How exactly do you call ./configure?
CC=clang CXX=clang++ ./configure <more options>
produces a working build on some systems at least.
In principle, gcc and clang are producing compatible binaries, but YMMV.
Not so for g++ and clang++, mixing them up is likely to fail. There's also
an extra complication involving clang's libc++ (its use depends upon your
configuration)
vs gcc's libstdc++
On Tue, 17 Aug 2021, 02:15 Kazuyoshi Furutaka, <furutaka.kazuyo...@gmail.com>
wrote:
>
> Dear sage experts...
>
> I'm sorry again for not showing the fix and only to report....
>
> As the subject says, when configured with CC=clang and CXX=clang++, the
> build of dochtml failed.
>
> > ***************************************************************
> > Error building Sage.
> >
> > The following package(s) may have failed to build (not necessarily
> during this run of 'make all-start'):
> >
> > * documentation: dochtml
> > last build time: Aug 13 11:37
> > log file:
> /home/furutaka/work/sage/sage-9.4.rc2-git-clang/logs/pkgs/../dochtml.log
>
> Looking at the log file, ImportError occured many times.
>
> > $ grep ImportError
> /home/furutaka/work/sage/sage-9.4.rc2-git-clang/logs/pkgs/../dochtml.log |
> sort | uniq
> > [combinat ] ImportError:
> /home/furutaka/work/sage/sage-9.4.rc2-git-clang/local/lib64/python3.9/site-packages/sage/geometry/polyhedron/combinatorial_polyhedron/
> face_iterator.cpython-39-x86_64-linux-gnu.so: undefined symbol:
> __kmpc_end_critical
>
> Then, I searched `face_iterator.cpython-39' in the log of entire make, and
> found that gcc was used to build the shared object even though it was
> configured with CC=clang...
>
> > [sagelib-9.4.rc2] gcc -pthread -shared -Wl,-z,relro -Wl,--as-needed
> -Wl,-z,now -g -Wl,-z,relro -Wl,--as-needed -Wl,-z,now -g
> -Wl,-rpath-link,/home/furutaka/work/sage/sage-9.4.rc2-git-clang/local/lib
> -L/home/furutaka/work/sage/sage-9.4.rc2-git-clang/local/lib
> -Wl,-rpath,/home/furutaka/work/sage/sage-9.4.rc2-git-clang/local/lib -O2 -g
> -march=native
> build/temp.linux-x86_64-3.9/build/cythonized/sage/geometry/polyhedron/combinatorial_polyhedron/face_iterator.o
> -L/usr/lib64 -lgmp -o
> build/lib.linux-x86_64-3.9/sage/geometry/polyhedron/combinatorial_polyhedron/
> face_iterator.cpython-39-x86_64-linux-gnu.so -fopenmp -lpari
>
> Searching for the use of "gcc" in the log file, it seems that gcc was used
> to build the folloiwng packages.
>
> > $ awk '$2=="gcc"{print $1}' make.log | sort | uniq
> > [cypari-2.1.2]
> > [cysignals-1.10.3]
> > [cython-0.29.21]
> > [gmpy2-2.1.0b5]
> > [markupsafe-1.1.1]
> > [matplotlib-3.3.4]
> > [memory_allocator-0.1.0]
> > [numpy-1.20.3]
> > [pillow-8.1.2]
> > [psutil-5.2.0.p2]
> > [pyzmq-22.0.3]
> > [sagelib-9.4.rc2]
> > [scipy-1.6.3]
> > [singular-4.2.0p3]
> > [tornado-6.0.4]
>
> Is the use of gcc normal/correct/intended when CC=clang?
>
> Thanks.
> Kazuyoshi
>
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sage-devel/08763baa-fa57-4c4f-b190-86bec21b1231n%40googlegroups.com
> <https://groups.google.com/d/msgid/sage-devel/08763baa-fa57-4c4f-b190-86bec21b1231n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/sage-devel/CAAWYfq0HOHqr5AVA_nd44sbN6QGr25bij1RfWYZAhkSsLjn-dQ%40mail.gmail.com.
