Re: Qhull 2012.1 questions

[David Sterratt]

Dear Brad &al,

I'm fine with all your changes as I don't actually use the build system
for incorporating Qhull into the R geometry package.

All the best,

David.

On Thu, 2012-02-09 at 23:18 -0500, Brad Barber wrote:
> Hi all,
> 
> Marco proposed the following changes to Qhull as part of his cygwin build.  
> Most of them look OK to me.
> 
> Please review the following changes.  Let me know which ones are good and 
> which are not so good.   I'd like to use the same conventions for all builds. 
>  There will probably be a rerelease of Qhull 2012-1 next week.
> 
> 1)  Move installed documentation from 
>      /usr/share/doc/packages/qhull
> to
>      /usr/share/doc/qhull
> 
> 2) Move installed man pages from
>     /usr/man/man1
> to
>     /usr/share/man/man1
> 
> 3) Remove the installation of user_eg* and testqset.  Continue to build them 
> into Qhull's local bin directory.
> 
> 4) Remove cmake's target VERSION from executables
> 
> 5) Set cmake's SOVERSION to 6 and remove $qhull_MAJOR.  The dll's would have 
> the following names.  A source of confusion is that qhull 2011 produced 
> qhull6.dll for qh_QHpointer.   Any users would need to refer to qhull_p-6.dll 
> (and not! qhull-6.dll).
> 
> >  /usr/bin/cygqhull-6.dll
> >  /usr/bin/cygqhull_p-6.dll
> 
> 6) Merge Road*.h and Road*.cpp from 
>      src/road/...
>      include/road/...
> to
>     src/libqhull/...
>     include/libqhull/...
> 
> 7) Keep libqhullstatic.a.   Marco prefers 'libqhull.a' and 
> CLEAN_DIRECT_OUTPUT.    Using libqhullstatic.a avoids conflicts between 
> libqhull.a and libqhull.so.   Renaming libqhull.so to libqhull-6.so would 
> remove this conflict.   Does anyone else prefer libqhull.a?
> 
> 8) Continue to build the qhull executables as static builds.   It makes them 
> standalone and simplifies moving them elsewhere.
> 
> 9) Retain the prolog for q_eg, q_egtest, and q_test.  It checks that the 
> script is launched in the top-level directory.   Maybe change the tests to 
> error only if qhull is inaccessible.
> 
> 10) Redo the error message for nearly-cospherical Delaunay sites (e.g., rbox 
> c | qdelaunay).   The option 'Qz' is particularly valuable in this case.
> 
>                         --Brad

-- 
David C Sterratt, Research Fellow http://homepages.inf.ed.ac.uk/sterratt
Institute for Adaptive and Neural Computation     tel: +44 131 651 1739
School of Informatics, University of Edinburgh    fax: +44 131 650 6899
Informatics Forum, 10 Crichton Street, Edinburgh EH8 9AB, Scotland  
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
NEW BOOK: Principles of Computational Modelling in Neuroscience
Sterratt, Graham, Gillies & Willshaw (CUP, 2011).
http://www.compneuroprinciples.org  


The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.