Robert Moynihan a écrit : > Luc Le Blanc wrote: > > >I finally managed to fix the bug by simply removing the line that would > >cause the crash (...) It turned out to be redundant, so no functionnality > >is lost and everything else works fine beyond that. I still can't explain > >what was wrong, nor why it only crashes on a T3 (I suspected color, but I > >had no problem on a color Clié) nor why Simulator cannot debug this but > >only crash. But as I have many other things I can worry about, I'll just > >enjoy this magic ending. > > > > > I would feel uncomfortable with that ending. If you have been having > crashes, and you can't explain why they have disappeared, then the bugs > are most likely just hiding, and will start acting up again when the > time is right. You mentioned that other things were acting wrong > (button mapping) and the simuator was crashing. This sounds like your > app is overwritting some of its memory bounds, and walking all over > other memory locations where it shouldn't. Such errors do not > necessarily cause a crash, depending on what memory is being > overwritten, and what you are mistakenly writing there. But there > almost surely is a bug that needs squashing. As Roger (I think) > suggested, you should use a debug version of the emulator or sim, as > these catch more stuff. In my opinion, the Treo650 sim seem to be > particularly unforgiving (as is the Tapwave), and I use that a lot. Its > tendency to just pack up ship and disappear when it runs across a bug > really is annoying, but you get used to that. When I'm chasing a bug > that is too-often causing the sim to close, I might switch to a 4.1rom > emulator, so that I can at least reset it without killing and > reloading. My 3cents worth. > > Since you seem to have resolved it by removing one redundant line of > code, perhaps you would tell us what that line said. Maybe there > actually IS a reason why the crash stopped. > > Bob
Thanks Bob. I too feel uncomfortable about this crash disappearing as if by magic, and I'd sure appreciate tools that detect a memory overwrite right as it happens. The line I removed was just calling FrmShowObject for a Gadget. The gadget exists (with its own handler) and it's already displayed from another routine using the same code (a remnant from previous modifications). -- Luc Le Blanc -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
