Reply-to: Mailing list for the LaTeX3 project <LATEX-L%DHDURZ1.BITNET@vm.gmd.de>
Message-id: <01HP12YC7Q9E9TDEI9@MZDMZA.ZDV.UNI-MAINZ.DE>
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Content-transfer-encoding: 7BIT
Date: Thu, 06 Apr 1995 18:14:33 +0100
From: Philip Taylor <CHAA006@alpha1.rhbnc.ac.uk>
Sender: Mailing list for the LaTeX3 project <LATEX-L%DHDURZ1.BITNET@vm.gmd.de>
To: Multiple recipients of list LATEX-L <LATEX-L%DHDURZ1.BITNET@vm.gmd.de>
Subject: Re: help needed for placing figures at the right place
Status: R

Dear Frank:

Although the following proposal may be completely incompatible with the
intended implementation of L3, I none the less feel that it is sufficiently
important to at least propose it publicly, even though I know in advance that
it will almost certainly be shot down:

I would like to propose that rather than being basically monolithic, with
modules only coming into play at the <package> level as in the New Standard
LaTeX (TM), LaTeX-3 be completely modular in nature, and furthermore that
those modules be based on the smallest possible kernel.

There are several reasons for this proposal apart from those obvious from
its nature: they include, for example, the current discussion raging on
Gutenberg about the relative merits of TeX and LaTeX; discussions with
Marek Rycko & Boguslaw Jackowski, both of whom are staunch advocates of
a `library' approach to TeX functionality rather than that of a monolithic
package; the current received wisdom in computer science concerning the
importance of re-usable code; and my own experiences of the (relative)
difficulty of modifying the appearance of a LaTeX document when compared
to the difficulty of achieving the same changes in a Plain TeX document.

In summary, LaTeX is full of wonderful ideas but is a pig to modify;
if only those wonderful ideas could be used in the context of a Plain
TeX document, one could have the best of both worlds.  For example,
I have tried (unsuccessfully) to extract the NFSS system from the New
Standard LaTeX; but its (undocumented?) dependencies leave me floundering:
there seems no end to the amount of code which I have to copy verbatim
from the LaTeX sources in order to get a working NFSS.  How wonderful
it would be if there were a file Nfss.Ltx, which started

        \needs {Ltx-Krnl}

and I would then know that all I needed in order to get NFSS functionality
in my Plain TeX document was Nfss.Ltx and Ltx-Krnl.TeX.  Even if it
started

        \needs {<Some other package>}

I could at least track back the chain of dependencies, and copy only the
modules that were truly needed.

OK, so it's radical: is it also totally off the wall?

                                        Philip Taylor, RHBNC.