CGAL users discussion list

[Zohar Levi <>]

Okay, thanks.

(Not sure if this message goes directly to you or shared on the mailing 
list--I'll know after it's sent.)

On 16/1/25 19:57, \"Sebastien Loriot (GeometryFactory)\"" 
( via cgal-discuss Mailing List) wrote:
> If you are outside of the arrangement, then maybe you can use the
> edge that is the closest to your query point. If you have some convexity
> hypothesis, then it's probably only a matter of some orientation tests
> on the edge on the boundary.
>
> Sebastien.
>
> On 1/16/25 03:44, Zohar Levi ( via cgal-discuss 
> Mailing List) wrote:
> > Sorry, I haven't considered it deeply. It's not my code, and I was 
> > aiming for a quick solution.
> >
> > I said triangulation, but it's a triangulation that is held in an 
> > Arrangement_2, which does some intersections. It's attached to 
> > Arr_landmarks_point_location for point location. I could probably 
> > rethink the whole design unless you have a quick tip. It just seemed 
> > weird that the point location can't give me any useful info about a 
> > point that is epsilon away from my shape.
> >
> > On 16/1/25 9:28, \"Sebastien Loriot (GeometryFactory)\"" 
> > ( via cgal-discuss Mailing List) wrote:
> > > If you have a triangulation, why don't you use it for the point 
> > > location?
> > >
> > > I'm probably missing something...
> > >
> > > Sebastien.
> > >
> > > On 1/15/25 21:21, Zohar Levi ( via cgal-discuss 
> > > Mailing List) wrote:
> > > > I have a triangulation embedded in a unit disk (after Tutte).
> > > > I'm doing point location using 
> > > > CGAL::Arr_landmarks_point_location<Arrangement_2>.
> > > > Due to numerical issues, a point that I'm looking for falls in the 
> > > > unbounded face but an epsilon distance from the boundary.
> > > > The return value is the unbounded face, which seems like returning 
> > > > failure. Since a search has already been done, is there a way to 
> > > > get more useful info, e.g., closest boundary edge? Otherwise, I'll 
> > > > need to search again with AABB or something.