Date: Sat, 22 Apr 2000 08:52:16 +0200
From: Ralf Corsepius <[EMAIL PROTECTED]>
This reminds me about proposing that we could consider to merge
http://www.airs.com/ian/configure or a reworked/derived version of it
into autoconf's docs.
[There might be copyright or other restrictions on this article, but ..
Ian?]
It's fine with me. I'm not sure that it necessarily makes sense, as
my notes were intended to describe the GNU configure and build system
as a whole. But if people want to use them, go ahead.
However this would go further than simply extending autoconf
documentation:
* This article describes Cygnus's (Ian's ?) conventions/standards.
* Cygnus's conventions/standards match well with existing means
(AC_CANONICAL_* etc.) in autoconf.
* Autoconf means (AC_CANONICAL_*) do not necessarily imply/require
Cygnus's conventions.
* Autoconf currently does not have explicit cross-compilation
canonicalization conventions, but silently implies Cygnus conventions.
The autoconf support for --host, et. al., were added when DJM was at
Cygnus, specifically to add the features of Cygnus configure to
autoconf. The idea was to merge Cygnus configure into autoconf, and
permit Cygnus configure to be replaced. This has mostly happened over
the years; it is likely that the only remaining use of Cygnus
configure is at the top level of the binutils/gdb/gcc distributions.
I don't know why anybody would want to adopt different conventions.
I'm not even sure what different conventions would look like. It's
not like they are being imposed by Cygnus; they were hammered out with
a number of people, and also drove the gcc configure script ideas
before that was converted to autoconf. Perhaps I'm simply too close
to the problem.
Ian
