Re: TreeView crashes X server?

[Joshua N Pritikin <vishnu pobox com>]

On Mon, Dec 17, 2001 at 02:00:52PM -0500, Jonathan Blandford wrote:
> > Ever since jrb committed the incremental reflow, GtkTreeView seems a
> > lot less stable.  Sometimes i get inconsistent behavior as if there are
> > some uninitialized variables.  Sometimes i seem to get random memory
> > corruption.
> 
> Erk.  That shouldn't happen.  I committed a bunch of RBTree corruption
> fixes last week.  Does it still happen with current CVS?

Yes, but i admit it is hard to reproduce and does damage to my application
very gradually.  i guess it could be my bug too.  Let's fix the obvious
stuff first ...

> > Generally, i am very careful and program as defensively as possible.
> > i never saw these types of problems before the incremental reflow
> > commit.  i haven't filed bugzilla reports because even the test
> > programs which come with gtk+ don't work in obvious ways.
> 
> Which ones?  Where?

It's too hard to test your own code, i guess.  Here are some hints:

* testtreecolumns: Add a bunch of columns.  When you toggle the visible
checkbox, the treeview only updates the header.  It doesn't redraw the
rows.  If you add an option to turn off the header then i can show
you another bug.

* testtreeview: Change the model to NULL.  Change the model to GtkListStore.
Notice how the incremental reflow doesn't happen until you click a checkbox
or something.

-- 
Victory to the Divine Mother!!         after all,
  http://sahajayoga.org                  http://why-compete.org