Re: [Plplot-devel] Make of plplot-5.2.0 fails on a freshly installed Red-Hat-8.0.

[Maurice L. <mj...@ga...>]

Alan W. Irwin writes:
 > On Mon, 27 Jan 2003, Rafael Laboissiere wrote:
 > 
 > > This is a very serious bug and I will try to find some time tonight to try
 > > to find a decent solution.  I agree with Alan that this must be fixed
 > > properly.  Workarounds today will only result on headaches tomorrow.
 > >
 > > I think that if/when this bug is fixed, a new 5.2.1 tarball will have to be
 > > released.
 > 
 > I suggest using a dual-track short-term and long-term strategy for solving
 > the problem.  Short-term so that those who have already downloaded 5.2.0
 > can get something out of it, and long-term for a more fundamental solution.
 > 
 > (1) Short-term (time scale of a day or so).  Joao and Maurice, will you
 > please try the suggested workaround to make sure it cures all the problems
 > you found? With the current plplot-5.2.0 tarball simply unpack it and touch
 > include/plDevs.h.in before running configure; make; make install.  If I am
 > right about this "touch" workaround, there should be absolutely no attempt
 > to run autoheader (or any other autotool) at the make stage.  I will do the
 > same test here (this evening after work).  If we all find no problems, then
 > I will immediately publish that workaround to plplot-general (and in the SF
 > release notes for 5.2.0) and also indicate we are working on plplot-5.2.1
 > which will solve the problem at a more fundamental level.

I've verified this approach does work on my RH 7.3 system.

 > (2) Longer term (time scale of a week or so). Find a fundamental solution. I
 > believe this is absolutely necessary because from my further reading of the
 > documentation of AM_CONFIG_HEADER and its autoconf counterpart
 > AC_CONFIG_HEADER, it is amazing our use of two files in the argument list
 > for AM_CONFIG_HEADER actually worked at all (albeit with the date problems
 > for plDevs.h.in).  There is still a lot of change going on with both
 > automake and autoconf so I would be most surprised if our two-file solution
 > worked in the slightest for future autoconf/autoheader versions.

I'm not sure about AM_CONFIG_HEADER, but AC_CONFIG_HEADER definitely works
given multiple args.  Probably AM_CONFIG_HEADER does too since it's just
a front-end to AC_CONFIG_HEADER.  There's also now some *_CONFIG_HEADERS
(note the trailing "S") but I don't know if these are the same or what.
Man, what a mess.

Anyhow, the current cvs HEAD configuration is now badly broken.  Adding
plDevs.h to AC_OUTPUT doesn't work -- headers have to be treated specially.
So we should go back to the previous version of configure.ac until this
is all sorted out.

-- 
Maurice LeBrun    mj...@ga...
Research Organization for Information Science and Technology of Japan (RIST)