Re: [Pyobjc-dev] Crash in PyImport_ImportModuleEx

[Bob I. <bo...@re...>]

On Jun 27, 2006, at 1:28 PM, Michael Tsai wrote:

> On Jun 27, 2006, at 6:41 AM, Ronald Oussoren wrote:
>
>> The outline-edit bug is the only know issue, and I seem to have fixed
>> that. I guess it's time to gear up for another release, this time one
>> that is properly announced and fully tested :-)
>
> I've been seeing a crash in restore_sem_to_pool for a while now, in
> all kinds of situations, and I've finally been able to reproduce it
> at will (in my app, not yet in a test app). It happens on PowerPC and
> on Intel. The crash log is at the end of this message. The way I can
> make it happen is to start a bunch of child threads using NSThread.
> Then I make them finish by falling off the end of method. Finally, I
> mouse over my NSTableView and let it display a bunch of tooltips,
> which are generated using the delegate method
> tableView_toolTipForCell_rect_tableColumn_row_mouseLocation_. This
> causes it to crash in PyImport_ImportModuleEx.
>
> I will try to produce a small test app in which to reproduce this
> crash, but in the meantime does anyone have suggestions? It's always
> KERN_INVALID_ADDRESS with the same address. Also, it seems odd to me
> that the stack for thread 0 (the only thread at this point) begins
> with _pthread_body/forkThreadForFunction rather than start/
> NSApplicationMain.

Are you using Python 2.3? This sounds like it could be one of the GIL  
bugs in Python 2.3.

-bob