CGAL users discussion list

[Mariette Yvinec <>]

The right way to handle PSLG with a CDT
is not to modify the constrained triangulation...
What I woul recommend here, is  the following :
- leave the triangulation unchanged, id est filling the whole convex hull
- mark facets which are not part of the PSLG interior. To mark facets,
  you may, for instance, use a
 CGAL::Triangulation_face_base_with_info_2 and add a boolean in the facets.

panayiotis foteinos wrote:

Ok.

What I am trying to do (and surprisingly seems difficult) is to remove the 
triangles outside the PSLG given the CDT of the PSLG's convex hull, and THEN 
create the appropriate infinite triangles (deleting the old ones) and link 
them together correctly. "Appropriate" infinite triangles are these that are 
indicent to the PSLG's boundary segments.

Is this feasible?

Thanks,
Panagiotis Foteinos

On Friday 03 April 2009 16:30:23 Andreas Fabri wrote:
  

The infinite vertex is SYMBOLIC, that is you shouldn't take its point
and evaluate geometric predicates with it

andreas

panayiotis foteinos wrote:
    

Thank you.

There is something else I do not quite understand. Could you explain how
I can orient an infinite triangle?

Why the predicate orientation(infinite_vertex()->point(), Point(0,0),
Point(0,6)) gives -1 (CLOCKWISE), but the test
orientation(infinite_vertex()->point(), Point(1,0), Point(0,6)) gives 1
(COUNTERCLOCKWISE) ?

Regards,
Panagiotis Foteinos

On Friday 03 April 2009 03:39:47 Olivier Devillers wrote:
      

panayiotis foteinos a écrit :
        

-----> is this because the resulting mesh does not triangulate the
convex hull
          

yes

(see the source of is_valid in Triuangulation_2.h)


either you only use the is_valid of the tds  (do not check orientation
of triangles)
either you rewrite your own is_valid keeping the geometric check only
fot finite triangles.
        

-- 
Mariette Yvinec
Geometrica project team
INRIA  Sophia-Antipolis