Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!nrl-cmf!ames!coherent!dplatt
From: dplatt@coherent.com (Dave Platt)
Newsgroups: comp.misc
Subject: Re: Looking for Computer Folklore
Message-ID: <20114@coherent.com>
Date: 6 Feb 89 18:04:20 GMT
References: <7143@pyr.gatech.EDU> <532@geovision.UUCP> <4575@tekgvs.GVS.TEK.COM> <319@itcatl.UUCP> <2986@ficc.uu.net> <6230@saturn.ucsc.edu>
Reply-To: dplatt@coherent.com (Dave Platt)
Organization: Coherent Thought Inc., Palo Alto CA
Lines: 54

Here's another one involving circuit breakers.  It's quite true... it
bit me several times.

Honeywell used to make a 30 character/second dot-matrix terminal called
the "ROSY".  This monster weighed about 50 pounds, was horribly noisy,
and was one of the least-favorite terminals with which to be saddled.
There were a couple of significant "misfeatures":

1) The designers of the terminal were apparently raised in the
   half-duplex world.  If you hit the terminal's "break" key, the serial
   chip would begin to send a "long space" (which is correct) and would
   also turn off the receiver logic in order to avoid garbling on
   half-duplex circuits (which isn't correct in the full-duplex world).
   The receiver was turned off in a very crude fashion... I believe that
   its input was clamped for as long as the transmitter was sending a
   long-space.  When the long-space ended, the receiver was reactivated
   and would interpret the next "space" bit it saw as a "start of
   character" signal, even if that bit was in the middle of a real ASCII
   character (as sent by the computer at the other end of the wire).
   Net result... if you hit "break" when speaking over a full-duplex
   connection, the terminal would miss several characters and would then
   print several characters of garbage.

   This was particularly troublesome when this terminal was used with a
   system that used "break" to request a soft interrupt (sort of like
   control-Z on BSD Unix).  The loss and garbling of characters would
   frequently obscure the "Break! C to continue" prompt from the current
   program, leaving the user waiting for a prompt that never appeared.

2) Early models of the ROSY had a nice, undocumented feature.  When they
   received a US (unit separator) control character, they would activate
   an SCR crowbar circuit across their power supply output, and would
   trip their circuit breaker... a nice "remote shutoff feature".

These two design misfeatures, in combination, added up to real trouble.
A programmer would be sitting at the terminal, editing a file, and would
ask the editor to type out a range of lines.  Partway through the
listing, the programmer would see the lines that s/he had wanted to
view, and would hit "break" to stop the listing.  The terminal would
miss several characters while sending the "break", would turn its
receiver back on at the wrong moment in the middle of the "Break! Hit C
to continue" message from the editor, "believe" that it had received a
unit-separator character, and would crowbar its power supply.  SNAP!
The terminal powers itself off, disconnecting the timesharing session
and discarding any changes that the user had entered and had not yet saved.

Needless to say, this problem led to the common believe that the ROSY
was unsuited for any use other than as a boat anchor.

-- 
Dave Platt    FIDONET:  Dave Platt on 1:204/444        VOICE: (415) 493-8805
  UUCP: ...!{ames,sun,uunet}!coherent!dplatt     DOMAIN: dplatt@coherent.com
  INTERNET:   coherent!dplatt@ames.arpa,    ...@sun.com,    ...@uunet.uu.net 
  USNAIL: Coherent Thought Inc.  3350 West Bayshore #205  Palo Alto CA 94303