CGAL users discussion list

[rytis <>]

Hello,

I am a beginner with CGAL which I like very much so far. I think there is an issue (of documentation) with constrained Delaunay triangulation. This is the issue that is not clear from the docs.

If someone could indicate how solve my indexation problem (at the end) that would be nice too.

I need to create a 2d mesh from "thin air" + boundaries (as if by Delaunay_mesher_2) but I must insert a few points which remain constant (that don't disappear or change their coordinates -- as if they were "punctual" constraints).

Here is the issue I can't figure out:
Is it correct to say that, at least conceptually
A) refine_Delaunay_mesh_2 takes vertices as constants, so that, in particular, vertex_handles remain valid throughout the refinement procedure, and that no new vertices are added, whereas
B) Delaunay_mesher_2 invalidates vertex_handles, by possibly adding and/or removing some of them in the process of refinement?

So far this is what I tried:

1: insert boundary nodes and set boundary constraints,
2: run A),
3: insert my "custom" vertices,
4: run B).

Moreover I need to generate two types of output for nodes and links, respectively
ID,x,y
...

ID,ID
...

For that, I keep vertex handles in a map like this:
map<string,Vertex_handle> node_map;

this map is updated  after each step like so: node_map[ID]=Vertex_handle(Vertex_iterator);

which allows retrieving of the vertex coordinates by their ID.

I noticed that during step 4 some of the vertex handles were not matched by an ID which makes me think that A) is incorrect. I checked that those vertices inserted at step 3 were still present. Help....

Thanks,
Rytis