Re: Compiling with Qhull

[John W. Eaton]

On 13-Jan-2012, Ben Abbott wrote:

| On Jan 13, 2012, at 2:17 PM, Muhali wrote:
| 
| >> I tried that version on MacOS 10.7.2 with gcc 4.5, but encountered an error
| >> with the "voronoi" diagram ...
| >> 
| >>    ../../run-octave -f -q -H -p . --eval "geometryimages ('voronoi', 
'eps');"
| > 
| > I have the same problem now when trying to compile the current stable tip
| > (81ee6b12adbc). Here is what I get (running Ubuntu Natty):
| > 
| > make[3]: Entering directory `/usr/src/octave/doc/interpreter'
| > ../../run-octave -f -q -H -p . --eval "geometryimages ('voronoi', 'eps');"
| > panic: Segmentation fault -- stopping myself...
| > attempting to save variables to `octave-core'...
| > save to `octave-core' complete
| > make[3]: *** [voronoi.eps] Segmentation fault
| > 
| > M.
| 
| What version of qhull do you have installed ?
| 
| If it is qhull-2011, you'll want to install the 2010 version, or apply the 
attached patch.

It would be great if we could have some way to determine automatically
whether

  #define qh_QHpointer 1

is needed.  I'm not sure how to do that.  I'd prefer to avoid checking
version numbers, as we have no way of knowing whether this might
change again with some future version of Qhull.  And even if we did
check versions, is this definition always needed for Qhull 2011 and
never needed for earlier versions?  Or does it depend on how Qhull was
configured/compiled?  So I would prefer an autoconf test that does not
check versions.  To make it work better for cross-compiling
situations, it would be best if we could test this feature simply by
compiling a program without having to run it.  But if that's not
possible, then I would settle for a test that requires running a
program.

jwe