[ih] Re: internet-history digest, Vol 1 #130 - 1 msg

Bob Braden braden at ISI.EDU
Sat May 17 08:34:21 PDT 2003

>RFC 97 states that one of the design problems initially for Telnet was
>basically line-at-time systems should universally interface with
>character-at-a-time systems and vice versa.  Obviously dealing with this
>problem was essential in Telnet's design, and I want to ensure that I
>fully understand the problem and its solutions.

A LAAT client was unable to use a CAAT server, but vice versa worked
fine. On input, a LAAT client could not send an isolated character, for
example.  However, there was really a more complex distinction -- LAAT
systems were nearly always half-duplex.  Once the user entered a line,
the keyboard would be LOCKED until the server sent a complete line
and explicitly unlocked the keyboard.

In the beginning this distinction was not made, because the LAAT people
and the CAAT people each lived in their own world.  The CAAT people
could simply not conceive that anyone might lock the keyboard -- it was
too contrary to their world view.  Eventually, we LAAT freeks managed to
get this point across.  I believe that Alex McKenzie was one of the first
CAATers to understand the problem.

It was this experience that led me to a great generalization about computer
cultures: there were paper-tape machines, and there were punched card
machines, and never the twain could meet.

>RFC 318 on page 5 gives two "USER TELNET SIGNALS": Transmit Now and
>Suppress End-of-Line.  I am assuming that these were the mechanisms in Old
>Telnet to allow the two kinds of systems to interface along with line mode
>and character mode, as described on page 14.  From reading the RFCs, I do
>not have a clear understanding of these signals' usage.  How did they
>work? Was this solution in Old Telnet abandoned because of the asymmetry
>problem that the protocol had?

I don't immediately recall, but it seems apparent today that these commands
could not be implemented on a half-duplex LAAT system, so they would have 
been of little
help.  What helped was making the DEFAULT be LAAT half-duplex in the New 
with the ability to negotiate up from there.

>For New Telnet, the Go Ahead command and the option Suppress Go Ahead were
>the new mechanisms to make this interface work between the different kinds
>of systems. This seemed to make sense because RFC 857 (page 3) and earlier
>documents like NIC 15930 (August 1973, page 4) state:
>    The echoing option alone will normally not be sufficient to effect
>    what is commonly understood to be remote computer echoing of
>    characters typed on a terminal keyboard--the SUPPRESS-GO AHEAD option
>    will normally have to be invoked in conjunction with the ECHO option
>    to effect character-at-a-time remote echoing.
>This statement helped me to understand this better because in the
>descriptions of GA use half-duplex and full-duplex terminals, which were
>not used in Old Telnet. Do these kinds of terminals have anything to do
>with character-at-a-time and line-at-a-time systems? What's the connection

See above.  By New Telnet Time, the CAATers had figured out LAAT.
Though they still privately thought LAAAT was brain-damaged, ARPA
had made it clear that the ARPAnet had to support IBM and other
mainframes that used LAAT.

Bob Braden  (who LIVED it!)

>As always, thanks for the continuing help with this research.
>Adriana Arrington
>mailto:aca at cs.utexas.edu
>internet-history mailing list
>internet-history at postel.org
>End of internet-history Digest

More information about the Internet-history mailing list