Poof! Where’d My App Go?
Posted May 21st, 2004 @ 07:56am by Erik J. Barzeski
A few people have written in and told us that PulpFiction sometimes quits on them without leaving the little crash log window with the "Submit to Apple" button (it'd be nice if developers could rewire that to send the crash log to them). In other words, the app goes poof, disappears from the screen, and that's that.
I'm not certain what could cause this - we'd like to try to track it down, but nobody has provided any real steps to cause this to happen - and I'm not certain what circumstances could cause the crash reporter to fail to pop up on the screen.
Does anyone have any ideas? I'm primarily looking for developers who have seen this problem and have useful information for tracking this down, not "me too" type repsonses.
Donate Life
Posted 21 May 2004 at 8:10am #
Happens to me too... I keep it running, go away, and next thing you know, it simply isn't running any more - without any further notice. Obviously, I wouldn't know how to reproduce that since I didn't know anything *actively* that would cause it...
Posted 21 May 2004 at 10:09am #
Would if be helpful if we submitted the stack trace that is part of the Apple crash report to you, or do you need the (non-existant?) core file?
FWIW, I've noticed that the open-source Adium (adium.sf.net) pops up it's own crash-submitting dialog that pre-empts Apple's widget whenever something goes TERRIBLY WRONG.
Posted 21 May 2004 at 10:21am #
Well, happens to me too! 😉
Seriously, though, it does always seem to happen to me as the result of some action, IIRC when selecting a post to view.
Unfortunately, I have no posts that definitely cause it, next time it happens, I'll try to remember which caused it. But I do know that the same posts do not always cause it to poof. Generally when I reopen PF, the post that poofed it will open fine. So I'm not sure if it would be reproducable.
(I'm using ForwardDirectly.css if that may be a cause?)
Posted 21 May 2004 at 10:43am #
It's happened to me a few times, but for reasons that I can't explain. I open PF check my feeds then minimize it. I'll check back some time later and find that it has crashed.
The only crash that I can duplicate semi-regularry is when I try to add a particular atom feed. When I get home tonight I'll test that feed again and send you the url if it is still crashing.
Posted 21 May 2004 at 11:09am #
Had it happen once. I was reading a feed, and a large ad took over my screen. (I was using ForwardDirectly.css at the time, I don't anymore because of this). I was looking at a Reuters article off of their feed at the time.
The flash ad began to cover everything on my screen and then it just disappeared, and the app went down. It's seemed to run a little slower since then, and lags a second or two when choosing articles.
Posted 21 May 2004 at 1:11pm #
The stacktraces aren't of much help to me, because they don't contain any symbols.
I'm getting (non-reproducable) crashes in _CFStringAppendFormatAndArgumentsAux (com.apple.CoreFoundation), but the traces don't point out the offending functions or methods in PF which cause it to crash on it.
Posted 22 May 2004 at 3:39am #
One possible cause for this kind of thing is if your application calls "exit" or "terminate". Odds are this isn't something you'd be doing yourself on purpose, but it can be triggered by the standard libraries under certain circumstances.
For instance, if you're using C++ a call to "unexpected" will be made when throwing from a destructor that is called while an exception is being handled (in the course of unwinding the stack). By default, unexpected is implemented in a way that includes a call to either terminate or exit (I forget which). If you think it might be this, you could plug in your own unexpected handler (via "set_unexpected") that would do something different (or that might at least let you see if this is what's happening -- perhaps via NSLog, ironically enough).
Also, in Carbon the API ExitToShell can also terminate an application early (on OS X this might be implemented via a call to "exit", but I don't really know). Unless you're calling ExitToShell yourself (or you're running a plug-in that is), however, this is probably not the cause.