we used to argue about that. I disliked autoconf because I felt 99% of
the work could be precomputed, which is what MIT X11 Makefiles did:
they had recipes for the common architectures.
-G
On Thu, Jun 20, 2024 at 3:15 PM David Arnold <davida(a)pobox.com> wrote:
On 20 Jun 2024, at 08:48, Kevin Bowling <kevin.bowling(a)kev009.com> wrote:
On Wed, Jun 19, 2024 at 11:59 PM Theodore Ts'o <tytso(a)mit.edu> wrote:
<…>
So I did everything using (only) autoconf,
including building and
using shared libraries,
This is The Way if you really care about portability. Autoconf, once you get your head
around what, why, and when it was created, makes for nice Makefiles and projects that are
easy to include in the 100 Linux distributions with their own take on packaging the
world.
For those of a certain era, autoconf was both useful and relatively simple to use.
In an era of many, divergent Unices, with different compilers, shared library
implementations, and varying degrees of adherence to standards, it made using FOSS a
matter of ‘./configure && make && make install’ which was massively easier
than what had been required previously unless you happened to have exactly the same
platform as the author.
And to use it, you needed to understand shell, make, and m4, and learn a few dozen macros
(at most). m4 was perhaps the least likely skill, but since it was used by
sendmail(.mc), twmrc, X11 app defaults and various other stuff, most people already had a
basic understanding of it.
In my view the modern rejection of autoconf as “incomprehensible” mostly suggests that
the speaker comes from a generation that never used the original Unix toolset.
d