[ODE] some quick notes on 0.6...

[Martijn Buijs]

David Walters wrote:

> Could you explain a bit about the RECOMPUTE_RAYNORMAL and DO_RAYDEPTH
> options.

Actually I'm not sure myself. It has been a while, but I believe it has to do with how the contact 
plane is calculated. Most of the dTerrainY collider code was left untouched, perhaps the original 
author (Benoit) can tell us more.
However, everything should work fine as it is, without changes.

 > I'm not familiar with GetDepth versus what seems like a ray
 > collision and would like to know which is likely to be faster, more
 > reliable, etc.

I haven't done any thorough tests with rays (I'm not even sure collisions with rays will result in 
contacts), you'll have to experiment with it.

> I've been using a trimesh for environment collision, but the inclusion
> of a solid object representing at least some sections of my game world
> should (if my early experiments with the plane geom are anything to go
> by) greatly reduce the chance of my game entities getting stuck in the
> ground!

dHeightfield proved completely stable for us (sphere, box, capsule, cone), and as it utilizes the 
plane collider internally, it should indeed be as stable as regular plane geoms. :)

Martijn