Re: [gtk+970606] problems with gtk_widget_destroy()




David Mosberger-Tang writes:

 David> Unfortunately, I don't have much time to track this down. It looks
 David> like inserting a vbox into an hbox via a gtk_container_add()
 David> increments the ref count in the vbox by two.

Peter Mattis writes:

 Pete> That is exactly what is happening. gtk_container_add calls
 Pete> gtk_widget_set_parent which is where widgets (objects) get
 Pete> referenced. But the box container_add functions also calls
 Pete> gtk_widget_set_parent, referencing the widget again. I've gone through
 Pete> gtk and fixed places where this occurs. I've also put code into
 Pete> gtk_widget_set_parent so that it will warn you if you set the parent
 Pete> of a widget to its current value (which is what happened in this case)
 Pete> or if you call gtk_widget_set_parent when the widgets parent field is
 Pete> non-NULL. (ie. It already has a parent).

 Pete> A patch to do all this follows. (gzipped and uuencoded)

The patch definitely does the trick for xscanimage.  Neat!

	--david



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]