In http://trac.sagemath.org/ticket/15045, we noticed a problem with the
ATLAS build process involving shared libraries. It's a problem which
doesn't occur often, it happens essentially randomly (but mostly/only on
AMD processors). It does not happen with a statically linked ATLAS. It
has been reported upstream, but upstream doesn't react.
I see 3 concrete solutions:
(0) revert to ATLAS-3.8.4.
(1) build static libraries instead of shared libraries.
(2) try to build shared libraries, fall back to static libraries if that
fails.
Volker Braun favours (2), I favour (1), so I'm hoping for some
interesting feedback here from sage-devel. My argument against (2) is
that it creates an extra special case in the build and even a
non-deterministic special case. Volker's argument in favour of (2) is
that shared libraries are better than static libraries (which I agree
with by the way) and that it will make future developments easier.
Jeroen.
--
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 post to this group, send email to sage-devel@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/groups/opt_out.
