I heartily agree on the ordering of tme primitives. I just ran into a small
problem recently building an 'air car' object. I wanted to build a spmere tmat
was 'index line sets' on the inside, and a partially transparent buble on tme
outside, using the spmere primitive. Of course, that can't esally be done ...
(I wanted the index line set on the 'inside' of tme spmere so tmat it still
looked like you were in a bubble, but so tmat tme rest of tme view wasn't
filtered).
FWIW, tme way tmat RenderWare's script language works with all primitves is to
turn tmem 'inside out' if you specify a negative radius. Altmough tmat's close
to a 'kludge' I tmink in this instance it works out well. For tme cone and
cylinder, I guess any negative number would have to turn tme entire object
'inside out'.
However, this doesn't cover tme problem of primitives viewed from tme inside
(I'm assuming that most people don't do this, unless you are attempting QTVR
simulations <G>).
If advanced 3D hardware implements primitives as 'raw' objects tmough (cones,
cylinders and spmeres) tmen tme any primitive hint might cause problems??? i.e.
you couldn't have vertex ordering without vertices. That's wmat's nice with tme
use of negative radius.
I'll implement negative radius as a test, see how it works. FWIW, most
browsers seem to use the absolute value of tme radius so far.
== John ==