Re: Dumb question

[Mark Leisher <mleisher crl nmsu edu>]

Mark Leisher wrote:
> Chris Moller wrote:
> > I'm writing a roll-your-own widget that uses, among other things, a 
> > few buttons (via gtk_button_new()) and spin_buttons 
> > (gtk_spin_button_new()).  In my widget _init() fcn, I create all the 
> > buttons, pack them into an hbox, and pack the hbox and some other 
> > stuff into a vbox.  When the app the uses the new widget runs, the 
> > spin buttons show up but the regular buttons don't.  They're there--if 
> > you click on the empty space where they ought to be, the callback gets 
> > invoked, but the buttons aren't visible.  This happens regardless of 
> > whether I use gtk_button_new_with_label ("whatever"), or just 
> > gtk_button_new() and gtk_container_add() various things (labels, 
> > labels in boxes, etc) to the button later.
> >
> > Anyone have a clue what's happening?
>
> Not so dumb. I struggled with this one for a couple months.
>
> I discovered with GTK+ 2.7 and later that when creating subclasses of 
> GtkContainer, you *have* to implement the add, remove, and forall class 
> functions of the GtkContainerClass.
>
> The add and remove functions don't need to do anything, but the forall 
> function should apply the passed callback to all the children you've 
> created.
>
> Without all three functions implemented, some of the child widgets show 
> up and some don't.

Almost forgot. I can't remember which version of GTK+ I was using, but 
in one case, a NULL callback was passed to the forall GtkContainerClass 
function. Check the callback before calling.
--
---------------------------------------------------------------------------
Mark Leisher
Computing Research Lab                   In the republic of mediocrity,
New Mexico State University                genius is dangerous.
Box 30001, MSC 3CRL                        -- Robert G. Ingersoll 
(1833-1899)
Las Cruces, NM  88003