Summary: | sys-devel/gcc add USE for libquad installation instead of hiding behind USE=fortran (was: sci-libs/fftw[quad]: missing dependency for libquadmath) | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Fredric Johansson <fredric.miscmail> |
Component: | Current packages | Assignee: | Gentoo Toolchain Maintainers <toolchain> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | ajak, aladjev.andrew, jlec, sam, sandino, sci, slyfox |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
See Also: |
https://github.com/gentoo/gentoo/pull/9754 https://bugs.gentoo.org/show_bug.cgi?id=734820 |
||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
build.log of 3.3.8-r1 on sparc
config.logs |
Description
Fredric Johansson
2017-02-11 19:11:33 UTC
Could you please add a build.log and the output of "emerge --info gcc"? Looking at toolchain.eclass: # newer gcc's come with libquadmath, but only fortran uses # it, so auto punt it when we don't care if tc_version_is_at_least 4.6 && ! is_fortran ; then confgcc+=( --disable-libquadmath ) fi so I guess fftw[quad] depends on gcc[fortran], even though the above comment is actually wrong, there are packages using quad without fortran. Thanks Chrostoph digging this out The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=461721fc58ad27051c1a3108f62a0e431288160e commit 461721fc58ad27051c1a3108f62a0e431288160e Author: Conrad Kostecki <conrad@kostecki.com> AuthorDate: 2018-09-01 13:49:27 +0000 Commit: Andrew Savchenko <bircoph@gentoo.org> CommitDate: 2018-09-01 22:23:44 +0000 sci-libs/fftw: bump to version 3.3.8 Closes: https://bugs.gentoo.org/651028 Closes: https://bugs.gentoo.org/608990 Package-Manager: Portage-2.3.48, Repoman-2.3.10 sci-libs/fftw/Manifest | 1 + sci-libs/fftw/fftw-3.3.8.ebuild | 181 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 182 insertions(+) I think that GCC maintainer will still be interested in fixing this properly in GCC (toolchain.eclass). Either a new USE flag should be added to GCC, or libquadmath should be installed unconditionally. Yeah. I think unconditional libquadmath install should be fine. Created attachment 650674 [details]
build.log of 3.3.8-r1 on sparc
The problem is still there with 3.3.8-r1
Created attachment 650676 [details]
config.logs
Still getting the same with gcc-9.3.0-r1[fortran] and fftw-3.3.8-r1 on arm (In reply to John Helmert III (ajak) from comment #9) > Still getting the same with gcc-9.3.0-r1[fortran] and fftw-3.3.8-r1 on arm Please file a separate bug. This bug is about missing support library in gcc. The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=a06e1246568c3c6043de3ae6a5042e5470be1fa8 commit a06e1246568c3c6043de3ae6a5042e5470be1fa8 Author: Sergei Trofimovich <slyfox@gentoo.org> AuthorDate: 2020-07-30 06:51:44 +0000 Commit: Sergei Trofimovich <slyfox@gentoo.org> CommitDate: 2020-07-30 06:54:52 +0000 toolchain.eclass: install libquadmath by default libquadmath is a small support runtime library for _Float128 types. sci-libs/fftw happens to use it and expects it to be provided by gcc. Let's provide it unconditionally. Reported-by: Fredric Johansson Closes: https://bugs.gentoo.org/608990 Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> eclass/toolchain.eclass | 6 ------ 1 file changed, 6 deletions(-) Please do not fix in that way, libquadmath cross compilation is broken. |