CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

Read this:
http://www.cgal.org/FAQ.html#inexact_NT

and then use Exact_predicates_exact_constructions_kernel

Sebastien.

On 02/27/2012 06:41 PM, Zohar wrote:
> Hi,
>
> Could it be that intersecting two 3D segments that overlap in one edge
> (s1[0] == s2[1]):
>
> s1: -1.10431 1.4678 0.379479 -1.10499 1.48276 0.428179
> s2: -1.10356 1.45133 0.325848 -1.10431 1.4678 0.379479
>
> with CGAL::Cartesian<double>  isn't good enough:
>
> Segment intersection (=s1): -1.10431 1.4678 0.379479 -1.10499 1.48276
> 0.428179 sqrLen=0.00259594
>
> and I need to use CGAL::Exact_predicates_inexact_constructions_kernel?
>
> Actually I found the problem when I intersected 2 triangles:
>
> Tris sharing 1 vertex,
>   t1:(-1.11051 1.47118 0.423155 -1.0993 1.4947 0.43336 -1.10431 1.4678
> 0.379479),
>   t2:(-1.10431 1.4678 0.379479 -1.09949 1.47281 0.327215 -1.10798 1.42799
> 0.324362)
>
> and I debugged the problem to intersection_collinear_segments().
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Intersecting-with-Cartesian-kerenl-tp4425392p4425392.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.