I have been debugging the infamous EXC_BAD_ACCESS error for a few days now. NSZombieEnabled = YES did not offer anything. The call stack was different everytime I received the error, which was once every 5 or 6 runs.
I saw a tip for enabling guard malloc (which is in the scheme editor now for Xcode 4) on Lou Franco's website: Understanding EXC_BAD_ACCESS. As soon as I did this, my program halted on the exact line that was causing this elusive error.
According to its description, guard malloc creates separate pages for every malloc and deletes the whole page when the memory is freed, thus crashing the program when the freed memory is accessed. For general development, why wouldn't I just keep guard malloc on all the time? It seems to catch certain types of memory errors easily. If I'm not testing memory management or performance specifically, is there some downside to using it?
Not only does it waste address space, but it will significantly slow down your program (potentially to the point where it's unusable, even on the simulator). I suppose for an iOS programme when you're running it on the simulator it's a bit moot (memory isn't a problem, and the performance hit isn't terrible either), but perhaps in the name of best practice you shouldn't run it constantly.