[ODE] cone twist joint & other stuff

[Yordan Gyurchev]

Hi,

first time poster...

Some background info:
I've used an old modified version of ODE with some custom collision system
of ours. I took over after the physics guy moved on.
I've just downloaded the 0.5 ver and managed to get it going in a test bed
(game state) at work.

List archives?
Any alternative location?
I was not able to download the 35meg mbox. It simply gives me a dead link.
Sorry for any duplicate questions.

Questions:
Have people used it on the PS2? Any problems?
I've managed to compile it on SN System's ProDG for PS2. It starts to go
less than 60 fps if I stack more than 25 boxes (not complaining!).
But if people are aware of issues or have some specific ps2 opts (and
willing to share ;) I would love to hear from them?.
Compilation for PS2 the VC6 config works fine spare the _copysignf. OPCODE
fires compile time asserts about some type sizes.

How do people usually represent their static physics world?
Can this be saved in native format with internal trimesh so I can shift the
tree build process in the tools? Any example code?

We have a notion of mutable bodies. I don't know why we call them like that
but basically means bodies that don't react in physical manner but are
directly controlled by the game: character avatars, animated physical
objects (elevators, doors, etc).
What is the usual approach for implementing such bodies in ODE?

Cone/Twist joint:
At some point I wanted to have a more intuitive joint for representing
shoulder like human joints so I came up with this. Without doubt this has
already been discussed and has some fatal flaw but here we go anyway. It
does what it says. Imagine a cone around your shoulder and that it was
restricting you arm. The twist part stops your arm rotating around its own
axis. It uses two Amotors...
I'm no good at explaining this thing but I wrote step by step code
explanation for a friend once. You can find it at:
http://www.gyurchev.com/docs/ConeTwistJoint.txt
Is anything wrong with that kind of joint? To me it seems a lot more
intuitive than euler angles...
In a real shoulder it would need a ball joint as well... but of course you
already know that.

There are few other issues I'm interested in but I'm hoping I'll get hold of
the archives before I bother the list with more questions.

Sorry for the long first post.

Thanks,
Yordan