Skip to Content.
Sympa Menu

cgal-discuss - Re: [cgal-discuss] Anonymous SVN access Jet fitting

Subject: CGAL users discussion list

List archive

Re: [cgal-discuss] Anonymous SVN access Jet fitting


Chronological Thread 
  • From: Thorben Kröger <>
  • To:
  • Subject: Re: [cgal-discuss] Anonymous SVN access Jet fitting
  • Date: Mon, 19 Oct 2009 07:09:20 +0000

On Thursday 15 October 2009 11:20:47 Marc Pouget wrote:
> ok, then if you send me the data for the puzzling plane example, I
> may try to dig this out. I have no more clue so far.

After lot's of debugging, I have found the root cause of this: MATLAB.
The problem was that the cgal code links to lapack, but Matlab ships it's own
version of lapack. So when I ran the same code outside of matlab, it picked
up
my correct system lapack library, while the same code running inside of
matlab
saw matlab's version in the path first and used that. I don't know why, but
somehow that version seems to be incompatible, and so the singular value
decomposition just did nothing when run from matlab.

My solution/workaround was to delete the matlab installed library and to
symlink it to my system library.

So, thanks for listening :-)

Thorben

>
> Best
> Marc
>
> On 14 , Oct , 2009, at 12:08 , Thorben Kröger wrote:
> > On Wednesday 14 October 2009 09:49:02 Marc Pouget wrote:
> >> Puzzling!
> >>
> >> can you try a real planar plane
> >
> > Attached smooth_plane.png calls the same matlab interface function
> > that I used
> > in my previous experiments. That looks very much OK.
> >
> >> and a simple smooth surface
> >> (parabolic cylinder or elliptic paraboloid there is a picture in the
> >> manual 53.2)
> >
> > Attached file smooth_ellipsoid.png This looks ok as well.
> >
> >> to see if the results are as expected and hence check
> >> there is not pb in the interface?
> >
> > Cheers,
> >
> > Thorben
>



Archive powered by MHonArc 2.6.16.

Top of Page