[Plplot-devel] OSF1 is looking pretty good with the new configuration scheme

[Alan W. I. <ir...@be...>]

Joao reported to me just now that the new configuration scheme works pretty
well both with static and dynamic drivers on his OSF1 system!

uname -a
OSF1 alf.fe.up.pt V4.0 1091 alpha

Previously, I had made sure the new configuration scheme worked for an
extremely limited solaris system on the SourceForge compile farm, but it
is really nice to see other good results start coming in as well.

All is not perfect for OSF1 (but it is pretty close....;-)) There are some
OSF1 issues with octave that Joao is still working on.  Also, g++-3.0.x on
OSF1 doesn't work very well with our c++ front end (but that compiler
version is notorious so we expect that).  There are no known issues with
static drivers.  For dynamic drivers there is only one issue, namely, a
change has to be made in the name of the dynamic drivers that are produced
on OSF1.  ("lib" is prefixed to the name, and that screws everything up.) I
am still looking at the most natural way to fix that since there was no such
problem for the dynamic drivers produced in either the Linux or solaris
cases.  Such renaming issues are bread and butter to libltdl so I am sure
there is an easy solution without fooling around with a bunch of mv commands
just for the OSF1 platform, but I have to find what that natural solution
is.

Anyhow, I am extremely pleased with Joao's hard work in iterating through a
fairly large number of rounds of OSF1 tests for me while I made the easy
changes that his tests showed were necessary to the various Makefile.am
files.  Together we have solved many common cross-platform issues so the
next round of testing on a new platform should be much easier.  Who is going
to be the next volunteer?

Even trying the new configuration scheme on a variety of Linux platforms
and/or with a variety of options would be most helpful.

Finally, my autoconf skills are virtually nil so there is also some work to
do in configure.ac and associated files to smooth out the warnings and get
rid of other glitches in the way that the configuration options are
processed.  Help in that area would be most appreciated!

Alan

email: ir...@be...
phone: 250-727-2902	FAX: 250-721-7715
snail-mail:
Dr. Alan W. Irwin
Department of Physics and Astronomy,
University of Victoria, P.O. Box 3055,
Victoria, British Columbia, Canada, V8W 3P6
__________________________

Linux-powered astrophysics
__________________________