CGAL users discussion list

["Laurent Rineau (CGAL/GeometryFactory)" <>]

Le vendredi 09 août 2013 05:20:49 Bear a écrit :
> Dear Laurent Rineau,
> 
> Yes, my polygons do intersect, but the unexpected vertexes are either
> intersections of constraints nor infinite vertexes.
> 
> Please see my replay to Sebastien.
> http://cgal-discuss.949826.n4.nabble.com/unexpected-vertexes-in-constrained->
>  Delaunay-triangulation-tp4657889p4657914.html

I think your incorrect points are intersections points of almost-colinear 
segments, whose constructions have a lot of rounding errors.

Can you test with the exact-exact kernel?

#include <CGAL/Exact_predicates_exact_constructions_kernel
typedef CGAL::Exact_predicates_exact_constructions_kernel K;


That kernel has exact constructions but is slower. If, using that kernel, you 
have a correct result, Sébastien and I will show you a trade-off between 
exact 
computation and speed, so you can have the best of the two solutions.

-- 
Laurent Rineau, PhD
R&D Engineer at GeometryFactory           http://www.geometryfactory.com/
Release Manager of the CGAL Project       http://www.cgal.org/