I strongly suspect this is another manifestation of the
large #define EPSILON 1E-5 in the config.h of Intel platforms....
on UNIX and Mac much smaller values are used.  This was also
responsible for the scaled sphere bug reported earlier in this
forum.  Is there a reason this large epsilon is retained in
3.1?

I also suspect, as was the case with the sphere.c problem,
the calculation in cone.c could be renormalized to avoid the
O(x^2) comparison with EPSILON, and use instead a O(x) comparison.
I will look into this second issue myself as soon as I am able.

Dan

Margus Ramst wrote:
> 
> Maybe... Did you also check this with the generated file whith only 3 decimal
> points? That doesn't seem very precise, but then I don't know much about the
> calculations involved in creating a cylinder.
> 
> Margus
> 
> Darius Davis wrote:
> >
> > Hi Margus,
> >
> > It appears to me that this is not a bug, but rather is a visible effect
> > of the limitations on the precision of floating-point operations in
> > hardware.

-- 
http://www.flash.net/~djconnel/

Post a reply to this message