CGAL users discussion list

Again following up on non-manifold Nef3 decoration. This has become quite
a bit more technical than it should for this forum. I'm posting an update
just to keep the thread consistent.

What I found that I had to do to guarantee consistent
halfedge_around_vertex adjacency ordering when constructing spheremaps in
non-manifold situations was to add another overload to SM_decorator in the
form:

   SHalfedge_handle new_shalfedge_pair(SVertex_handle v1, SVertex_handle
v2, const Sphere_direction sd);

This assumes that geometry has been set up for existing svertices and
shalfedges. It will simple link the new halfedges in such a manner to
respect strict ccw halfedge ordering at V1 and V2 respectively. The
Sphere_direction is required to uniquely describe the sphere segment.

This appears to fix my remaining construction issues, but leaves me with a
couple of questions for Peter (when/if he shows up):

- My impression is that the existing SM decoration implementation is
incorrect because it doesn't enforce shalfedge ordering. I'm guessing this
just hasn't been a problem up to now because it's only ever been tested
for manifold constructions - i.e. exactly two shalfedges per svertex. Is
this correct?

- Does my approach make sense, e.g. have the decorator perform what
amounts to an insertion sort based on the strictly_ordered_ccw_at()
predicate at the source svertices? The decorator otherwise seems concerned
only with adjacency. The geometric considerations could be hidden away in
a sedge-ordering functor given as a parameter.

Regards,
Fred