Re: Segmentation violation in GC - infinite recursion

David> I've gotten two segmentation violations in the garbage David> collector. It appears to be the result of an infinite David> recursion in the mark phase. The stack trace is from the David> first occurence in an XEmacs that had been running for a David> few days doing language editing plus running VM. The David> second occured within a few minutes. All that had happened David> was an edit of a text file, a few file insertions, a few David> pastes from the clipboard plus starting VM. The crash David> happened while VM was processing an HTML formatted message. Michael> Hm, to me this doesn't look like infinite recursion: I Michael> don't see any duplicate Lisp objects along the way. I Michael> suspect this is simply a stack overflow. It's certainly Michael> likely that the current, temporary kkcc marker consumes Michael> more stack than the old one. (The plan is to replace it Michael> by one that doesn't use any stack at all.) What puzzles me is that there are about 10000 stack frames of just marker objects without interspersed objects of other types. But I haven't had the time to familiarize myself with the creation of these objects. How soon do you expect the next phase of the new garbage collector implementation? for now, I've quit using it because of the frequent segmentation faults. I'll give it a try with the GC threshold cut in half. david