User-agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.4) Gecko/20030624 Netscape/7.1
Many thanks to all who responded to my original post about some very
active electronic display simulators blocking the update of a Swing
based control panel. If you recall, the electronic displays simulate
things like spectrum analyzers, o-scopes, etc. and when three or more of
them were opened up (these are VERY active drawing programs) the Swing
control panel that would send them parameters to change things (like
center frequency, bandwidth, timing) would really be blocked and it
would look like a couple of gray rectangles (eventually it would fill
in, but end up in the same state if another window covered the Swing
stuff).
In any case, an analyisis of the code showed that ALL of the work for
updating the electronic display simulators was done in the paint()
method. In p-code this would look like this:
1. Obtain data to be graphed from source.
2. Initialize graphics, etc.
3. Load into buffer
4. Analyze timing and scale diagram properly, set sleep value for
"run()" here
5. Draw image
paint() is of course triggered by re-paint and this is issued by a
main-thread in a run() method, which when simplified is little more than:
sleep(<time calculated from step 4>)
repaint();
The code in steps 1-4 above is a TON of code. It was ALL running on the
event dispatching thread.
The problem was resolved by cutting a lot of code out of the paint()
method, putting it in a new method called inside the run() method which
is running in a main-thread, NOT the event dispatching thread, and once
the image is prepared, draw it inside the paint() method (on the event
dispatching thread).
Now the p-code looks like this
paint():
1. Draw image
newMethod():
1. Obtain data to be graphed from source.
2. Initialize graphics, etc.
3. Load into buffer
4. Analyze timing and scale diagram properly, set sleep value for
"run()" here
run() (looping of course)
1. call newMethod() to do all the heavy work
2. sleep allocated time from newMethod()
3. repaint().
Needless to say, a wait/notify set had to be introduced.
NOW IT WORKS FINE!!!
Thanks everyone.
P.S. : For any of you future book writers out there please STRESS the
importance of the event dispatching thread and its effect on graphics
updates.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Java-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/java-dev/email@hidden
