Summary: | >=dev-python/numpy-1.10.4: parallel build issue | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Thomas Deutschmann (RETIRED) <whissi> |
Component: | Current packages | Assignee: | Gentoo Science Related Packages <sci> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | alexander, alonbl, creideiki+gentoo-bugzilla, frp.bissey, hydrapolic, klaus818, luca.chiampo, me, python, viklevin2 |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
URL: | https://github.com/numpy/numpy/issues/13080 | ||
See Also: |
https://bugs.gentoo.org/show_bug.cgi?id=679298 https://bugs.gentoo.org/show_bug.cgi?id=623790 https://github.com/gentoo/gentoo/pull/15315 https://bugs.gentoo.org/show_bug.cgi?id=907718 https://github.com/numpy/numpy/issues/13080 |
||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
build.log
failed build.log of numpy-1.18.3 with MAKEOPTS=-j2 |
Description
Thomas Deutschmann (RETIRED)
2018-07-08 21:45:26 UTC
Is it after a recent upgrade of gcc? Mh, don't think so: Mon Jun 25 17:41:35 2018 >>> dev-python/numpy-1.10.4 Sun Jul 1 21:25:20 2018 >>> dev-python/numpy-1.10.4 Wed Jul 4 01:34:20 2018 >>> dev-python/numpy-1.10.4 Wed Jul 4 01:42:25 2018 >>> dev-python/numpy-1.10.4 Fri Jul 6 23:17:27 2018 >>> dev-python/numpy-1.10.4 Wed Dec 20 23:40:20 2017 >>> sys-devel/gcc-6.4.0 Fri Jan 12 16:51:32 2018 >>> sys-devel/gcc-6.4.0-r1 Sat Mar 3 23:53:15 2018 >>> sys-devel/gcc-6.4.0-r1 Tue Jun 19 13:34:22 2018 >>> sys-devel/gcc-7.3.0-r3 (In reply to Thomas Deutschmann from comment #2) > Mh, don't think so: > > Mon Jun 25 17:41:35 2018 >>> dev-python/numpy-1.10.4 > Sun Jul 1 21:25:20 2018 >>> dev-python/numpy-1.10.4 > Wed Jul 4 01:34:20 2018 >>> dev-python/numpy-1.10.4 > Wed Jul 4 01:42:25 2018 >>> dev-python/numpy-1.10.4 > Fri Jul 6 23:17:27 2018 >>> dev-python/numpy-1.10.4 > > > Wed Dec 20 23:40:20 2017 >>> sys-devel/gcc-6.4.0 > Fri Jan 12 16:51:32 2018 >>> sys-devel/gcc-6.4.0-r1 > Sat Mar 3 23:53:15 2018 >>> sys-devel/gcc-6.4.0-r1 > Tue Jun 19 13:34:22 2018 >>> sys-devel/gcc-7.3.0-r3 OK so that's not change of fortran soname. That would have been so easy. Instead it looks more like the parallel building bug that has been plaguing scipy for a while /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-i686-3.6/numpy/linalg/lapack_lite/python_xerbla.o: file not recognized: File truncated collect2: error: ld returned 1 exit status /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-i686-3.6/numpy/linalg/lapack_lite/python_xerbla.o: file not recognized: File truncated collect2: error: ld returned 1 exit status Try again at "-j1". Yup, looks like a parallel build issue: Building with MAKEOPTS=-j2 successful. (In reply to Thomas Deutschmann from comment #4) > Building with MAKEOPTS=-j2 successful. I can build only with MAKEOPTS=-j1. Got the same error as you with MAKEOPTS=-j2: /usr/lib64/python3.6/distutils/dist.py:261: UserWarning: Unknown distribution option: 'test_suite' warnings.warn(msg) /usr/lib64/python3.6/distutils/dist.py:261: UserWarning: Unknown distribution option: 'define_macros' warnings.warn(msg) error: Command "x86_64-pc-linux-gnu-gcc -O0 -fno-omit-frame-pointer -ggdb -pipe -shared -Wl,-O1 -Wl,--as-needed -shared -O0 -fno-omit-frame-pointer -ggdb -pipe -fno-strict-aliasing /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_litemodule.o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/python_xerbla.o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/zlapack_lite.o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/dlapack_lite.o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/blas_lite.o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/dlamch.o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/f2c_lite.o -L/usr/lib64 -L/var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/temp.linux-x86_64-3.6 -lpython3.6m -o /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/lib/numpy/linalg/lapack_lite.cpython-36m-x86_64-linux-gnu.so" failed with exit status 1 (In reply to Alexander Tsoy from comment #5) > (In reply to Thomas Deutschmann from comment #4) > > Building with MAKEOPTS=-j2 successful. > > I can build only with MAKEOPTS=-j1. Got the same error as you with > MAKEOPTS=-j2: > > /usr/lib64/python3.6/distutils/dist.py:261: UserWarning: Unknown > distribution option: 'test_suite' > warnings.warn(msg) > /usr/lib64/python3.6/distutils/dist.py:261: UserWarning: Unknown > distribution option: 'define_macros' > warnings.warn(msg) > error: Command "x86_64-pc-linux-gnu-gcc -O0 -fno-omit-frame-pointer -ggdb > -pipe -shared -Wl,-O1 -Wl,--as-needed -shared -O0 -fno-omit-frame-pointer > -ggdb -pipe -fno-strict-aliasing > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_litemodule.o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/python_xerbla.o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/zlapack_lite.o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/dlapack_lite.o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/blas_lite.o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/dlamch.o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6/numpy/linalg/lapack_lite/f2c_lite.o -L/usr/lib64 > -L/var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > temp.linux-x86_64-3.6 -lpython3.6m -o > /var/tmp/portage/dev-python/numpy-1.10.4/work/numpy-1.10.4-python3_6/build/ > lib/numpy/linalg/lapack_lite.cpython-36m-x86_64-linux-gnu.so" failed with > exit status 1 Got the same error on ~amd64 with numpy 1.14.5. MAKEOPTS="-j1" emerge numpy works. The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0b5b4b4c8770a61a9eb28ca5fc45ceca65a7ee6d commit 0b5b4b4c8770a61a9eb28ca5fc45ceca65a7ee6d Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2018-09-20 12:50:41 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2018-09-20 12:58:48 +0000 dev-python/numpy: force MAKEOPTS=-j1 Looks like nobody cares. To allow users to build that package in the meanwhile, let's force MAKEOPTS=-j1. Bug: https://bugs.gentoo.org/660754 Package-Manager: Portage-2.3.49, Repoman-2.3.10 dev-python/numpy/numpy-1.10.4.ebuild | 2 ++ dev-python/numpy/numpy-1.13.3.ebuild | 2 ++ dev-python/numpy/numpy-1.14.5.ebuild | 2 ++ dev-python/numpy/numpy-9999.ebuild | 2 ++ 4 files changed, 8 insertions(+) The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=885d5f81e25739a1b088e0d69369d2db992454cc commit 885d5f81e25739a1b088e0d69369d2db992454cc Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2019-03-02 22:57:21 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2019-03-02 22:57:21 +0000 dev-python/numpy: drop parallel build workaround Cannot reproduce with >numpy-1.13.3 anymore. Bug: https://bugs.gentoo.org/660754 Package-Manager: Portage-2.3.62, Repoman-2.3.12 Signed-off-by: Thomas Deutschmann <whissi@gentoo.org> dev-python/numpy/numpy-1.14.5.ebuild | 4 +--- dev-python/numpy/numpy-1.15.4.ebuild | 2 -- dev-python/numpy/numpy-9999.ebuild | 2 -- 3 files changed, 1 insertion(+), 7 deletions(-) *** Bug 679298 has been marked as a duplicate of this bug. *** Looks like this is still happening (bug 679298). But it looks like it is only happening for USE=-lapack. I'll push an updated ebuild shortly which will force MAKEOPTS=-j1 only for USE=-lapack... please test and report. The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=09f3824f30a37092664909b512043b0321e3d14d commit 09f3824f30a37092664909b512043b0321e3d14d Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2019-03-03 16:23:05 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2019-03-03 16:25:31 +0000 dev-python/numpy: re-add parallel make workaround for USE=-lapack Bug: https://bugs.gentoo.org/660754 Package-Manager: Portage-2.3.62, Repoman-2.3.12 Signed-off-by: Thomas Deutschmann <whissi@gentoo.org> dev-python/numpy/numpy-1.13.3.ebuild | 4 ++-- dev-python/numpy/numpy-1.14.5.ebuild | 2 ++ dev-python/numpy/numpy-1.15.4.ebuild | 2 ++ dev-python/numpy/numpy-1.16.1.ebuild | 2 ++ dev-python/numpy/numpy-9999.ebuild | 2 ++ 5 files changed, 10 insertions(+), 2 deletions(-) The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=d5337dedc68adc4dfcbfe8bb045c96554512aabf commit d5337dedc68adc4dfcbfe8bb045c96554512aabf Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2019-03-03 16:29:22 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2019-03-03 16:29:22 +0000 dev-python/numpy: force MAKEOPTS=-j1 for all USE configurations ...even failing for USE=lapack on some systems. Bug: https://bugs.gentoo.org/660754 Package-Manager: Portage-2.3.62, Repoman-2.3.12 Signed-off-by: Thomas Deutschmann <whissi@gentoo.org> dev-python/numpy/numpy-1.13.3.ebuild | 2 +- dev-python/numpy/numpy-1.14.5.ebuild | 2 +- dev-python/numpy/numpy-1.15.4.ebuild | 2 +- dev-python/numpy/numpy-1.16.1.ebuild | 2 +- dev-python/numpy/numpy-9999.ebuild | 2 +- 5 files changed, 5 insertions(+), 5 deletions(-) It seems the dev-python/numpy-1.18.3 ebuild removed the forcing of MAKEOPTS=-j1, so now I'm seeing the same build failures again. When manually specifying MAKEOPTS=-j1 to 1.18.3 it builds successfully. *** Bug 623790 has been marked as a duplicate of this bug. *** Created attachment 634836 [details]
failed build.log of numpy-1.18.3 with MAKEOPTS=-j2
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0d784579f7d2fa66ab44c80e2ff0fba353f08dbf commit 0d784579f7d2fa66ab44c80e2ff0fba353f08dbf Author: Benda Xu <heroxbd@gentoo.org> AuthorDate: 2020-04-27 10:03:44 +0000 Commit: Benda Xu <heroxbd@gentoo.org> CommitDate: 2020-04-27 10:06:39 +0000 dev-python/numpy: fix regressions introduced by 48dd16c7839fd. Bug: https://github.com/gentoo/gentoo/pull/15315 Closes: https://bugs.gentoo.org/660754 Package-Manager: Portage-2.3.88, Repoman-2.3.18 Signed-off-by: Benda Xu <heroxbd@gentoo.org> dev-python/numpy/numpy-1.18.3.ebuild | 17 +++++------------ profiles/arch/powerpc/package.use.mask | 4 ++++ profiles/arch/sparc/package.use.mask | 4 ++++ 3 files changed, 13 insertions(+), 12 deletions(-) Scipy is in process of switching to meson: https://github.com/scipy/scipy/issues/7112 https://github.com/scipy/scipy/issues/13615 and numpy & al will probably follow suit. With some luck, in a few years this limitation will be gone! |