While working myself through the 3D-API, I found a problem with the rendering on
multi-screen workstations, as myself one is with a notebook and an external monitor.
The problem occurs on Windows Vista as well as on Windows XP. I tested
thoroughly on Vista. (haven't tested on any other OS though)
When starting a J3D-application, it starts where I intended it to start. If
there is an animation, the animation starts, but may abruptly stop. The
rendering is either continued on another part of the dual-screen, is continued
invisibly, or is being continued on parts of the screen. (for example, it may
render over an open Skype-window, or over the mailclient; however, it always
renders on the same monitor, the application-window currently is on. There is no
rendering on the secondary screen, while the window resides on the primary
screen and vice versa)
When moving the application to the secondary screen, the rendering usually is
resumed within the window (working properly again - for a short period). Moving
it back to the primary monitor usually presents the window in the
background-color of the universe.
The same is true when starting the program as an applet (for example, using the
Eclipse-functionality to run the program either as an applet within an
applet-viewer or as an application)
Running the applet within a browser, it presents the exact same behavior as
described by Issue 361 and it's comments. Therefore, those two issues may be
related to each other.
Normally, I'd think that this would be a problem within my applications, but
this behavior occurs with the J3D-Sample programs and the programs from the Java
3D-Tutorial as well. However, the rendering is more likely to fail if there is
user action involved.
For example, the "KeyNavigatorApp" from the Tutorials is almost guaranteed to
fail while moving around, whereas "static" examples like the "AxisApp" seem to
not fail at all.
Even after terminating the application, there are black spots all over the
screen, where the application has rendered itself. This link
(http://www1.siconet.at/Kastna/stuff/rendering.jpg) shows my Skype-window, after
the application rendered itself partly over it. (there are actually names
visible, I just white-blanked 'em out. The black spots are the ones, that were
created by the application)
What confuses me most, is, that this behavior also appears, even if the
secondary screen is disabled (by not extending the desktop to the secondary
screen in the windows display-settings). Although I have to add, that the screen
was still connected to the notebook (via a docking station).
There also seems to be a correlation between the runtime of the operating system
and the frequency of the issue to appear: The longer a system is up, the more
likely it is, that the rendering fails, but that may be a subjective opinion.