> On 18/10/2016, at 09:35, Steven Trogdon <strog...@d.umn.edu> wrote:
> On Tue, 18 Oct 2016 09:11:59 +1300
> François Bissey <frp.bis...@gmail.com> wrote:
>> On 18/10/16 06:10, v...@ukr.net wrote:
>>> On Mon, 17 Oct 2016 21:55:18 +1300
>>> François Bissey <frp.bis...@gmail.com> wrote:
>>>> OK so I checked on my other machine. liblinbox.so should have been
>>>> linked to (c)blas and so should have matrix_modn_dense_float.so
>>>> independently. Because you didn’t get any complaints from ldd
>>>> on liblinbox.so I’ll assume that it was missing/not properly
>>>> detected when linbox was built. The way matrix_modn_dense_float.so
>>>> is built also suggest that cblas was missing when built.
>>>> What does
>>>> pkg-config —libs cblas
>>>> pkg-config —libs blas
>>>> pkg-config —libs lapack
>>> $ pkg-config --libs cblas
>>> $ pkg-config --libs blas
>>> $ pkg-config --libs lapack
>>> -lreflapack -lblas
>> Smells like a system using the main tree blas/lapack rather
>> than the science overlay. I may have to enforce something
>> there. Anyway those are very strange answers nonetheless.
>> I would expect "-lgslcblas" not "-lgsl". If you have
>> "-lreflapack" what blas are you using if not "refblas"?
>> So does /usr/lib64/libgslcblas.so exist on your system?
>> Is /usr/lib64/libblas.so a dangling symlink (point to nothing)?
> OK, the blas does appear to be perhaps dangling, but the the cblas looks like
> what was reported at https://github.com/cschwan/sage-on-gentoo/issues/429.
Good point Steve! I had forgotten about that. And since no one seems
to care about https://github.com/gentoo-science/sci/pull/638
It is likely to stay that way unless I include it in the sage-on-gentoo
Remember folks, gsl from the science overlay is broken.