[ODE] Auto-Disabling?

[Adam D. Moss]

Aras Pranckevicius wrote:
> On the same theme: IIRC, the outcome of the discussions after my 
> latest "patch" was that auto-disable-steps are bad, and auto-disable-time 
> would be far better.

I kinda keep forgetting that we don't have ODE-level
autodisable/enable (since personally I do this at the application
level... I only get reminded when it comes up here again ;)).

What I ended up having to do there is autodisable based
on minimum quiescent time AND minimum quiescent framecount, since
at least a 3-frame sampling is necessary to not accidentally freeze
an object near the apex of its arc but a minimum length of time
is also necessary to not freeze things too eagerly
when the frame-rate is high/low. :)  If that makes sense...

I'd accept something which did that and (naturally) thresholds
angular and rotational velocity separately (if that involves
breaking the documented API for autodisable/autoenable
then IMO break it).

I'm personally all for conditionally-compiled ODE features (since
a 'universal' DLL isn't something that's useful to most of our
users IMO) which are 'on' by default.  Then people can slim down
their library and structuers to support only the features that they
actually need... at some mythical future point when I have some time
I'd like to do this for all Geom types and the two different Steppers,
if it doesn't make the source overly untidy.

Thanks,
--Adam
-- 
Adam D. Moss   . ,,^^   adam at gimp.org   http://www.foxbox.org/   co:3