On Mon, Oct 7, 2019 at 6:00 PM E. Madison Bray <erik.m.b...@gmail.com> wrote: > > On Mon, Oct 7, 2019 at 5:57 PM E. Madison Bray <erik.m.b...@gmail.com> wrote: > > > > When building from scratch on Cygwin I am getting some errors related > > to OMP-related symbols being missing when linking a Sage module that > > uses fflas-ffpack; specifically sage.libs.linbox_flint_interface: > > > > [sagelib-8.9] g++ -shared -Wl,--enable-auto-image-base > > -L/opt/sagemath-8.9/local/lib -Wl,-rpath,/opt/sagemath-8.9/local/lib > > -L/opt/sagemath-8.9/local/lib -Wl,-rpath,/opt/sagemath-8.9/local/lib > > build/temp.cygwin-3.0.7-x86_64-2.7/build/cythonized/sage/libs/linbox/linbox_flint_interface.o > > -L/opt/sagemath-8.9/local/lib > > -L/opt/sagemath-8.9/local/lib/python2.7/config > > -L/opt/sagemath-8.9/local/lib -llinbox -lntl -liml -lfflas -lffpack > > -lgivaro -lblas -lflint -lmpfr -lgmp -lgmpxx -lstdc++ -lpython2.7 -o > > build/lib.cygwin-3.0.7-x86_64-2.7/sage/libs/linbox/linbox_flint_interface.dll > > [sagelib-8.9] > > build/temp.cygwin-3.0.7-x86_64-2.7/build/cythonized/sage/libs/linbox/linbox_flint_interface.o: > > In function `FFPACK::rns_double::init(unsigned long, unsigned long, > > double*, unsigned long, Givaro::Integer const*, unsigned long, > > unsigned long, bool) const [clone ._omp_fn.0]': > > [sagelib-8.9] > > /opt/sagemath-8.9/local/include/fflas-ffpack/field/rns-double.inl:95: > > undefined reference to `GOMP_loop_ull_runtime_start' > > > > it goes on like that for several related errors. I've never seen this > > one before. This is with gcc 7.4.0. > > > > This was doing a build from scratch for a release, in a clean Cygwin > > install, so I'm not sure where the difference is. One thing I can > > see, comparing to a previous build log of my development sage from one > > of the 8.9 release candidates (with fflas-ffpack 2.4.3) that that > > module was compiled *without* -fopenmp, whereas on my clean build it > > is passing -fopenmp and I'm not sure where the difference is coming > > from. > > Ahah, I can also see that in my development build, configuring > fflas-ffpack output: > > checking for OpenMP... no > > whereas on my release build it has > > checking for OpenMP... yes > > > For now I don't need or care about OpenMP support, as I have not > tested that at all on Windows. I just need to figure out why that > flag is being forcibly set...
It appears we used to configure fflas-ffpack with --disable-openmp, however this ticket removed that: https://trac.sagemath.org/ticket/27444 without bumping the package patch level. It also didn't remove the comment about disabling OpenMP support, even though it no longer disables it. I'm not sure it make sense to make that the default; perhaps there should be a global config flag for whether or not Sage and its dependencies should be built with OpenMP support. > > On Mon, Sep 30, 2019 at 1:19 AM Volker Braun <vbraun.n...@gmail.com> wrote: > > > > > > The "master" git branch has been updated to Sage-8.9. As always, you can > > > get the latest beta version from the "develop" git branch. Alternatively, > > > the self-contained source tarball is at > > > http://www.sagemath.org/download-latest.html > > > > > > There was no change over 8.9.rc1 > > > > > > -- > > > You received this message because you are subscribed to the Google Groups > > > "sage-release" group. > > > To unsubscribe from this group and stop receiving emails from it, send an > > > email to sage-release+unsubscr...@googlegroups.com. > > > To view this discussion on the web visit > > > https://groups.google.com/d/msgid/sage-release/ea48d604-eff9-4b8c-ab44-a0929a7e99f9%40googlegroups.com. -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/CAOTD34aDCyYgi6Q6aD1BixY0g-cf1G_1J8sZcbu58YmwcuHsYw%40mail.gmail.com.
