[EMAIL PROTECTED] writes:
> I just tryed the "new" calculate_bounds3 and calculate_bounds2, but it does
> not solve the floating overflow/invalid problems on my VMS-box.
Sorry. I assumed it was the same bug because it fixed a seg fault on
the same test on Linux.
I think the problem is that projection skips clipped vertices but
window mapping doesn't. You can "fix" it by initializing vProj to 0.0
for clipped vertices in clip_tmp.h/cliptest_points4(). To fix it
correctly, either the cull mask has to mask off clipped vertices, or
we need to handle the window mapping in a separate function that does
check the clip mask.
Josh
