Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site l5.uucp
Path: utzoo!linus!decvax!decwrl!sun!l5!gnu
From: gnu@l5.uucp (John Gilmore)
Newsgroups: net.info-terms
Subject: Re: terminal emulators and invalid ESC sequences
Message-ID: <372@l5.uucp>
Date: Mon, 30-Dec-85 06:31:58 EST
Article-I.D.: l5.372
Posted: Mon Dec 30 06:31:58 1985
Date-Received: Tue, 31-Dec-85 00:59:42 EST
References: <171@uscvax.UUCP>
Organization: Nebula Consultants in San Francisco
Lines: 24

When a terminal emulator (or a terminal emulator running dedicated
in a small micro commonly called a "terminal") receives an invalid
escape code it can do roughly what it pleases.

If it follows the X3.64 standard (ANSI or VT100 compatible) then it
is required to ignore the whole sequence.  The encoding of the
sequences is designed so that a simple state machine can parse them
even before the terminal knows what they mean -- so if it ends up
not knowing what they mean, it can just toss them without trouble.

If you are emulating a specific terminal, about the only thing to do is
buy or borrow one of those terminals and try it.  I screwed up the
first release of Sun's firmware 4014 emulator that way -- when I got an
invalid control char, I dropped out of graphics mode back to alpha
mode.  Seems like various programs were used to being able to send ^V's
randomly scattered throughout their output [never did figure out why],
and this broke my emulator.  We didn't have a 4014 to test against (and
of course the manual didn't say...of course!) so it took a customer to
find the bug.

I also consulted to Anderson-Jacobson on their VT100 clone years ago.
There sure are a lot of things that you can ask a terminal to do, that
aren't in the manual so you can see what it *ought* to do...or, in
other words, "Specs?  The *code* is the spec..."