RE: Release candidate for Qhull 2012.1

[Brad Barber]

Hi Melvin,

Many thanks for testing convhulln on Lion and Fedora.

Your results for Lion indicate that Qhull triangulated the output instead of 
retaining the cube's facets.  My guess is that for Lion, someone had set 
options 'QJ' or 'Qt' which request triangulated output.  You can test this 
case, by adding 'FO' to the option list (look for Qtriangulate or QJoggle in 
the returned "Options selected").  

Try running the qhull program directly on Lion.   Here are the results with and 
without triangulation.  

# Convex hull of a cube without triangulation
rbox c | qhull i
6
4 6 2 0
1 5 4 0
5 7 6 4
3 1 0 2
7 3 2 6
3 7 5 1

# Convex hull of a cube with triangulation
rbox c | qhull Qt i
12
6 2 0
4 6 0
5 4 0
1 5 0
5 6 4
6 5 7
2 3 0
3 1 0
6 3 2
3 6 7
3 5 1
5 3 7

If you do not find the reason, please send me the log file from option T4.

                                --Brad

At 01:13 AM 1/29/2012, you wrote:


>________________________________________
>From: address@hidden address@hidden on behalf of Brad Barber address@hidden
>Sent: Saturday, January 28, 2012 11:54 AM
>To: address@hidden
>Subject: Release candidate for Qhull 2012.1
>
>Hi all,
>
>See below for the Qhull 2012.1 release candidate.  It is not deployed to 
>qhull.org, because I need feedback on the changes.   The goal of this release 
>is to clean up the builds and make them suitable to Octave, R, and other users.
>
>If something is broken for you in Qhull 2012.1, please let me know.
>
>Qhull 2012.1 adds libqhull.so for Octave and other users.  This shared library 
>uses a static global data structure .   It is the same as the debian build, 
>Qhull 2009.1.  To help avoid confusion, libqhull6_p.so is the new name for 
>qhull with a dynamic global data structure (-Dqh_QHpointer).
>
>Qhull 2012.1 fixes strict aliasing in qset.c and adds a qset test program.
>
>
>Qhull 2012.1 includes a draft debian build in the 'config/' directory.  
>Someone will need to verify that it works.   As before, it produces 
>libqhull.so and the qhull executables.  It could be updated to match the other 
>builds.
>
>A later version of Qhull may include a third interface to get rid of global 
>variables.  It would pass qh_qhT as a parameter to each function call.  There 
>would be a single source tree using a macro to enable the extra parameter.  
>There would be a third shared library, tentatively named libqhull7_t.so for 
>"threading".   If so, the C++ interface will probably switch to this library.
>
>   http://gitorious.org/qhull/
>   http://gitorious.org/qhull/qhull/blobs/master/src/Changes.txt
>   http://www.qhull.org/download/qhull-2012.1-src-tgz.md5sum
>   http://www.qhull.org/download/qhull-2012.1-src.tgz
>   http://www.qhull.org/download/qhull-2012.1-zip.md5sum
>   http://www.qhull.org/download/qhull-2012.1.md5sum
>   http://www.qhull.org/download/qhull-2012.1.zip
>
>Many thanks for helping with qhull.
>
>                                        --Brad
>
>
>
>I built Octave on Lion with this version of Qhull and this went OK.  It did 
>fail one of the make check tests on convhulln:
>octave:1> test convhulln
>  ***** testif HAVE_QHULL
> cube = [0 0 0;1 0 0;1 1 0;0 1 0;0 0 1;1 0 1;1 1 1;0 1 1];
> [h, v] = convhulln (cube);
> assert (size (h), [6 4]);
> h = sortrows (sort (h, 2), [1:4]);
> assert (h, [1 2 3 4; 1 2 5 6; 1 4 5 8; 2 3 6 7; 3 4 7 8; 5 6 7 8]);
> assert (v, 1, 10*eps);
>!!!!! test failed
>assert (size (h),[6, 4]) expected
>   6   4
>but got
>   12    3
>values do not match
>
>The same error also occured with the Qhull package from MacPorts (2011.2).  
>The difference was that I did not have to apply Ben's patch using when 
>building against this version.  Not applying Ben's patch to 2011.2 resulted in 
>many more make check failures.  
>
>Any idea what could be happening to cause this to fail on Lion, but pass on 
>Fedora?  
>
>
>
>-----
>No virus found in this message.
>Checked by AVG - www.avg.com
>Version: 10.0.1416 / Virus Database: 2109/4773 - Release Date: 01/28/12