Re: Tricky treemodel question

[Ben Martin <monkeyiq users sourceforge net>]

On Thu, 2002-07-18 at 22:19, Paul Davis wrote:
> >it goes. These events trigger a gtk_main_iteration_do() so that the gtk
> >progressbar is updated, but GTK also tries to update the treeview which
> >is not ready to be updated because its has called
> >tree_store_iter_children() and is not in a state to be redrawn.
> 
> i haven't used the treeview code yet (i'm still at gtk+ 1.2) but some
> of its equivalent in 1.2 had "freeze/thaw" methods to prevent this
> >from happening. you called them before and after the updating.

Yep, I noticed these methods in 1.2 and spoke with a few of the authors
of the tree widget on #gnome about them for gtk2 because of some
performance issues I was having with adding 1000 children in a "live"
treeview. The update issue only seems to happen for expanding a subtree,
and then its somewhat unpredictable (I have no threads in the app at the
moment). I have disabled the subtree expansion in the client for now,
because of the above and I also find that the root of the subtree
sometimes becomes very big (the whole row is ~800 pixels in size from
~25) for no reason that I can find. I am forcing the height of most
cells and the widths of about 75% of them at the moment and my cell
renderer get_cell() method returns sane data.

> 
> --p
-- 
-----------------------------------------------------
http://witme.sourceforge.net/libferris.web/