[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Thu, 13 Apr 2006 16:39:57 +1000
Wim Oudshoorn & Chris Vetter both provide nice solution.
However I have problem with both.
Wim method, while working in a test program, fail to bring any information
when I hacked lib foundation so that NSZombie log the error stack.
I ended up with stack like
0 (nil) (nil)
1 (nil) (nil)
2 (nil) (nil)
And Chris Vetter method doesn't compile with GNUstep.
It uses functions defined in objc/class.h which is missing in GNUstep.
In both case that didn't work.
Anyway I have uncovered a bug in GNUstep and I give up trying to solve it
but I can explain it too you.
This bug show in "some circumstances", on WIndows, when it's a first install
of GNUstep, that is: when there is no user default file yet saved on the
To reproduce the bug I delete the Application Data\Gnustep directory.
I have a simple command line ObjectiveC program which deosn't bug!
I have a complex .NET application using my ObjectiveC.NET binding which
crahs all the time when the aplpication get first idle (when I'm running
something like my 3rd event loop).
When I do [theAutoreleasepool release]
plenty of things happend and eventually the variable
static NSFont *placeHolder = nil;
in NSFont.m: 188
get deallocated twice..... (while it should not be deallocated at all,
That happend only when I have no initial NSUserDefault file.
I have no idea why.
The stacktrace would sure have helped me abit.
Now I have a work around, I retain it a few times when it is created...
(another NSFont hack)
But I would like to solve it, so that a SVN update doens't dismiss my
- StackTrace problem,
Lloyd Dupont <=