[TUHS] Curly braces: An evolution of UNIX and C

[Clem Cole via TUHS]

below

On Wed, May 20, 2026 at 7:07 PM G. Branden Robinson <
g.branden.robinson at gmail.com> wrote:

> Hi Clem,
>
> ...
> Thanks for raising this.

You're welcome.

> ...
>
> I had thought that the CSRC's hands-off approach to video terminals was
> that (a) it "wasn't research", and (b) more a matter of the DEC VT100
> showing up and commencing to eat the world, and Berkeley manifesting a
> resolve to attack the problem of character-cell video terminal support
> with its termcap library.
>
Well, it probably was a number of things.  As I said for a long time,
inside the Bell System, well into the 1970s, many/most people used a WE212
and a phone line to access the system down the hall. That was just the way
the phone company did things (think about hammers and everything being a
nail).   Also, Teletype is an AT&T company.   So getting one of their
products was different from buying something from an outside vendor. This
was different from the way Universities and others.  By the late 70's, most
places I experienced had started to switch.  I'm sure there were pockets,
but as I said, by July 1976, an ADM3A "glass tty" was the least expensive
way to attach to any timesharing system, not just UNIX-based ones.  Even in
the Bell system, things did start to change by the mid-last 70s - you
started seeing TI "Silent 700s" which were introduced in 1971 for $2780, in
offices within the Bell System.  By the late 70s, the HP 2621A and 2621P
[which cost $1450 and $2550 in 1978] were definitely popular at the Labs,
as they were slightly cheaper than DEC's August 1978 VT-100 release at
$1900.

The point is, as late as the development of Research Seventh Edition, the
use of a "glass tty" was less common within the Bell System, while the
Universities had already started to switch, and at many places (certainly
my experience at CMU and USB) had already switched nearly 100%.   But that
was the problem.  Most "video" editors like Cornell's Fred (friendly ed) or
the Rand Editor were hard-coded for specific terminals.  Joy's termcap was
a huge (and practical) solution to an issue that was felt outside the Bell
System before it was really needed inside it.



>
> Your comment extends my timeline backward, which isn't inconsistent with
> my story but leaves me wondering what the thinking was in earlier days,
> prior to 1978/9.  AT&T did in fact hire the termcap developer away from
> Berkeley to craft terminfo--but that was in the 1980s.
>
I think Mary Ann is on this list, so I let her comment directly.   Certainly,
by the late 1970s and early 1980s, the older scheme was beginning to change
within the Bell System, but in my experience, it was lagging behind the
outside.

>
> What's the mid-1970s rationale for AT&T ignoring this segment of the
> market, when it cannot have been a secret that people were using these
> devices to interface with Unix?

There are a couple of things.  First, please remember that AT&T is not
allowed to be in the computer business.  They had "abandoned on your
doorstep" the sources to UNIX and left it to you to figure out what you
wanted to do with it.  So the only pressure they would have felt was from
internal customers within the Bell System.  AT&T "products" were from
Teletype Corp.

The interesting thing is that Teletype did make a "rugged" terminal
targeted to the Bell System, insurance firms, and the US Military, called
the Model 40 or Dataspeed 40, and sold in 1973 between $3K-4.5K.  This was
a modular, heavy-duty industrial data terminal. It featured a large,
character-driven cathode ray tube (CRT) display, a detached keyboard, and
an optional high-speed friction or tractor-feed line printer, and of course,
could run up to 120 cps, just like what a WE212 modem could support.

In the early 1980s, after ANSI X3.64 was released, Teletype tried again
with the 4400 and 5000 families.  The 1983 release of the 5410 and 5418
were traditional text-only character mode terminals that conformed to the
ANSI X3.64 standard and cost $995 and ~$1300. They had no bit-mapped
graphics capability, but they could switch between 80 and 132 columns of
text and featured horizontal split-screen functions.   The 1984 release of
the 5420 and 5425 were intended for line-oriented data entry, allowing an
operator to type out a full screen of text locally into memory before
sending the raw text block to the host computer all at once — IBM mainframe
style and cost $1495 and $2215.

Of course, by 1983, the Wyse-50 was selling for $695 in ones and twos and
substantially less in quantity ($350 was typical).  By 1986, Wyse had their
100% ANSI-compliant Wyse-60 [VT-100s are only partially ANSI-compliant BTW,
Wyse-60s could "dumb down" to be VT-100 too], which was listed at $599 with
a street price in the $400-$450 range.

To be fair, Pike and Locanthi created the JERQ in 1981.  By 1984, it had
been reengineered by Teletype to become the DMD 5620 [Dot Matrix Display]
with a list price of $6115, and by 1986, you could get one from a VAR for
around $4K.

But fate would have it for Teletype again, as in Sept 1985, Sun released
the Sun 3/50 for $8K sans disk and $14.9K as a standalone system.

So, back to your question: by the time AT&T realized that "glass ttys"
would be important, they were still inwardly focused as they developed
products for their traditional market.  Once they were allowed to be in the
computer business officially (*i.e*., post Jan 1, 1984) the products they
had were not competitive.


> I think I get why Research didn't dirty
> their hands with it.  But it should have been solidly within USG/
> "Program Generic"'s wheelhouse.  And eventually it was, but took
> something like 7 years to get there.
>
It really was "TPC" in action. The development and bringing-to-market
thinking inside AT&T was based on a history of having a captive market that
told its customers what was needed.  When the chains of the consent decree
were cut, the problem was that they did not have a structure to recognize,
much less understand and value, the same things their new customers
wanted.

The best example of that was the AT&T System V "UNIX Consider it Standard"
campaign.  The actual users outside of the Bell system scratched their
heads, and so no.  That was not something that AT&T's internal mechanism
could handle.