Re: GtkTreeModel

[Christer Palm <palm nogui se>]

Hello again!

Christer Palm wrote:
> edscott wilson garcia wrote:
>
> > Use gtktreestore model. Add a dummy for non expanded directories. That
> > makes the expander appear. When the user expands, if nothing is in
> > there, collapse the node and remove the dummy. Since there will be no
> > children, the expander disappears.
>
>
> Hmm.. Yeah - that should work I guess. Although it feels a bit backward 
> from an object-oriented point-of-view.

In case anyone is interested, I just want to report that I've 
implemented Edscotts suggestion, and it works great! Thanks!

I've also explored the row_has_child_toggled() and row_deleted() 
functions of GtkTreeModel a bit further, and ironically, while GTK 
immediately calls back on my interface to get the new child and node 
status when I call these functions, it still messes up and tries to 
refer to the deleted node as well as complain about inconsistency with 
the has_child status. At least when I call these from within the 
iter_children_vfunc() implementation.

Even more surprising is that GTK attempts to dereference the GtkTreeIter 
iterator returned by my iter_children_vfunc() implementation, even 
though I have this function return false (and set the GtkTreeIter to an 
invalid value), as explained in the docs. So this does simply not appear 
to work correctly [or as documented :-)].

--
Christer Palm