I am also using 1.3. After a long time trying to overcome repainting
using a combination of Swing and 2D graphics, i gave up and basically
wrote my own 2D graphics routines *and* avoided using any of the
clipping routines provided. (I mention clipping because i believe Swing
uses these underlying routines.) I need reliable frame output since i am
doing motion-based experiments. I now do get reasonable repainting
performance in all except fonts. 

The initial iconification time lag, which i have noticed in all the
applications i have developed, is almost certainly due to caching
behaviour.

After trying to look at the Swing source, I came to the conclusion that
there are severe issues with the "coalescence" that is supposed to
happen in the event queue with respect to repainting. Before you caution
me, i know that Swing is not thread-safe, i know that repaint and update
automatically are executed on the event queue thread, and no, that was
not the cause of my problems. I do not think the event/repainting queue
yet works 100% in Swing, and if these performance issues are not well
addressed soon, i believe it jeorpardises Java/Swing as a feasible
platform for many highly-interactive applications of anything larger
than small game size. Which is too bad, it's a nice language and
environment in which to develop.
-- 
-------------------------------------------------------------------------
Lyn Bartram                                 Voice: +1 604 291 4369
Graphics, Usability and Visualization Lab   FAX: +1 604 291 4424
Simon Fraser University                     email: [EMAIL PROTECTED]
Burnaby, BC                                 WWW:
http://www.cs.sfu.ca/~lyn
CANADA V5A 1S6
-------------------------------------------------------------------------
He uses statistics the way a drunk uses lampposts: for support, not
illumination.
_______________________________________________
Advanced-swing mailing list
[EMAIL PROTECTED]
http://eos.dk/mailman/listinfo/advanced-swing

Reply via email to