Date:	Wed, 7 Jan 1998 11:17:15 +0200 (IST)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To:	John Velman <jrvelman AT mail DOT hac DOT com>
cc:	djgpp AT delorie DOT com
Subject:	Re: Guile again, and "no accepatable ld.."
In-Reply-To:	<9801062103.AA13451@hacgate2.hac.com>
Message-ID:	<Pine.SUN.3.91.980107111448.4858M-100000@is>
MIME-Version:	1.0

On Tue, 6 Jan 1998, John Velman wrote:

> I think I need
> to do some reading about autoconfig and m4 in order to try to find out
> what is going on.  The part of the configur script that is looking for 
> ld seems to start out in aclocal.m4.

Actually, you don't need to know about Autoconf to understand this
problem.  Autoconf is just a way to generate the configure scripts
without reinventing the gory details of how to test for certain
features which lots of programs need.  It works by using the m4 macro-
processor, whereby the maintainer of the package writes a macro, like
"AC_CHECK_FUNCS(bcopy getcwd getwd)", and m4 expands this into
commands that are run by the shell (Bash) to test for the availability
of the named functions in the C library of the target machine.

The output of Autoconf is a huge shell script called "configure", and
it is the part of that script that tests for LD which is of interest.
Looking at the *.m4 files isn't always enough to understand what's
going on, because guessing into what the m4 macros will expand is not
always easy.

> >See my other message for what I think is a better way to solve this.
> 
> Tried it.  Same results.

Then the "test -f" is not the problem.  That script is testing for
something that I don't understand, and have never seen before.  (Why
would a package care about the linker invoked by GCC?)

> Eli - I'll send you the parts of configur that are checking for LD as a 
> separate message.

Please send the parts of configure script itself, not of the *.m4
files which are input to Autoconf.

- Raw text -