On Thursday 08 May 2008 04:39:57 NotFound wrote: > Last night in #parrot particle informed of a bug in a msvc++ build. By > his suggestion, me and others were able to reproduce it in other > platforms by using the gcdebug core: > > ./parrot --runcore=gcdebug t/pmc/orderedhash_9.pasm > > Looks like the problem is that parrot_mark_hash in hash.c expects that > buckets always contains a valid key, but DESCRIPTION in > orderedhash.pmc says: > > C<delete_keyed> never removes items; they are just nulled.
That to me seems like the real problem. Did anyone experiment with using hash_delete? > The problem can be fixed by checking in parrot_mark_hash that the key > is no null before calling pobject_lives. The attached path does it. That seems like treating the symptom to me. I'd like to see if really deleting the item works, first. -- c