On Mon, Nov 5, 2012 at 3:25 PM, Simon King <simon.k...@uni-jena.de> wrote: > Hi Robert, > > On 2012-11-05, Robert Bradshaw <rober...@gmail.com> wrote: >> +1. I've always been meaning to get back to this for ages, but just >> haven't found the time. If we're going to make a big push to get this >> in, I'll do what I can to help. > > I'd appreciate your support! > >> For testing, I would propose we manually insert gc operations >> periodically to see if we can reproduce the failures more frequently. > > How can one insert gc operations? You mean, by inserting gc.collect() > into doctests, or by manipulating the Python call hook?
I was thinking about inserting it into the doctesting code, e.g. with a random (know seen) x% chance between any two statements. >> We could then marks some (hopefully a very small number) parents as >> "unsafe to garbage collect" and go forward with this patch, holding >> hard references to all "unsafe" parents to look into them later (which >> isn't a regression). > > That actually was what we tried: There was some bug that has only > occurred on bsd.math, and could be fixed by keeping a strong cache for > polynomial rings (which is inacceptable for my own project, but which > is at least no regression). > > Anyway. I did not look into the new problems yet. If it is (again) about > libsingular polynomial rings, then I think we should really make an > effort to get reference counting for libsingular rings right. True, but I'd rather no particular ring hold us back from getting the general fix in. - Robert -- 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.
