CGAL users discussion list

[Chris Hamilton <>]

>> Yeah, a little bit of a rash statement on my part.  Definitely possible
>> to come up with counter-examples.  However, the whole point of
>> Nef_polyhedra is to get around these ugly cases (ie: where the
>> intersection are or contain surfaces, etc).  Is it possible for two
>> nef_polyhedron_3 without holes to produce an intersection with holes?
> 
> I can think of an example of that, yes (both 3D holes and 2D holes). But I
> seem to recall that the wider issue is that Polyhedron_3 can only
> represent manifold objects, and there are plenty of ways to wind up with
> non-manifold Nef3's.
> 
> Try computing the closure() of the interior() of your Nef. This should
> eliminate any hanging geometry.

Yeah, I've done that.  The actual intersections I'm computing are
relatively simple. I have a finite bounded volume, and I want to take
slices of it along a given axis (effectively, I'm intersecting an object
A with the region {(x,y,z) in R^3 : x1 <= x <= x2}*).  Thus, I know the
intersection will be closed, bounded and without holes.  The problem is
the 'intersecting constraints' in the answer, whatever that means.

(*Not exactly that simple, because I'm not along one of the cardinal axes.)

Cheers,

Chris