Re: gtkwindow changes

[Havoc Pennington <hp redhat com>]

Tim Janik <timj gtk org> writes: 
> i gave a description to trigger this already, i can give it again:
> revert gtkwindow.c to before my recent changes, run a gtk app in
> gdb, fire up the GLE window, click on "Selector" and get a backtrace
> once it segfaults.
> 

If I remove the need_default_position = FALSE line, I don't get an
infinite loop on the GLE selector window. I'm not arguing with any
other part of your patch, I just don't like the need_default_position
= FALSE, at least until we know the exact infinite loop code path.

i.e. if I write down one cycle of the infinite loop, what is that
cycle.

Havoc