CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

I can reproduce it. I investigate and let you know.

Sebastien.

On 11/28/2013 03:13 PM, Hugo Mercier wrote:
> Hi,
>
> The behaviour of CGAL::intersection( Triangle_3, Triangle_3 ) changes in
> 4.3.
>
> With these two triangles :
>
>      CGAL::Triangle_3<K> tri1( CGAL::Point_3<K>( 0, 0, 0 ),
>                                CGAL::Point_3<K>( 0, 1, 1 ),
>                                CGAL::Point_3<K>( 0, 1, 0 ) );
>      CGAL::Triangle_3<K> tri2( CGAL::Point_3<K>( -1, -1, 0 ),
>                                CGAL::Point_3<K>( 0, -1, 0 ),
>                                CGAL::Point_3<K>( -1, 0, 0 ) );
>
> it returns the point (0, -1, 0). Which is wrong, the two triangles do
> not intersect.
>
> If I understand Triangle_3_Triangle_3_intersection.h correctly, the
> intersection between the two supporting planes is computed (here a line)
> and then the intersection of each triangle with this line (here a
> segment for tri1 and a point for tri2). Then ... the point is returned.
> I think the point should only be returned if it intersects the other
> segment (which is not the case here).
>
> Do you reproduce / agree ?