Re: [OctDev] convhulln options.

[Pedro T. <pt...@te...>]

Rafael,

> I just cvs committed the changes for the convhull and convhulln functions.
> They accept now an extra (string) argument containing extra options to be
> passed to the underlying qhull command.

Sorry for the delay but I've been away a few days. Let me point a few thing=
s=20
about the new convhulln:
        - It accepts a string but it is added to the default "s Qt Tcv".I=20
would rather use these options only in the case in which a string hasn't be=
en=20
parsed, so that I can use QJ. I don't know the reason but Qt is not returni=
ng=20
the same output from octave than outside octave (if you test qhull with a 3=
D=20
cube you will get 12 triangulated facets from qhull and 6 facets and extra=
=20
vertices from within octave).
        - Output. The only output the command gives right now is limited an=
d=20
biased:
                - Biased: Since the output matrix is limited to dimension=20
"dim", extra=20
vertices for non simplicial facets get discarded. My knowledge of C is very=
=20
little, but.. =BFis it possible to increase the dimension of the returning=
=20
matrix (maybe filling it with zeros) in the case of finding a non-simplicia=
l=20
facet?. It would be something in line 97 like that:
                        if (j > dim) {
                                dim =3D dim + 1
                                redim idx(n, dim)
                                for k =3D 1 to i-1{
                                        idx (k, dim) =3D  0
                                }
                                idx(i,j++) =3D 1 + qh_pointid(vertex->point)
                        }
                (As you can see my knowledge of C is scarce and I've=20
reinvented the syntax=20
;) ).
                - Limited: Parsing a string to qhull allows to obtain some=
=20
other info on the hull. Some info that is specially interesting for me is t=
he=20
normals to each facet (option n). Since qhull can return various output set=
s=20
=BFis it possible that it returns an octave struct with the different resul=
ts?.=20
(I have tried to parse to qhull the "TO file" string so that I could get th=
e=20
normals from the file created but it does not work. This could also be a=20
solution, in fact THE solution in the case of OFF output to feed Geomview).

Regards,
Pedro



                               =20


>
> I did some minimal tests, but additional checks are welcomed.  The next
> step will be changing the voronoi* functions.