CGAL users discussion list

[Jane Tournois <>]

Insertion is done before removal, but the logic remain the same.

The function that actually does the job is

Vertex_handle
  move_point_topo_change(const Vertex_handle& old_vertex,
                         const Point_3& new_position,
                         OutdatedCellsOutputIterator outdated_cells,
                         DeletedCellsOutputIterator deleted_cells) const;

this function collects the removal AND insertion conflict cells,

then, it inserts `new_position`, removes `old_vertex`,

and finally collects the outdated cells with the function 
`get_conflict_zone_topo_change` : the union of cells incident to the 
newly created vertex, with the cells in conflict with the re-insertion 
of `old_vertex`'s position.

It is indeed equivalent to what I said before.

Best,

Jane.



Le 09/09/2016 à 18:36, Brown, James 2 a écrit :
> >    Vertex_handle move_point(const Vertex_handle& old_vertex,
> >                             const Point_3& new_position,
> >                             Outdated_cell_set& outdated_cells_set,
> >                             Moving_vertices_set& moving_vertices) const;
> > does exactly what you need.
> Yes, this is the function that I am looking at.
>
> > - collect the incident cells to v
> > - remove v
> > - triangulate the cavity --> triangulation T1
> > - collect cells in conflict with the insertion of newpoint : T2
> > - insert newpoint
> > - collect incident cells to new vertex : T3
> > the newly created cells after removal + insertion is T1-T2+T3.
> Sorry if I am missing something, but this is not quite the process that is 
> happening in move_point (though it must be equivalent?):
> - In move_point the new (outdated) cells are just the conflict cells of the 
> old position in the final triangulation.
>   - As well it is doing removal after insertion of the new point (so the 
> insertion conflicts are calculated from the original triangulation before 
> removal)
> Is move_point taking a shortcut? It is not obvious to me that it yields the 
> same results as your suggested process.

--
Jane Tournois, PhD
R&D Engineer at GeometryFactory
http://www.geometryfactory.com/