Re: gtk_notebook_set_current_page and the switch-page signal

[Tristan Van Berkom <tvb gnome org>]

Francesco Montorsi wrote:
[...]
> > > My switch-page handler in fact performs some code which should be 
> > > executed only on user-actions, not on selection changes by GTK.
> >
> >
> > from the docs about "switch-page":
> >     "Emitted when the user or a function changes the current page"
> >
> >  a.) "switch-page" maybe isnt exactly where all that logic belongs ?
>
> yes, I know; but which event handler can I put that logic, then ?
> There's a documented "select-page" signal but docs say almost nothing 
> about it... so I don't know how it differs from "switch-page"
>
>
> >  b.) maybe your code actually _knows_ that it is in the process of 
> > manually
> >      setting the page of the notebook and should just behave 
> > accordingly ?
>
> the only thing I can think to is to set a flag just before calling 
> gtk_notebook_set_current_page:
>
> bSkipNextSwitchEvent = true;
> gtk_notebook_set_current_page(...);
>
> static void my_page_change_callback(...)
> {
>     if (bSkipNextSwitchEvent)
>         return;
> }
>
> However I think that's rather unsafe as the user could interfere with 
> that and I fear that some user-generated events could be lost that way...

Impossible:

{
  bSkipNextSwitchEvent = TRUE;
  gtk_notebook_set_current_page (<this will recurse into my_page_change_callback>);
    my_page_change_callback () { if (bSkipNextSwitchEvent) return; }
  bSkipNextSwitchEvent = FALSE;

  /* rest of your code... */
  return;
  /* Events can only be recieved by gtk+ after this line */
}

If you are executing code - you are not reading a socket from the X
server and you are not recieving events - once you return to the
mainloop another event or other GSource can be processed.

Cheers,
               -Tristan