On Oct 26, 2005, at 11:17 AM, Mark D. Gerl wrote:
No, I haven't. I'm pretty sure that we're OK in the memory stomping department, because our code is well tested in CFM and during this app-init sequence, there's basically no changes between the CFM and MachO code.
That's a dangerous assumption. Compiler changes can easily cause previously non-stomping code to become deadly. Everything is thrown up in the air, even when you just change a few build options. When you're changing compilers and code format, you've got a different flavor of soup altogether!
I'll try MallocDebug, just for kicks... but I don't think it will help.
That's a dangerous attitude. You should be excited to try something already installed on your system, which could potentially track down your bug quickly and without the hassle or delay of waiting for mailing list responses. The response you've gotten here sort of hints at the fact that "Mach-O projects don't just crash themselves."
I also strongly recommend trying "Guard Malloc," which is a much more accurate (and painstakingly slow) tool for detecting overruns. I don't know how easy it is to do outside of Xcode, but it is great for catching "Oh my gosh, I didn't *think* I had any memory stompers!" bugs.
I think it sends the wrong message when you enter a desperate plea for help to a mailing list, and then to follow up with expressions mild dismissal when people offer good suggestions.
Good luck catching this bug!
Daniel
Do not post admin requests to the list. They will be ignored.
