Re: forcing immediate redraw on widget

[Gus Koppel <gtk spamkiller bytechase cx>]

Harring Figueiredo wrote:

> > /* Process and remove all previously pending events */
> >     while (gtk_events_pending ())
> >         gtk_main_iteration ();
> > /* Verify that 0 == 0 still */
> >     printf ("gtk_events_pending (should be 0): %d\n",
> >             gtk_events_pending ());
> > /* Invalidate widget, should generate new event */
> >     gtk_widget_queue_draw (widget_to_redraw);
> > /* Check whether GTK+ knows about a new event pending */
> >     printf ("gtk_events_pending (should NOT be 0): %d\n",
> >             gtk_events_pending ());
> >
> > If especially the second result is "0" then I would suspect an
> > erroneous or missing implementation of gtk_widget_queue_draw () in
> > GTK+ 1.2. 
>
> Err, I beg to differ. This is a very erroneous assumption -- The
> events are assynchronous and they could happen just "in between"
> the call from gtk_envents_pending and printf..  This is very likely
> since printf actually causes a hardware interrupt.
>
> Again, don't assume that the program data will be unchanged on a
> assynchroneous environment (Even if it is running on a single CPU).

Yes, I know about that situation. This test isn't 100% precise but it
should suffice. I think that chances are not too bad that no GDK / GTK+
event occurs between leaving the first while-loop and the 2nd printf ()
call. The test should be executed a few mnumber of times, of course.
Besides, the 1st printf () call is really dispensable.

For this test it is not important that gtk_events_pending () returns 0
at the first printf () call. It's rather important that 1 is returned at
the 2nd printf () call. Just discard the 1st printf () call and the
result of gtk_events_pending () without a prior call to printf () will
be printed. No falsification by printf () will be possible then.

I didn't expect + know that printf () could cause a hardware interrupt
in which GTK+ could be interested, since it's neither related to GTK+ /
GDK nor even an X server itself. Which interrupt do you have in mind?