Ok, we've started now…
But I wanted some ideas as to how you worked with ODE and ODEJava
How much of ODEJava did you use/change?
I'm just looking for some pointers as we look to tackle this beast known as PhysX. 
Ok…
I see I'm going to have problems with stuff like this:
virtual NxMat34 getLocalPoseVal() const = 0;
virtual NxVec3 getLocalPositionVal() const = 0;
virtual NxMat33 getLocalOrientationVal() const = 0;
The types for those functions...I don't think swig or JNA handles that for you... :?
It would be unrealistic for us to try to do that by hand
I think swig handles this but…
JNA looks so much better…yet to say that the documention sucks is an understatment…
I'd recommend converting that on native side from Matrix4f/Vector3f/Matrix3f to NxMat34 etc. and vice versa. Reusing the native structs where possible.
I'm not really sure what you're saying…
I was looking through all of the Foundation includes and basically everything is defined except for the floats and ints.
This wouldn't be a problem save for it being stated as NxU32 and NxF32 there's no direct translation for that, even though we know what they are.
And it doesn't appear to be defined by the includes as I said…so my issue is if I wrap the core dll as is…will it work?
Also is the issuse of classes and instancing…I wonder if these wrapping programs handle that as, accounting for these by hand is unrealistic…so I'm sure some of them do.
I know but there are certain things that are an absolute pain to code in JNI. And given the size of this PhysX it would require more than just us two. Besides Aegia I guess wanted to be cool and create their own type names. This is confusing and annoying. Hopefully we will have SOMETHING by the end of this semester. But I've got school so… I don't know…
Yes the implementation of this, even halfway, would be be a great step for JME as there are so many great things that you could acomplish with Physx that would be a pain to implement on your own… It's the cloth tearing that motivates me… just awesome…