Steve Litt wrote:
> I was a DEC PDP/11 TSX over RT-11 guy back then, but as I remember, a
> terminal was a television that printed letters and numbers plus a
> keyboard on which you could type.
I have to disagree a little bit in that actual TVs were too low-rez for
good 80-column text, which has a longer tradition. However, TV-compatible
40-column or lower text modes did come into play during the microcomputer
revolution, at the Homebrew Computer Club, and with the TV Typewriter
(which was more of an electronics hobbyist project, but could be turned
into a terminal, though not a "professional" one). The DOS PC CGA also
supported 40-column modes--it had a composite/TV output besides RGBI. Of
course, if by television you basically meant CRT, then I'd agree.
I'll add some more, actually.
(This isn't first-hand knowledge, but I forget where I got it from --
possibly various sources.)
Here's some
-=PREHISTORY=-
which explains more of how terminals came to be:
The kind of early electronic data processing that lead to terminals really
started with (electromechanical) tabulating machines. Those gave us the
standardized punch card. Edwin Black's "IBM and the Holocaust" contains
(not very detailed and technical) descriptions, and even photos of such a
machine and some punch cards. An early use for these evolving machines was
the census, in the US and later in (Nazi-) Germany.
The way this worked was, these cardboard punch cards got holes stamped into
them to encode information on them. This was done in dedicated card punches
(keypunches: https://en.wikipedia.org/wiki/Keypunch), which generally were
a cross between a hole punch and a typewriter. There was an array of rows
and columns and a clerical worker would type and thereby punch the right
holes in the right positions. IBM soon standardized on the 80-column card,
and that's why most terminals, screen fidelity permitting, got 80-column
text. One punch card could encode one such line of text. (ASCII text? Well,
sort of, similar. Let's forget that EBCDIC ever existed.)
https://en.wikipedia.org/wiki/Punched_card#IBM_80-column_punched_card_formats_and_character_codes
I don't know if there's a specific reason for the 24 (later sometimes 25)
lines of most terminals, but I guess it had to do with being a multiple of
8 and just how many 80-column lines could legibly be fit onto a 4:3 aspect
ratio CRT display. But that came later. Back to the cards.
Those punch cards were fed to a reader, which was an array of contacts that
would be lowered onto and --holes permitting-- through a card. Early
tabulators had little pockets of liquid mercury underneath each position,
and the electrical circuit was closed by the contacts dipping into that.
(Hello OSHA.) Those tabulating machines weren't computers, and some early
ones just had clock dial counters, on which hands would advance when a
contact was closed. This already allowed electromechanical addition though:
Feed in a card, read it, have the hands advance. Feed in a second card,
read it, have the hands advance some more.
https://en.wikipedia.org/wiki/File:HollerithMachine.CHM.jpg
Other tabulating machine setups included sorters, where cards would be
sorted in different output stacks depending on their hole-encoded
information. (There's a link from that to VisiCalc and Excel, but that's OT
to this OT post.)
When computers were invented, these 80-column cards quickly became the
industry-standard input/output mechanism.
https://en.wikipedia.org/wiki/Punched_card_input/output
The programmer/user would write a program or calculation on paper, mail or
hand that paper to the clerical staff who would type/punch that information
into cards, and then a whole pile of punched cards would be delivered to
the computer operators, who would carefully deposit them into the punch
card reader, and these machines had feeders which could quickly process a
whole batch of cards, one after another. This is what gave us batch
processing.
For output, these computers could either themselves punch cards and/or they
could use a line printer. https://en.wikipedia.org/wiki/Line_printer
(Sometimes the output was the way the cards got sorted, but that was mostly
a tabulating machine thing.)
It was pretty standard then to hand in your program sheets and get back
your printed results back on continuous stationery.
https://en.wikipedia.org/wiki/Continuous_stationery
This was technically multi-user, but you had to queue and wait maybe hours
or a day for your cards to be processed and to get your results back.
Longer if it went through the mail. (DO NOT BEND was very important if
punch cards were mailed, because punch card readers really don't like
warped cards.) https://en.wikipedia.org/wiki/Time-sharing#Batch_processing
Meanwhile, the telegraphy industry had developed the teletype, which was a
teleprinter, basically a typewriter that could electronically transmit
typed characters, across a room or across the country, and type out (on
continuous stationery) characters that it received.
https://en.wikipedia.org/wiki/Teleprinter
https://en.wikipedia.org/wiki/Teletype
These technologies were combined, and soon you had teletypes connected to
computers, so a user could themselves remotely type in their lines and
--irrespective of whether the actual computer used punch cards or magnetic
tape internally-- get the results back by having them printed out on their
teletype. It became possible to connect multiple teletypes to one computer,
thanks to the development of time-sharing. These teletypes that were hooked
up to computers were --literally-- terminal devices, at the end of a
telegraph wire / serial cable.
At some stage someone had the bright idea to add an electronic display (a
CRT) to the teletype, and now the thing wasn't such a loud impact printer
anymore and didn't need to waste so much paper. But the concept of having
that printed record of the interaction with the computer still somewhat
survives in our log files and script(1). The first CRT-based terminal
devices also became known as "glass terminals", and they came to replace
many teletypes.
(Of course, if you really didn't have a teletype anymore, you now needed a
printer...)
Once you added modems to marry up the telephone network with your terminals
and computers, you could make the one expensive computer at your university
or corporation accessible to many users, on- and off-site. Developments in
time-sharing and multitasking technology gave birth to Unix.
In part, these developments preceded the microcomputer revolution, however,
once that hit, many users found that they could complete their computing
tasks entirely on their desktop computer, and single-tasking, single user
DOS PCs became a big hit for a while.
https://en.wikipedia.org/wiki/Microcomputer_revolution
However, the I/O and networking technology established with terminals
remained useful and led to the development of terminal emulators and BBSes
and FidoNet (all usable even with DOS).
https://en.wikipedia.org/wiki/Bulletin_board_system
https://en.wikipedia.org/wiki/FidoNet
With improving Internet-connectivity, systems arose that were sort of
nephews of terminal emulators, but they now established the text mode
interface via existing network stacks rather than though dedicated
serial/modem connections. Telnet was one of those.
The text terminal-compatible Gopher system briefly rose and quickly fell,
and we know why.
https://en.wikipedia.org/wiki/Gopher_%28protocol%29#Stagnation
Then right around 1994, '95, and '96, four things happened: BBSes started
to decline, OpenBSD struck out on its own, FidoNet peaked, and the WWW and
Internet started to grow exponentially.
OpenBSD proceeded to lead the way in replacing the famously insecure telnet
with SSH. (It's open to argument how much OpenSSH even has to do with
terminals. I'll probably get stick for that association, because pty(4),
not tty(4), etc. Technically an entirely separate component you could
argue, though it functionally enables--but anyway.)
As you know, serial consoles and terminal emulators still have their uses,
though their use is declining.
But there remain many design echo(1)es of terminals and teletypes; from
ex(1) to banner(1) to wump(6)...
And if anyone has ever operated the OpenBSD installer via a teleprinter, I
want to hear that story.
PS: https://www.youtube.com/watch?v=bHLt2NrSEuU&t=7m22s
