I think that just eliminating this void* data feature and providing an example of copying the diagram into a different graph data structure and looking up input sites in a map to populate some field would be the best way to handle this. I don't know that we currently have code that uses this feature of the cell/edge, do we Andrii?
Well, I have examples in the basic tutorial on how this data member could be used. Personally I used it to do DFS on the Voronoi graph, by setting visited edges data pointer to themselves. I found this functionality to be very handy: void dfs(const VD::edge_type* edge) { if (edge->data()) return; // do some edge processing here. edge->data(&edge); edge->twin()->data(&edge); const voronoi_diagram<double>::edge_type* e = v->incident_edge(); do { dfs(e); e = e->rot_next(); } while (e != v->incident_edge()); } I think building the Delaunay triangulation of a set of points from the
voronoi diagram data structure should be provided as example code in the documentation in addition to being a documented API that populates a collection of polygon_concept with triangles. That way people can adapt the example code to populate their own mesh data structure if a container of polygon_concept doesn't map well to what they need. We can make the example code use concrete types for ease of reading and implement the API as a generic algorithm with concept overloading on polygon_concept that would be confusing if shown in example code.
True. I am also going to implement dedicated Delaunay triangulation data structure as soon as I have some time. Regards, Andrii