Strange physics-related lockup

Do you probably alter the scenegraph structure in the updateGeometricState method? (add, remove spatials, set parent etc.)



First switch off your own application code which reacts on collisions, just to sort that out. If it's still locking up, try to reproduce it in a short test and post that here.



A general not for finding lock ups: instead of debugging by system out println  :P, simply run in debug mode, hit the lock up, and pause your application. Then watch the stacktrace of the interesting thread (usually "main", with game states it's some worker thread)…

OK, brief update on this. Initially I stated that the problem goes away when I make the new object that triggers the lockups spherical. I've done a bit more testing, and it seems to work when the bounding volume itself is spherical, even if the physics geometry is a box. So I'm guessing it's an issue in JME, and not at all physics-related. I'll post the question in the general JME boards shortly, perhaps with a bit of extra research.



But here's a question to tie it all back to physics, though. What effect does the bounding volume have on the physics code? Is it OK to bound everything in spheres even if they're boxes, or do the bounding volues factor into the createPhysicsGeometry call somehow?



And, to answer some of the points addressed earlier, I've disabled my application's own collision code, and am not (to my knowledge) modifying the scene graph. I'll work more on this tomorrow and will post something more substantial to a more appropriate board.

thewordnerd said:

But here's a question to tie it all back to physics, though. What effect does the bounding volume have on the physics code? Is it OK to bound everything in spheres even if they're boxes, or do the bounding volues factor into the createPhysicsGeometry call somehow?

boundings affect only generateCollisionGeometry, the create calls don't care about bounding (like the rest of the physics).