axiom-developer
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Axiom-developer] autoconf revisited


From: Tim Daly
Subject: [Axiom-developer] autoconf revisited
Date: Mon, 8 Aug 2016 22:27:52 -0400

John,

...[snip]...

re: why doesn't Axiom use autoconf.

Poul-Henning Kamp wrote an essay that contains the essence
of the reply:
A Generation Lost in the Bazaar
http://queue.acm.org/detail.cfm?id=2349257

  "Even more maddening is that 31,085 of those lines are in a
   single unreadably ugly shell script called configure. The idea
   is that the configure script performs approximately 200
   automated tests, so that the user is not burdened with
   configuring libtool manually. This is a horribly bad idea,
   already much criticized back in the 1980s when it appeared,
   as it allows source code to pretend to be portable behind the
   veneer of the configure script, rather than actually having the
   quality of portability to begin with. It is a travesty that the
   configure idea survived."

autoconf adds languages like M4, shell scripts, and makefiles.
It hardens the shape of the source tree. New versions are
released that break old versions. The "new" is added and
the "old" is never removed. It is complexity by accretion,
black paint over veneer.

Axiom's goal has been, and continues to be, removing
dependencies. Meta is gone. Boot is gone. Noweb is gone.
The directory-based pile-of-sand (POS) is disappearing into
books.

In the reasonable future C will be gone. Work is underway
to build hyperdoc and graphics in the browser (with
minimal standards-based _javascript_/xhtml).

Make will be gone as more of the work is merged into lisp.

Latex exists with only minimal chunk support macro
extensions (a few dozen lines).

The eventual goal is Axiom as a literate, common lisp
program built using lisp with an optional browser front-end
served from lisp.

Simplify, simplify...

Tim
 

reply via email to

[Prev in Thread] Current Thread [Next in Thread]