[Gnu-arch-users] Re: autoconf

[Stephen J. Turnbull]

>>>>> Sascha Silbe writes:

    >> So my recommendation is: If you can avoid using autoconf, do
    >> it.  AFAIR there are alternatives, but I had no need for one in
    >> the past (for all my projects a simple Makefile suffices),
    >> so I did not evaluate them.

Alternatives with a similarly huge database of different systems and
features?  That's the big advantage to autoconf, AFAICT: it knows
about almost everything XEmacs can build on (Windows NT native,
lacking a POSIX shell, is the single exception), and (if we could only
use autoconf 2.5x) most of the features.

>>>>> "wave" == wave  <wave++> writes:

    >> So my recommendation is: If you can avoid using autoconf, do
    >> it.  AFAIR there are alternatives, but I had no need for one in
    >> the past (for all my projects a simple Makefile suffices),
    >> so I did not evaluate them.

    wave> Autoconf scripts tends to autodetect things without keeping
    wave> in mind particular situations. IT'S TRIVIAL to do autoconf
    wave> tests that BREAKS "portability". It's also a pain to trick
    wave> it, if you have particular requirements.

We really haven't found this to be a problem.  Basically, a convention
of wrapping everything that is a workaround in

dnl For systems with breakage in FEATURE-X, add cases here.
case $sys in
  *a-broken-system* )    dnl work around breakage
    ;;
esac

prevents autoconf-induced portability problems.  If that doesn't work,
then there is an underlying portability problem anyway.  That's not
autoconf's fault, although it would be nice if provided a
straightforward way to deal, rather than being an obstacle as it
sometimes is.

As for tricking it, I dunno.  But I suppose you're right, as we're
still using 2.13 since eliminating or porting our various autoconf
hacks to 2.5x is turning out to be tedious.


-- 
Institute of Policy and Planning Sciences     http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.