Comments on building GrADS from source

[Patrice Dumas]

> Re: GrADS configure system vs. "other" packages

In the changes I proposed I left absolutely untouched what happens when
--enable-dyn-supplibs isn't given, I didn't touched a bit of acinclude.m4
(I didn't even corrected obvious m4 quoting and typos to help reviewing,
I'll propose a patch for these, but only after the other changes are
included).

The reverse apply when --enable-dyn-supplibs is given: with my system,
all the library detection stuff is handled by new macros when
--enable-dyn-supplibs is given, and the acinclude.m4 aren't used anymore
in that case.

What I had in mind is that COLA/Jennifer can customize what happens when
--enable-dyn-supplibs isn't given, such that it plays well with the
provided static supplibs. On the other hand when --enable-dyn-supplibs is
given the macros are much more complicated, as they have to adapt to
many setups and library versions, but they may be maintained by the
community, especially the packagers who try to do binary packages adapted
to their platform using their libraries. That's exactly what I did for
fedora: trying to package grads let me see where better detection could be
done.

--
Pat