4.0081 TeX (50)

Elaine Brennan & Allen Renear (EDITORS@BROWNVM.BITNET)
Sat, 19 May 90 20:04:46 EDT

Humanist Discussion Group, Vol. 4, No. 0081. Saturday, 19 May 1990.

Date: Thu, 17 May 90 23:40:44 -0400
From: amsler@flash.bellcore.com (Robert A Amsler)
Subject: TeXies

Excuse my not replying sooner to Sebastian's preemptive strike against
TeX criticism in any form. In part it was due to my being a little
upset when he took a message I sent him off-line to use as the source
for extracting bits to use in his rebuttal. It wasn't so much that I'd
said anything in that message that couldn't have been said to Humanist
at large, but that I was myself replying to an off-line message he had
sent me and frankly didn't think this type of TeX discussion was worth
everyone's time on Humanist. His reply didn't convince me I was wrong.

It seemed a little strange to have Sebastian lambasting Scribe while
admitting he had never used it. Probably the appropriate stand for a
religious warrior; ``Of course I don't know anything about THEIR
system--do you think I want to be contaminated by infidels?''

There are a couple things I would like to say. I just finished a rather
small Snobol4 program to convert a particular 150 page document from
Waterloo GML (pretty much the closest working typesetting system to SGML
from what I have seen) and it once again proved to me the rightness of
the SGML-level of text description. The conversion was basically a set
of simple substitutions with some bells and whistles thrown in for the
attributes of some tags. I would attribute this in part to Scribe's
high-level operators. It most certainly could also have been done in
TeX, but that it exactly the whole point. From the higher level
langauges one can translate downward easily into the lower level
languages. The important principle is to believe in the worthiness of
that translation step for the future of machine-readable text.

It is really tempting to fall under the sway of a single typesetting
system and come to believe that it has reached perfection itself because
it lets you manipulate everything at the level of XXXX. I mean, Fat
Bits on the Mac are wonderous to add that Incunabula look to a document
to make sure each letter E is slightly different; simulating the wearing
down of wooden type. (And Sebastian would add.... ``Yes, that's what I
like about TeX, it has Metafont available to make it possible to do
things like that more professionally than on the Mac'')

The point is that it is very hard to climb back UP to the high-level
description from the more specialized forms. You have to infer what the
author intended by those little changes they invariably seem prone to
make. As someone who spends time trying to UN-typeset documents to make
their text usable for things like database extraction and extracting
useable data, I can assure you it is worse to try and unravel TeX and
troff than anything short of Postscript and Impress.

(And for Sebastian: Scribe is indeed a system for designing document
styles and does let you ``sink down into raw typesetter speak'' to
specify things like character widths, select individual characters to
make up your own alphabets and font families. It lacks Metafont
capabilities though, if that makes you happy.)