Re: NSActionCell _control_view

[Rob Burns]

On 2003-12-29 18:35:45 +0700 Fred Kiefer <fredkiefer@gmx.de> wrote:

> Rob Burns wrote:
> > when the app functions correctly, a description of the NSTableView 
> > is 
> > printed. When the _control_view variable has gotten set to something 
> > else, 
> > it will sefgault on the NSLog. The program can be found here:
>
> It did not segfault for me, but printed the following:
>
> 2003-12-29 11:57:56.052 Pimzler[3377] cell control view - 
> <GSControlTemplate: 
> 833c4b0>
> 2003-12-29 11:57:56.053 Pimzler[3377] cell control view - 
> <GSControlTemplate: 
> 833c4b0>
>
> Looks like the Gorm prototype is still stored in the NSActionCell 
> when you 
> are about to display it. One easy fix for this could be not to 
> encode/decode 
> the control view with the NSActionCell. I am not sure why this gets 
> stored at 
> all.

oops, incomplete description again. I too get GSControlTemplate 
returned sometimes. usually only for the drawing of the first two 
cells. After that, it reports the NSTableView subclass. niether of 
these things make it crash for me though. I did manage to find one 
other person that it also crashed for, so I don't think its unique to 
my machine/os/library setup. And I've filed a bug report

Thanks
Rob


> On the other hand I also don't see, why your application should fail 
> over 
> this. Hmm, maybe the GSControlTemplate has by that time already been 
> freed, 
> as the control view is not retained and you are now referencing an 
> invalid 
> object. Yes, that might do.
>
> I just tested with setting _control_view to nil in [NSActionCell 
> initWithCoder:] and that would work. But we need more input on the 
> side 
> effects of this change or better solutions to it. Best you report 
> this as a 
> bug in the Savannah form, so it wont get lost.
>
> Cheers
> Fred
>
>
>
> _______________________________________________
> Help-gnustep mailing list
> Help-gnustep@gnu.org
> http://mail.gnu.org/mailman/listinfo/help-gnustep