Re: [OctDev] convhulln options.

[Rafael L. <rla...@us...>]

* Pedro Tarrafeta <pt...@te...> [2003-12-17 14:52]:

> - Output. The only output the command gives right now is limited and 
>   biased:
>
>   - Biased: Since the output matrix is limited to dimension "dim", extra 
>     vertices for non simplicial facets get discarded. My knowledge of C is
>     very little, but.. ¿is it possible to increase the dimension of the
>     returning matrix (maybe filling it with zeros) in the case of finding a
>     non-simplicial facet?. It would be something in line 97 like that:
>
>         if (j > dim) {
>             dim = dim + 1
>             redim idx(n, dim)
>             for k = 1 to i-1 {
>                 idx (k, dim) =  0
>             }
>             idx(i,j++) = 1 + qh_pointid(vertex->point)
>          }

Instead of doing that, we might return a cell array instead of a matrix when
non simplicial facets are found.  For instance, the following result:

    $ cat cube
    3
    7
    0 0 0
    0 0 1
    0 1 0
    0 1 1
    1 0 0
    1 0 1
    1 1 0
    $ cat cube | qhull o
    3
    7 7 12
         0      0      0 
         0      0      1 
         0      1      0 
         0      1      1 
         1      0      0 
         1      0      1 
         1      1      0 
    4 2 3 1 0 
    3 6 3 2 
    4 4 6 2 0 
    3 3 5 1 
    3 5 6 4 
    3 6 5 3 
    4 5 4 0 1 
    
would be represented in Octave like this:

    A = {[3 4 2 1],
         [7 4 3],
         [5 7 3 1],
         [4 6 2],
         [6 7 5],
         [7 6 4],
         [6 5 1 2]}


(Notice that vertex indices in Octave start at 1 instead of zero.)

I do not know how backward incompatible would be such a change.

> - Limited: Parsing a string to qhull allows to obtain some other info on
>   the hull. Some info that is specially interesting for me is the normals
>   to each facet (option n). Since qhull can return various output sets is
>   it possible that it returns an octave struct with the different results?.
>   (I have tried to parse to qhull the "TO file" string so that I could get
>   the normals from the file created but it does not work. This could also
>   be a solution, in fact THE solution in the case of OFF output to feed
>   Geomview).

To avoid backward incompatible changes, what about having extra output 
arguments, one for the facet normals, other for the OFF string result, and
so on?

-- 
Rafael