On Friday, November 30, 2012 5:21:24 PM UTC+1, Simon King wrote: > > Hi! > > I don't know if the following is related with the recently discussed > problems in the graph backend or whether it is from one of my dreaded > "fix memleaks by weak caching" patches - can experts please have a > look at #13779 ? > > The problem: If one does "export MALLOC_CHECK_=3", then the tests for > sage/graphs/graph_generators.py crash because free() is applied to an > invalid pointer - at least that's what is happening on my openSuse laptop, > both with sage-5.5.rc0 and an older (but patched) version. > > Speaking of MALLOC_CHECK_: > > Based on work of Nils Bruin and with the help of Hans Schönemann, > I created a new Singular spkg at #13731 (not yet totally ready for > review) that contains two upstream bug fixes and moreover provides the > option to replace its memory manager "omalloc" by a thin compatibility > layer on top of malloc, called "xalloc". > > *Without* the two upstream fixes and with xalloc, even fairly simple > computations would crash with MALLOC_CHECK_=3 - and the point of xalloc > is that it makes debugging easier. > > *With* the two upstream fixes and xalloc, running the full Sage test > suite under MALLOC_CHECK_=3 crashes only once, namely in the afore > mentioned graph_generators test. > > In particular, even when using a memory manager for Singular that should > immediately complain upon a corruption, #715 and #11521 (contained in > sage-5.5.rc0) are responsible for at most one crash (at least on > openSuse). So, I guess the problems with libsingular and weak caching > came from later tickets (#12215 and #12313, for example). > > And this is the good news of my post... > > singular-3-1-5.p2.spkg from #13731 will soon be ready for review, if > someone wants to have a look. > > Best regards, > Simon > > I think the original graph_generators problem is unrelated to singular, caching, etc. and has been fixed here: http://trac.sagemath.org/sage_trac/ticket/13719 isn't it?
Anyway, that is really good news! -- 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.
