Message-ID:  <199511171945.OAA01067@aleph.bu.edu>
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@VM.URZ.UNI-HEIDELBERG.DE>
In-Reply-To:  Your message of "Fri, 17 Nov 1995 16:21:15 +0100."
              <199511171521.QAA19884@frank.zdv.uni-mainz.de>
Date:         Fri, 17 Nov 1995 14:45:27 -0500
From: Matthew Swift <swift@BU.EDU>
Sender: Mailing list for the LaTeX3 project
              <LATEX-L@VM.URZ.UNI-HEIDELBERG.DE>
To: Multiple recipients of list LATEX-L
              <LATEX-L@VM.URZ.UNI-HEIDELBERG.DE>
Subject:      Re: distribution conditions on packages and other LaTeX software
Status: R

>>>>> "F" == Frank Mittelbach <mittelbach%MZDMZA.ZDV.UNI-MAINZ.DE@BUACCA.BU.EDU>
 writes:

    F> btw the biggest contribution in the winter release will be a
    F> complete rewrite of docstrip which now will produce up to 16
    F> (surprise surprise :-) files within one go. This will be
    F> extended (probably by june) with a configuration mechanism
    F> which then will allow to even distribute the unpacked files
    F> into different directories!

It sounds like an excellent improvement; is a summary of its syntax
available?  I will wait for this before releasing a revision of my
emacs mode for doc.sty documents.  (Still no integration with AUC-tex
-- it's too big a task for now -- but I'm working on compatibility
(not easy either).)

I am still vaguely troubled by the the self-contradiction at the core
of docstrip, namely that the model

    latex foo.sty             -=>  foo-documentation.dvi
    \includepackage{foo}

is mutually exclusive with the (more common) model:

    latex foo.dtx             -=>  foo-documentation.dvi
    tex foo.ins               -=>  foo.sty bar.sty baz.sty
           ^
           |
        foo.dtx

    \includepackage{foo, bar, baz}

The goal if I see things rightly is to have a syntax on which TeX
itself can be the equivalent of both weave(1) and tangle(1).  The two
models above are quite different solutions.  Docstrip implements
both. Neither is superior in every way.  You could have a much simpler
syntax and parser if you wrote a solution for model two that didn't
also have to implement model one.  It could be written in TeX or even web.

Now, I know you shouldn't fix something that's not broken, but there
are two reasons to think about it.  I find the practical details of
authoring docstrip files pretty obtuse (number of %'s, worrying about
\iffalse's and \fileinfo's, the column of \endinput, etc.).  Also,
David Carlisle emphasized the speed of the new docstrip.  If that's a
concern, it might be worth it.

regards

  matt