CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

Edoardo Milotti wrote:
> in my application I triangulate large sets of randomly scattered points 
> (in 3D), and I must keep track of all of them. Since insertion of the 
> whole vector of points also shuffles them at random to avoid 
> degeneracies, I cannot use this direct method, and I have to insert them 
> one by one to keep the original order. Since in my case the points are 
> in general position and I don't have to deal with degeneracies I would 
> like to use the faster insertion of the whole vector: is there a way to 
> disable the internal shuffling?
>
> Edoardo Milotti
Hello,

The insert by range method has nothing to do with geometric degenerate 
cases. The random shuffle of the input points is made to have a kind of 
uniform random distribution of points in the space. Then a spatial 
sorting algorithm is applied to input points so that upon insertion of 
an element, a cell of this element can be used as an hint to insert the 
next element (which makes the speed up).
Thus even without the shuffling, points are sorted and not inserted in 
the range order.

You can associate to each inserted point an index so as to keep track of 
them. We should have an integrate solution for this kind of problem in a 
future release of CGAL (the one after 3.6).

In the meantime, you could refer to this post 
https://lists-sop.inria.fr/sympa/arc/cgal-discuss/2010-01/msg00004.html
for a temporary solution.

S.

> ----------------------------------------------------------------------------
>
> Edoardo Milotti
>
>
> Address: 
>
> Dipartimento di Fisica dell'Università di Trieste
>
> and I.N.F.N. - Sezione di Trieste
>
> Via Valerio, 2
>
> I-34127 Trieste, Italy
>
>
> e-mail: 
>
>  
> <mailto:>
>
> phone: +39 040 558 3388
>
> fax: +39 040 558 3350
>
>
> http://www.ts.infn.it/~milotti/index.html
>
> ----------------------------------------------------------------------------