On Jan 29, 5:15 pm, mabshoff <mabsh...@googlemail.com> wrote:
> On Jan 29, 3:17 pm, William Stein <wst...@gmail.com> wrote:
>
> <SNIP>
>
> > Built and fails spectacularly:
> > * Fedora 64 bit, with dozens of failures like this (about 1 in 6
> > doctest files):
> > sage -t "devel/sage/sage/misc/all.py"
> > A mysterious error (perphaps a memory error?) occurred, which may have
> > crashed doctest.
>
> I think I tracked this down to
>
> http://bugs.gentoo.org/show_bug.cgi?id=218378
>
> which leads to
>
> https://bugzilla.redhat.com/show_bug.cgi?id=451494
>
> and then
>
> http://bugs.python.org/issue3115
>
> It fits the problem well:
>
> * open in Python 2.5.2 which we ship
> * it only happens on 64 bits since Jaap [our endless 32 bit Fedora
> tester - thank $DEITY we have him :)] has never seen it on 32 bit
> Fedora 8/9/10 in the last couple years
> * we have seen it on Fedora 9 64 bit boxen - it went away by
> downgrading the kernel to some FC8 kernel
Unfortunately this wasn't it. My next suspect is libSingular, but when
using malloc instead of mmap it all blows up on exit every time with a
double free since we end up calling the wrong dtor in a bunch of
cases. The same problem exists on OSX in 64 bit mode (where we are
forced to use --with-malloc=system for libSingular to even work), so
this bug needs to b e fixed either way. Unfortunately I have no clue
how to approach this at this point.
> I meant to update to Python 2.5.4 in alpha4 anyway, so I will see if
> the problem cannot be made to go away.
Upgrading Python is still a potential way out of this. Aside from the
above issues I also suspect prelinking, but have not tried to disable
it. Either way, if the issue goes away with disabling prelinking it
isn't very useful anyway since disabling it requires root priviledges
and I have been unable so far to find a way to disable it on a per
program basis.
> Cheers,
>
> Michael
Cheers,
Michael
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
