re: on_expose_event not being called when expected (Garth's KidStuff)

["Freddie Unpenstein" <fredderic excite com>]

From: "Garth's KidStuff" on 06/11/2008 :

>> This logic works impeccably (*grin*) until the CPU is under heavy load. In
>> that case, I can see the invalidate_rect calls happening but
>> on_expose_event doesn't get called until the cpu load is back to normal.
>> I've been banging my head against this for a few days now and would
>> love some new ideas about how to attack the problem.
> I found a function -- gdk_window_process_all_updates() -- that seems to
> solve this problem. If I call this immediately after invalidating the rectangles,
> then on_expose_event always gets called even under the adverse
> conditions described above.

Out of interest, have you tried playing with event priorities at all...?
Shift the periodic timer to a very low priority (lower than GTK+'s book keeping and I/O handlers).


If that doesn't work, set up a see-saw between it and the expose handler, something like this; increment an integer flag from the timer.  Only if the flag is now 1, perform the action and allow the timer to persist.  From the expose handler, invoke the timer action and set the timer if the flag is 2, and in any case reset it now to zero.  Under good conditions the timer will always see a zero and raise it to a one, and the expose will always see a one and drop it back to zero.  If things mess up, the timer will raise it to two and stop re-triggering until the expose handler runs, which will see the two, perform the timer action (which should likely be in a common function), restart the timer, and again, reset the counter back to zero.


Someone else may have a better idea, but hopefully that helps.

Fredderic


------------------------------------------------------------
Domain Registration
Cheap Domain Registration - Click now.
http://tagline.excite.com/fc/JkJQPTgLMJA6FHZ1HYp57nkgeZWMJKPTqcYYTVrv3iUzLDQcbgMYUz/