[Groff] Re: typographical history

[Steve Izma]

> From: Peter Schaffter <address@hidden>
> Date: Sat, 22 Jan 2005 12:28:24 -0500
> To: address@hidden
> Subject: Re: [Groff] Werner's Margin Notes
> 
> On Sat, Jan 22, 2005, Jorgen Grahn wrote:
> > Never heard of them before, but wouldn't they be likely to be based on
> > runoff or troff, and thus siblings of groff? The Quadritek (see e.g.
> > http://macpro.freeshell.org/quadritek/q1.html, but look out for the Mac
> > advocacy) is from 1977.
> > 
> > I would love to hear more about the ancient phototypesetters and how they
> > related to the world of troff.
> 
> I just checked out the Quadritek article.  Wow!  A blast from the
> past.  The article's good.  In fact, the whole thing captures the
> flavour of the machine and the times perfectly.
> ...
> Another thing missing from the article is the fact that type was
> literally set one line at a time.  You entered your codes and text
> in a small section at the bottom of the screen, then hit Return.
> What you'd just typed then jumped to the upper portion of the screen
> (so you could keep track of what you were doing) and the line was
> immediately typeset.  If you were doing a big chunk of justified
> copy, the machine automatically added the return (and hence typeset
> the line) at the appropriate place, which could be very annoying
> if you made a mistake because there was no way to recover from
> it.  (Later system upgrades made it possible to code an entire job
> without it actually being typeset, giving you the then-unheard-of
> luxury to correct your mistakes before actually running the file!)

I'm just catching up to this thread. Am I correct in suggesting
that the main issue here is at what point relative to the
end-of-line event does the position advance to the next baseline?

In hot type (which I never used) I assume the baseline was
built into the type, although it was modified with the addition
of leading between the slugs of type for each line. When I
started typesetting in the early 70's on an early Compugraphic
phototypesetter, I'm pretty sure that when the machine determined
when to end the line, it set the type for the current line
then advanced ("film advance") to the next baseline using
the current value for leading (which for cold type meant the
baseline-to-baseline distance, not just the space between the
slugs). Hitting the return key (or quad left, or whatever) to
finish a line manually had the same effect. So if you changed the
leading in the middle of the line, it didn't take effect until
the next line.

Troff has the opposite behaviour. The baseline is effectively not
established until a line is complete; then the current output
point advances according to the current leading (which may have
been reset by a .vs command in the middle of the line).

Peter was earlier referring to how significant this difference
becomes at the start of the page. This used to always screw me up
when trying to position a running head. On one hand, this appears
to be merely a philosophical difference, but I'll bet that it
derives from the difference between traditional cold-type typesetting
systems (which were not page-layout systems but produced only
galleys that were later cut and pasted into pages) and roff-style
systems that were from the beginning designed to move not just
down a column but to a specific point on the next page.

> The Quadritek was never a major player, though. The hands down
> winners in phototypesetting were Compugraphic and Linotronic.

By 1975 the typesetting co-op I was part off had acquired a used
Merganthaler-Linotype VIP, probably the most popular of the
higher quality phototypesetters in the world at the time and also
the last generation of the technology that began in the
mid-sixties and was dying by 1979 and thoroughly dead by about
1985. At about that latter date, we acquired about three of them
from a bankrupt Toronto typesetting company for either nothing
or close to it. Five years before they would have cost about
US$50,000.

About two years before this, some typesetting friends of mine were
starting up SoftQuad and had just bought the rights from AT&T to
use the source code to troff (acutally ditroff, I believe). These
people had been using VIPs for at least ten years by then and had
been driving them by computer (probably some sort of DEC
minicomputer). Along the way, another friend who shall remain
unnamed but was heavily involved in pushing the capabilities of
VIPs to the limit, had acquired the source code for the VIP's
internal computer, which included all the typesetting
algorithms. Let's just say his acquisition involved the
serendipitous recitation of some lines from either Monty Python
or the Firesign Theatre at a trade show in the presence of a
like-minded Merganthaler technician. Sort of like a secret
handshake. This made it a lot easier to modify the VIP program
(you booted it from an eight-level punched paper tape) so that it
could interact with external computers much more effectively.

One of the SoftQuad people had been involved in this a few years
before they got the licence for the troff code. I clearly recall
him telling me how he was nagged by a certain deja-vu while
studying the troff algorithms -- he soon recognized many
similarities to the VIP code.

The VIP code was probably developed in the mid-sixties. The troff
algorithms were probably worked out in the early seventies to
drive a CAT phototypesetter, I believe, which was not related to
the VIP in the corporate ownership sense. Did Joe Osanna have
contact with the Merganthaler people? Or were both programs
derived from some earlier common source? If we were biblical
scholars, we could call it Q.

Of course, groff was written from the ground up, but I know for a
fact that James Clark was in frequent contact with the SoftQuad
folks while he was writing groff, and the ways in which they both
differ from original troff bear a lot of resemblance.

I guess there's really nothing much new under the sun.

-- 
Steve Izma
    Computing Systems Administrator       (519) 884-0710 ext. 6125
    Wilfrid Laurier University Press      FAX: (519) 725-1399
    Waterloo, Ont., Canada N2L 3C5        address@hidden