Ah, I must have hit enter at the same time. Cool to read your way of expressing it, cwright.
Best, gt On Wed, Nov 16, 2011 at 8:25 PM, Christopher Wright < christopher_wri...@apple.com> wrote: > Thanks Chris that was it. > > How to know which item in Crash log is the offending item? > > > secret crash log decoder ring. > > The top part tells us you're on 10.5.8, a PPC machine. > > The next part tells us that the bad address was 0x4 (basically a NULL > dereference - 4 is sufficiently close to 0 that it's probably trying to > dereference the second field of a structure that failed to get allocated), > and that Thread 0 was where the exciting stuff happened. > > Thread 0's backtrace says that a QCView was getting ordered on screen > (document was opened, QC was launched, or you opened the viewer for an open > editor), and that QCRenderer's renderAtTime:arguments: method crashed. > This method has executed for every single frame of every single > composition that has ever been rendered in the history of Quartz Composer, > so it's probably not broken. So we look at the next line (called a > "frame", as in "stack frame"): something undefined from > net.kineme.CompositionLoader. That seems like a plausible candidate for > possibly being broken. > > Therefore, The Hypothesis: By not asking the potentially broken piece > (whatever that plugin is) to do a potentially broken thing (doing something > with an invalid object/pointer), you probably won't end up breaking - cheap > and easy to test, too. Seemed to work in this case. Didn't fix the > problem (that's a third party's problem), but it safely worked around it > (by not causing it to execute). > > Further up (frame 18 of Thread 0), there's another plugin, which would be > another plausible candidate for removal if the above didn't work. > > And then at the bottom, there's a number of plugins that could be > removed/disabled to test (if removing the above two wouldn't solve it). > If, after removing all plugins, the composition crashed on launch, _then_ > and only then would we have an interesting crash that I'd be interested in > seeing the composition for (because it would indicate a malformed > composition, or a bug in composition loading or something along those > lines). > > Hope that helps! > > -- > Christopher Wright > christopher_wri...@apple.com > > > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) > Help/Unsubscribe/Update your Subscription: > > http://lists.apple.com/mailman/options/quartzcomposer-dev/gtoledo3%40gmail.com > > This email sent to gtole...@gmail.com > >
_______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
