[ODE] Quadtree space.

[Erwin de Vries]

> Erwin de Vries wrote:
> >>If so, the critical question is -- which two dimensions?  The
> >>dimension along which it does not scale would ideally be the
> >>'up' dimension for landscapes, sure.  But which dimension this is
> >>will depend upon whether the user adopts a left-handed or right-
> >>handed convention, which ODE has traditionally been agnostic
> >>towards.
> >
> > This is something that will be fixed in future versions. I already
though
> > about fixing it. We simply let the user define the 2 axii, or we get
them by
> > extracting the largest 2 from the extents parameter from the constructor
> > (which i think this is the best).
>
> When I first read that I thought 'oh yeah!  that's cute, that'll
> be the best solution!'.  But now I don't think so... I think that
> it's best if the user explicitly specifies their favoured
> two axes.  This is because it's quite possible for the working
> volume to be (for example) taller than it is wide, but be very
> sparsely populated across the vertical axis.

The user could simply provide zero for the up axis. If we document this,
clearly it wont give errors.

Erwin