On Feb 24, 12:10 am, "Georg S. Weber" <georgswe...@googlemail.com>
wrote:
> On 23 Feb., 18:53, mabshoff <mabsh...@googlemail.com> wrote:
Hi Gerog,
> Hi Michael,
>
> I could not put that much effort into it yesterday evening, but one
> overnight complete build & testlong of Sage 3.3 with a
> "singular...p3a.spkg" dropped in instead of the original "p3" one did
> go through flawlessly. So either the #4181 patch (malloc=system) made
> the minpoly issue go away again if it had been introduced by the #5344
> patch (remove potential "four double free" problem), or the minpoly
> issue is not introduced by it resp. does not occur at all with my
> setting (32 Bit OS X 10.4.11).
All is good and you got positive reviews for
#4181: Georg Weber: Singular/omalloc: do use the system's malloc
#5344: Georg Weber: Singular/omalloc "double free" problem
and hence by extension
#3760: creating 666 rings in singular fails with an out of memory
error on 32-bit intel os x
You Singular.spkg fix did also partially resolve
#5278: On modern fedora 64 installs, sage exists frequently when
omalloc thinks there is no memory left
but there was another issue I found a fix for in the Python tracker.
The fixed python.spkg isn't uploaded yet, but it will be there in a
couple hours.
The "minpoly" issue is still there on 64 bit OSX, but is it unrelated
to your work, i.e. the double frees also disappear on OSX 64 bit, but
the segfaults remain. They are caused inside oMalloc, but as Carl
Witty pointed out right above in
http://groups.google.com/group/sage-devel/msg/4ea57c8579828bce
the might likely be caused by timing issues with garbage collection,
etc. This will probably remain unsolved for 3.4, but it seems like a
great lead to get one of the final 64 bit OSX bugs fixed.
In total I spend 16 hours reviewing and testing your patches. I also
did valgrind all of Sage's doctests with your Singular.spkg (thank the
NSF again for giving us 24 core systems since I just used 18 of them
to do this in a couple hours), so I am signing off on all of your
fixes and will merge them in a couple hours in 3.4.alpha0.
Sorry again for giving you a hard time for the review, but in the end
it made me work on the review very hard to still poke a hole into your
patches and I failed. But I am glad I failed :). Given that this is
rather critical infrastructure work for a very important Sage release
I wanted to be extra careful.
> Cheers,
> gsw
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
-~----------~----~----~----~------~----~------~--~---
