The indicated method does not detect a very clear collision with the bounding box for a TriMesh plane. BoundingBox.intersectBoundingBox() does detect the collision between the same 2 objects.
There are some comments in the method, but considering the lack of decipherable variable names (names like "fAWxDdu" are used), it isn't enough for me to reverse engineer the algorithms.
The collision is incorrectly rejected by the conditional "fR > fR01".
My work-around is to use a BoundingBox instead. That's too bad, because OBBs are a case where jME has a solution when Java3D had nothing. Looks like I'm back to custom coding like in the good 'ole days.
Hi blaine!
I think that BoundingBox and OrientedBoundingBox need some more work, I thought it came from me but I was wrong. When I put a tiny OrientedBoundingBox inside a larger BoundingBox, the merge of the both should give approximatively (because of some rounding errors) an OrientedBoundingBox with the same volume than the larger BoundingBox but it only works fine when the OrientedBoundingBox is aligned on the axis. Lol Ardor3D uses exactly the same source code (except that it uses some object pooling). Maybe it would be fine to have a precise test case to reproduce the bugs.
Edit.: I need to detect the collision correctly between a TriMesh plane (a portal composed of coplanar triangles) and an OrientedBoundingBox (the bounding volume of a player). The workaround should consist in using BoundingBox but for the moment, I see almost no difference.