From help-octave-request at bevo dot che dot wisc dot edu Fri Dec 19 12:58:17 2003
Subject: Re: Investigating delaunay() segfault
From: Solignac <solignac at altern dot org>
To: Dirk Eddelbuettel <edd at debian dot org>
Cc: help-octave at bevo dot che dot wisc dot edu, rafael@debian.org
Date: Fri, 19 Dec 2003 19:59:03 +0100

On Fri, 19 Dec 2003 10:30:15 -0600
Dirk Eddelbuettel <edd at debian dot org> wrote:

> >   I've encountered several octave crashes using Delaunay's triangulation
> > (qhull).
> > 
> >    Arch: Debian GNU/Linux 3.0r2 (ix86)
> >  Octave: 2.1.50 compiled from source (--enable-shared flag)
> >   Qhull: latest version (2002) compiled from source
> > 
> >   delaunayn.oct is built against libqhull.
> 
> As others have said, all of these are available pre-built for Debian
> testing. Note that you _can_ mix testing and stable; many people do.

  Yes I do know I can mix several dists with "apt-get install
package/unstable"-like commands. That's what I'm used to do with my own
machine (powerpc-arch). I'm not sure it is a good idea to do so on a server.

> Now, for the bug: why don't you send out a _small reprocible example_ so
> that we can see if the bug exists also in the packages produced by Rafael
> and myself?

  Ok, here it is, the code runs OK in matlab-5.3. That's weird, you'll
certainly think I drink too much coffee:

<<<<<
 ## -*- Mode: octave -*-  
 clear all ;
 T = [3.689  70.3 74.239543
      3.689  74.0 74.938502
      3.689  77.7 75.478510
      3.689  81.4 75.906632
      3.689  85.1 76.247444
      3.689  88.8 76.517692
      3.689  92.5 76.730036
      3.689  96.2 76.894593
      3.689  99.9 77.019751
      3.689 103.6 77.112587
      3.689 107.3 77.178749
      3.689 111.0 77.221041
      3.689 114.7 77.233106
      3.689 118.4 77.173737
      3.689 122.1 76.960004
      3.689 125.8 76.638131
      3.689 129.5 76.296164
      7.378  70.3 77.892947
      7.378  74.0 78.386866
      7.378  77.7 78.740757
      7.378  81.4 78.997984
      7.378  85.1 79.18418
      7.378  88.8 79.31554
      7.378  92.5 79.403366
      7.378  96.2 79.456302
      7.378  99.9 79.481573
      7.378 103.6 79.485755
      7.378 107.3 79.47527
      7.378 111.0 79.456664
      7.378 114.7 79.436388
      7.378 118.4 79.41271 
      7.378 122.1 79.083988
      7.378 125.8 78.648553
      7.378 129.5 78.241684 ]  ;

 X = T(:,1) ;
 Y = T(:,2) ;
 #Z = T(:,3) ;
 ##
 A = X + Y ; ## stupid
 ## this one works
 delaunay(A,X) ;
 ## this one fails
 delaunay(X,Y) ;
>>>>>

  Delaunay's triangulation does not seems to work for regulary disposed
points...  Maybe I'm mistaken somewhere...

  Ok I've red qhull's doc a bit more precisely and added 'QJ' flag to
delaunay() instead of 'Qt'; now it does not segfaults (great ;o) ). Does my
code segfaults for you?

  IMHO it may would be safer that delaunay() returns an error message a
little bit more explicative that "core dump" (someting like "Maybe next time
you'll read the man pages" :o) ).

  Thank you very much for your replies.

\bye

-- 
 BOFH Excuse #OAB
 parallel processors running perpendicular today



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------