[ODE] Recent version?

[david@csworkbench.com]

> Hello David, ODE community,
>
> I have just finished catching up with the past 400 ode mails making sure
> that all patches are in my install and things seem to be working- the
> only headache was a conflict betwen the old dRay contrib and the new
> built-in one.
> Thanks for hinting at that stepfast routine, but I am already
> encountering instability with ODE as it is (the wobbly wheels, some
> contact problems with tricolliders,etc.)- trading off more accuracy does
> not seem to be the right thing IMO.

Well, in some cases, the stepfast solver can be more accurate than the old
solver.  For example, if you make a wall of equally massive (that's
important) blocks, at about 8 iterations, there is less jitter in each
block (since the stepfast solver runs at a smaller stepsize
(stepsize/iterations to be exact), a less accurate solution at a .001
timestep may be more accurate than a fairly accurate solution at a .01
timestep).  I'm not sure if it will work the same for the wobbly wheel
problem, but it might be worth a shot.  It's biggest enemy is large mass
ratios though, so if you insist on putting 10 kg tires on a 2000 kg car,
it will almost definitely be useless.

> Also while going over those past mails I noticed that the question for a
> GetHinge2Angle2() has turned up four or five times- but no one replied.
> So what is up with that? I do not understand how the GetHinge2Angle1
> function works, or else I would have implemented it a while ago. My
> question to you joint experts out there: can such a function be created?
> Why is it missing despite what it says in the ODE manual?

No clue on this one.  The function the Angle1 function calls looks like it
returns the the 2nd angle to me....

> Thank you all,
>   Marco Grubert
>
> http://www.conitec.net/a6info.htm
>
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode