Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!samsung!emory!att!cbnewsl!npn
From: npn@cbnewsl.att.com (nils-peter.nelson)
Newsgroups: comp.text
Subject: Re: Price of DWB 3.1
Summary: A restrained reply to Jay Plett
Message-ID: <1990Nov28.193202.19684@cbnewsl.att.com>
Date: 28 Nov 90 19:32:02 GMT
References: <1990Nov5.022533.29625@nixtdc.uucp> <16706@letni.UUCP> <652@silence.princeton.nj.us>
Organization: AT&T Bell Laboratories
Lines: 50



I hope this is of interest to the general community. I have
no intention of boring the world with interpersonal chit-chat.
Jay Plett writes an excellent essay, and I find relatively
little fault with what he says.
First, the nits.
Plett upholds the history of troff. So do I. My criticism
was of the externally packaged DWB 2.0 release compared
with what was being used internally at the time. The internal
package was considerably more robust and modern. The goal
in bringing out 3.1 was to give the outside world access to
the same technology afforded to Bell Labs people.
Plett and Dunn call DWB a 'little' piece. There are 112 files
in the version on this system, and that doesn't count the font
tables and files. 35,000 blocks of source code hardly qualifies
at little. There are 23 new postscript programs alone.
It was certainly my own personal preference to package DWB
with UNIX. I lost. As a result, the current UNIX shipped by
USL doesn't even include the 'man' command (because man
relies on nroff).
Having lost that battle, what is a reasonable price for
60,000 or so lines of well-tested, portable source code?
As mentioned earlier, some VARs paid $4,000 for the 2.0
source license, compiled it, and sold binaries for $8,000
apiece (Amdahl price)!
The grand strategy, so to speak, is that large companies
would do well to buy a source license for their own internal
use; it's cheaper and gives them the obvious benefit of
source code.  Most important, they get *identical* software
on Amdahl, AT&T, DEC, IBM, Sun, hardware.
Smaller companies will have to rely on VARs selling binary.
AT&T should be coming out with a complete set (386, 486,
3B2, System 7000) early next year.  I've been unable to
get the attention of anyone at Sun and the others, but
perhaps this will help.
And, last, the university license of $1,000 for source is,
to my mind, generous. A guy who works for me has *two* kids
at Princeton, and that's $40,000 in tuition room and board.

My own personal vision is for DWB to be something you can
count on wherever you compute. It's not the sexiest formatter
around, but it is the most dependable. I can ship troff files
to any of several thousand people inside AT&T via e-mail with
confidence that they will be able to run off exactly what I
intended on their local printer. Unlike PostScript, these
files are revisable and so promote multiple-authored, inter-
location documents.  That's our target: distributed,
multi-machine, multi-vendor environment with "industrial
strength" needs (dozens to thousands of pages).