From: utzoo!decvax!harpo!npois!ucbvax!C70:editor-people
Newsgroups: fa.editor-p
Title: Re: ^S/^Q but not DTR
Article-I.D.: ucb.1385
Posted: Sat Jun 19 02:41:28 1982
Received: Sun Jun 20 01:03:29 1982
Reply-To: ople

>From Frankston.SoftArts@MIT-MULTICS Sat Jun 19 02:39:12 1982
     editor-people at SU-SCORE, Goldberg at RUTGERS
Remailed-date: 19 Jun 1982 0001-PDT
Remailed-from: J.Q. Johnson 
Remailed-to: Editor People: ;

{More discussing Xon/Xoff issues, skip this if you are not interested}

There is a big difference between something that just happens to work
in one case for random reasons and some that works.  A propeller
airplane is a lousy means of getting to the moon.

DTR might just happen to work between two pieces of hardware.  On the
other hand it is more likely to hangup your phone.  At least ^S/^Q
can be passed over a communication line.  I would guess that Xon/Xoff
is considered much harder to implement because you are speaking to
hardware people who seem to be unable to deal with things that are
not implemented solely in chips.  All it takes is an interrupt
handler that is able to operate within the skid time of your
terminal.  For a VT100 that is about 16 character times which as
19200 is 1/19200*10*16 seconds which 8.32 milliseconds.  To be
conservative, let's assume 4 milliseconds.  That is a very long
period of time.  A real VT100 is closer to operating at 4800 baud
anyway so that cranking it up to 19200 doesn't really help you.
Probably 9600 is better for it.

This is not to say that Xon/Xoff is a good protocol.  It is a rather
poor one, but it is available on many terminals.  If we were redoing
the world, then there would be a transport level that deals with such
matters and does it out of band.  Such mechanisms to exist, but are
not in the domain of this discussion.