Hi David

Thanks for the info.

So I did try to run the test on osx and it work.
But I realise that I was using the CFRetain, CFRelease, function.

I guess that in gnustep the function is not the same.

--Mathieu




----- Message d'origine ----
> De : David Chisnall <[email protected]>
> À : Discussion list for Étoilé desktop environment <[email protected]>
> Envoyé le : Ven 15 avril 2011, 14h 34min 45s
> Objet : Re: [Etoile-discuss] SIGSEGV on objc_msg_lookup
> 
> On 15 Apr 2011, at 13:30, Mathieu Suen wrote:
> 
> > Hi
> > 
> > I  am trying  to debug a segmentation fault when sending the message 
> >  NSWindow>>setTitle:
> >> From gdb the backtrace is:
> > 
> > ...
> > #5  <signal handler called>
> > #6   0x002185ab in objc_msg_lookup () from /usr/lib/libobjc.so.2
> > #7   0x008200e1 in -[NSWindow setTitle:] (self=0x8477c28, _cmd=0x9fb128, 
> >  aString=0x852b338) at NSWindow.m:1203
> > ...
> > 
> > The frame 7  is stuck in the first message send:
> > 1203      if  ([_windowTitle isEqual: aString] == NO)
> > 
> > But every thing seems  normal:
> > 
> > (gdb) p _windowTitle
> > $5 = (struct NSString *)  0x65637275
> > (gdb) po 0x852b338
> > <unnamed class> new  "0x852b338"
> > 
> > 
> > If someone have an idea?
> 
> Crashes in  message sending are usually caused by the receiver being invalid 
>(so the runtime  tries to dereference its isa pointer, but gets some nonsense 
>instead).   Have you tried running the app in valgrind?  It is probably a 
>missing  -retain message somewhere.  In this case, it looks like _windowTitle 
>has  been deallocated prematurely, and you have a dangling reference to it.  
>
> 
> David
> 
> -- Sent from my  STANTEC-ZEBRA
> 
> 
> _______________________________________________
> Etoile-discuss  mailing list
> [email protected]
> https://mail.gna.org/listinfo/etoile-discuss
> 

_______________________________________________
Etoile-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/etoile-discuss

Répondre à