[PyKDE] Segmentation fault with dynamic creating of widget content

[Wido Depping]

On Tuesday 17 June 2003 22:54, Wido Depping wrote:
> On Tuesday 17 June 2003 14:39, Phil Thompson wrote:
> > On Tuesday 17 June 2003 10:40 am, Wido Depping wrote:
> > > On Tuesday 17 June 2003 10:26, Phil Thompson wrote:
> > > > On Monday 16 June 2003 11:34 pm, Wido Depping wrote:
> > > > > I have a widget whose children are created dynamicly. if i get new
> > > > > data i have to refresh the widget, delete all it's children and
> > > > > create the child-widgets again. Most of the time this method works,
> > > > > but sometimes i get a segmentation fault.
> > > > > This is the method for the deletion of the children:
> > > > > for x in  self.attributeWidget.children():
> > > > > 	name = str(x.name())
> > > > > 	# this ensures, that the layout is not deleted
> > > > > 	# every label, lineedit and button has a name which starts with
> > > > > 'LDAP' if name[:4] == "LDAP":
> > > > > 		x.destroy()
> > > > > 		self.attributeWidget.removeChild(x)
> > > > >
> > > > > So, is my code wrong or is it PyQt's fault? Attached is a
> > > > > backtracke i got from gdb:
> > > >
> > > > What happens if you call removeChild() before you call destroy()?
> > >
> > > Then I get a segmentation fault, too. This time it happens, if I
> > > refresh the widget and then press 'Alt'. It also happens if I minimize
> > > and restore the window. Also with the same backtrace.
> > > That's why I executed the destroy() before removeChild().
> >
> > You should definately do the removeChild() first. I would also "del x"
> > rather than "x.destroy()". Other than that, have you got a small, but
> > complete, example that demonstrates the problem.
>
> Ok, I think I have a testcase, which should trigger the error. Needed some
> time to find out, what was triggering the segfault.
> I create for each row a button and install a eventfilter. After leaving the
> eventfilter, the event is normally processed. And this might be the cause
> of the error.
> If you try to switch between the openend windows often and fast enough, the
> errorr occurs with a similar backtracke I get with my real programm.
> Attached is the sourcecode. Run with 'python TestWidget.py'

Sorry, that was the wrong script. Now the right test case should be attached.
And by the way, the current pyuic seems to have problems with producing extra 
code for testing. The widgets show() method is never called and the 
QApplication exec_loop is missing, too. The generated .py file is executable, 
but nothing happens ;)

Wido
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TestWidget.py
Type: text/x-python
Size: 4598 bytes
Desc: not available
Url : http://www.riverbankcomputing.com/pipermail/pyqt/attachments/20030618/ad03eb80/TestWidget.py