On Saturday 14 October 2006 23:57, William Stein wrote: > On Sat, 14 Oct 2006 14:43:55 -0700, David Harvey > > <[EMAIL PROTECTED]> wrote: > > That's pretty impressive! > > > > So it seems if we want to get any faster than that, we need to work > > on the Python object construction stuff. To be honest, I'm a bit sick > > of thinking about that at the moment. It's something we could come > > back to later. I think we have enough work to do implementing more > > widely the stuff we worked on at SAGE days 2. > > I completely agree with this. I really really want to get on with doing > some implementation more widely. > > > Try a sage -t on the whole library with omalloc, and if that works, I > > say patch it in permanently. > > There will be a lot of places to change from Pymem and traditional malloc. > One should go through the whole code base, change them all to use a macro > like sage_malloc, then set that to omalloc, do "make test", and finally > send me a patch.
This will be a slow and painful process (but I'm willing to do it, it might take some time, however). Right now - if I only replace GMP's memory functions SAGE segfaults all over the place. So I have to replace pretty much ever malloc/PyMem_Malloc (*) etc. with sage_malloc and hope that the SEGFAULTs are not caused by a programming error in omalloc (which might only be triggered if it's built as a shared library so the Singular team wouldn't have noticed). (*) e.g. a string generated my mpfr is freed using "free" right now. This should be wrong even for PyMem_Malloc as MPFR builds on GMP and uses it's memory functions IIRC? Martin -- name: Martin Albrecht _pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99 _www: http://www.informatik.uni-bremen.de/~malb _icq: 177334829 _jab: [EMAIL PROTECTED] --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
