On Nov 15, 11:59 pm, Jeroen Demeyer <jdeme...@cage.ugent.be> wrote: > Could you try again with sage-5.5.beta1?
Same behaviour. Was there a reason to expect differently? I guess something is different on sextus. Bad memory/other hardware problems? I was surprised by how little issues arose from inserting garbage collections between all doctests. That should upset the memory usage patterns so much that I would expect it to shake out many problems. Only things like singular's omalloc would be immune, because it hides alloc/dealloc operations from the OS. You really need to wait for an actual corruption to see a problem. The guarded malloc experiment on OSX and similar operations took care of that. See http://trac.sagemath.org/sage_trac/ticket/13447 for a dirty singular package that switches out omalloc for a system malloc, which then allows normal OS tools to check memory allocation/ access/deallocation. See also the ticket for notes on how the approach taken there can be adapted to let Singular use the system malloc under linux (one singular malloc routine needs to know the size of an allocated block, which is a non-POSIX malloc feature that both OSX and linux support in different ways). Do we have other memory managers in sage that play tricks like omalloc? Things run a lot slower when you switch back to system malloc for these, but it does enable conventional memory sanitation tests. Valgrind produces way too much warnings to be useful. All you want is a segfault on any access-after-dealloc or double-dealloc (out-of- bounds access would be nice too). OSX's libgmalloc is perfect for that. Is there a linux equivalent (or a way to configure valgrind to do just this)? I pose it as a challenge that no-one is able to do a comprehensive testing of memory alloc/dealloc in sage. Even though I outline the exact approach above that would make it a relatively straightforward process to go through, no-one has the stamina and heroic hacker skills to pull it off. Prove me wrong! -- You received this message because you are subscribed to the Google Groups "sage-devel" group. 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. Visit this group at http://groups.google.com/group/sage-devel?hl=en.
