For your physics engine, have you considered ODE (http://opende.sourceforge.net)? ODE is a free, industrial quality library for simulating articulated rigid body dynamics - for example ground vehicles, legged creatures, and moving objects in VR environments. It is fast, flexible, robust and platform independent, with advanced joints, contact with friction, and built-in collision detection.
pyODE (http://i31www.ira.uka.de/~baas/pyode/) is a Python wrapper for ODE. I've only seen the demos of pyODE, but they look nice. Don't know if this can help on your quest for a physics engine.
ODE is in fact exactly the physics engine I'd like to use with Soya. Jiba wanted to integrate it into Soya at a high level. Personally, I don't care if it's integrated or not, as long as I can access it from Python. When I thought Jiba was on the verge of actually doing the incorporation, I waited around for him to get something out there. Alas, that hasn't happened yet. Good thing you reminded me about pyODE, though. I will use that for now. The collision detection and physical objects themselves will unfortunately be completely separate from Soya, so I will need to subclass everything in Python.
Tonight I managed to send Speex-encoded speech data through the network with Python. The delay is longer than I'd like, but my next step is to use pyopenal to play the speech data and get my 3d demo working in client/server mode (parallel tasks). I haven't decided exactly what to use for the client/server protocol yet; probably binary messages over UDP. I may do speech as peer-to-peer using the same UDP socket as is used for server messages, with the server sending port and IP pairs of clients that might want to exchange data and then having the clients send empty UDP datagrams to open up holes in their respective firewalls for one another.
pgp00000.pgp
Description: PGP signature
