"Cousin Ricky" <ric### [at] yahoocom> wrote:
> The cylindrical axis is required to orient the discs that cap the cylinder.  The
> axis, in turn, requires 2 distinct end points.  With a zero-length cylinder, the
> discs cannot be oriented, leaving no way to know what to render.  My guess is
> that the POV Team decided it improper to proceed under that circumstance.
> Degenerate triangles do not have this problem.

Yes, the reason is probably that a degenerate triangle *unambiguously* collapses
to a single infinitely thin line (or even a single point) which cannot be seen
from *any* angle, and therefore just ignoring it a perfect way to handle it.
The warning is just there to make the user aware, in case it was *not* intendet
that way.

A degenerate cylinder with non-zero radius, however, *should* be an infinitely
thin disk, which *should* be visible unless viewed edge-on, but cannot be
defined unambiguously using the define-by-endpoint approach of POV SDL.

The bottom line is that it is clear how a degenerate triangle should look like,
but not so for a degeenrate cylinder.

Post a reply to this message