From: "Pierce, Gregory (TBS)" <[EMAIL PROTECTED]>
Sent: Monday, July 30, 2001 10:30 AM
> What is the resolution of the events that come from behaviors?
If you mean time resolution, then here's how I think it works:
When a wakeup criterion is met on an armed Behavior [1.)
attached to a live scene graph, 2.) scheduling bounds
intersecting a ViewPlatform's activation volume, and 3.) wakeup
condition set], an event is triggered, and that event goes onto
a stack. The Behavior scheduler checks the stack every X
milliseconds, picks off triggered Behaviors, and executes their
processStimulus() methods.
'X' appears to be OS-related, and it seems to have the same
granularity as Java Object.wait(). For WinNT and Win2000, 'X'
is a little over 15 milliseconds. I think for Solaris, it is 20
milliseconds. I haven't been able to design experiments to find
the pattern used for the other Wins.
In the (common) case in which Behavior is set to
wakeupOnElapsedFrame(0), which is to say that it wakes up as
soon as the renderer has completed the last frame, then I think
a different approach may be taken. The triggered Behavior stack
may be checked at that time. I haven't probed this setup in any
depth.
Disclaimer: This description is completely without authority.
It's a collection of suppositions based on the results of
experimental probes.
Cheers,
Fred Klingener
Brock Engineering
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
