Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!rutgers!sri-unix!hplabs!decwrl!pyramid!prls!philabs!tg!scott
From: scott@tg.UUCP
Newsgroups: comp.text
Subject: Re: Why troff?
Message-ID: <106@tg.UUCP>
Date: Sat, 24-Jan-87 01:24:08 EST
Article-I.D.: tg.106
Posted: Sat Jan 24 01:24:08 1987
Date-Received: Sun, 25-Jan-87 14:39:01 EST
References: <362@linus.UUCP>
Reply-To: scott@tg.UUCP (Scott Barman)
Distribution: comp.text
Organization: Townsend Greenspan & Co., Inc.
Lines: 60
Keywords: TeX, troff

Just some insights on (di)troff:
I work for the American Physical Society (this is a graciously borrowed
id, for which I am thankful) and we publish the 45,000+ journal pages
typesetting them with troff/ditroff and supporting preprocessors tbl and
eqn.  Now let me say these versions are severely hacked to cure some
"ills" with these programs.  However, with the problems found (including
the 4-font restriction of the original C/A/T troff), troff seems to be
the best way to publish the high volume of journals as quickly as we do
with few problems.

Ditroff provides flexibility in the area of output support where we have
to produce output for a VideoComp 500 phototypesetter and an Imagen laser
printer (for proof reading).  Ditroff allows us, with some minor hacking
in table sizes, to support up to 20 named fonts and others that we
"mount" as we need them.  We also have the availibilty of using the many
very well written preprocessors like tbl, which is very important in
displaying tables of information.

Another advantage to ditroff is the ASCII output file it produces.  This
output allows for the output destined for one device to be previewed on
a Tektronix 4014 display as well as allowing some editing of this file
to convince a driver for the Imagen that this is output destined for it
(this is necessary to check page/line breaks).  This ASCII output is
also a very good, quick aid to try to determine what ditroff is doing
without printing each piece of output (which can get expensive).

I'll admit the biggest disadvantage of ditroff vs. TeX is TeX's ability
to typeset mathematics.  While I do not think that eqn was designed for
the type of work we force on it, with some hacking (the only program
that has undergone a near rewrite), we have it produceing three-page
physics proofs on a regular basis.  Without the extensive hacks to eqn,
we would not be able to do this job as well as we do.

In conjunction with eqn and ditroff problems, there are many problems
with accents/diacritics.  Since I do not know how TeX handles these, I
can only say that we have new routines and other additions to eqn
grammar to add things like bars, hats, dots, accute and grave accents,
cedillas, and (before ditroff) creating an angstrom from an A and a
degree-symbol.  I think that the American Mathematical Society uses TeX
to produce its journals, maybe someone from that group could give us
insight on TeX and publishing.

THE biggest disadvantage of TeX vs. ditroff I can see is the requirement
of TeX to use its own generated fonts (metafont).  We have looked into
the possibility of using TeX, but have resisted up until now because the
Computer Modern fonts of TeX does not compare with the Times Roman
availble on the VideoComp.  We cannot download fonts to this typesetter
nor is it fiscally feesable to have Information International Inc. (the
makers of the VideoComp) digitize these fonts for us.  For APS, this is
a big factor!

This is just one perspective, not necessarily correct!  I would be
interested in hearing about the uses of TeX on very high volume of
output also requiring the quality necessary for publishing.  I am
leaving APS for "greener pastures" but would be willing to pass on any
information that might help in the TeX decisions (I am being retained as
a consultant--neat arrangement ;-)).

Scott Barman
philabs!tg!scott