Date: Tue, 16 Apr 1996 14:24:51 -0600
Message-Id: <199604162024.OAA00336@creche.cygnus.com>
From: Tom Tromey <tromey@creche.cygnus.com>
To: new-httpd@hyperreal.com
Subject: Re: conditional HTML
In-Reply-To: <199604161943.PAA19709@volterra.ai.mit.edu>
References: <199604161943.PAA19709@volterra.ai.mit.edu>
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com
Status: O

Robert> I'll second this.  As long as we're grousing, a similar, well,
Robert> glitch is the way it reruns all the tests even when (IMHO) it
Robert> ought to be able to determine that nothing relevant has
Robert> changed --- in particular, when all that is different is
Robert> --prefix.

Caching should help out here.  If you use caching (all the standard
tests do), then the tests don't actually get re-run -- the results are
just taken from the cache.  This makes re-running configure go much
more quickly (and helps out if you are building a bunch of packages
all at once -- they can all share cache files.  This is what I do).

The prefix and other directory variables are not cached.  No doubt
this is intentional.

Robert> BTW, regarding configure --help, I can think of two ways it
Robert> could be more helpful [...]

You should send these ideas to bug-gnu-utils@prep.ai.mit.edu, or to
Roland McGrath <roland@gnu.ai.mit.edu> (the current autoconf
maintainer).  It would probably be good to put the values of CC,
CFLAGS, etc into config.status, too, if it isn't already done.

Sadly, the autoconf docs also seem to be somewhat lacking when it
comes to describing what environment variables are recognized -- it
only lists the ones used by autoconf itself, and not by the resulting
configure script.

Tom
-- 
tromey@cygnus.com                 Member, League for Programming Freedom