versions. I tested your fix: https://bugs.gentoo.org/show_bug.cgi?id=395009#c3 this seems to work fine on arch amd64 (builds the library with the same deps, tests pass with USE=threads). So I committed this fix, thanks. Sorry for the delay (I didn't notice earlier, found it now when I searched for polyml bugs because I was looking at bumping it to 5.5, but 5.5 fails a test, asked upstream about that).
