From:	SMTP%"Info-VAX-Request@Mvb.Saic.Com" 10-SEP-1994 18:29:58.10
To:	EVERHART
CC:	
Subj:	Re: clear screen

From: Jerry Leichter <leichter@lrw.com>
X-Newsgroups: comp.os.vms
Subject: Re: clear screen
Message-Id: <9409092236.AA06157@uu3.psi.com>
Date: Fri,  9 Sep 94 18:21:46 EDT
Organization: Info-Vax<==>Comp.Os.Vms Gateway
X-Gateway-Source-Info: Mailing List
Lines: 130
To: Info-VAX@Mvb.Saic.Com

	=I create a symbol
	=
	=	$ CLS	:== WRITE SYS$OUTPUT """<ESC>[2J<ESC>[1;1H"""
	=	$ HOME	:== WRITE SYS$OUTPUT """<ESC>[1;1H"""
	=	$ RESET	:== WRITE SYS$OUTPUT """<ESC>c<ESC>[2J<ESC>[1;1H"""
	=
	=The <ESC> is produced by pressing the escape key twice in EDT.  I
	=know that this is limited to VT terminals.   I hope this helps.

	Once more:  THE FIRST TWO ARE *NOT* LIMITED TO VT TERMINALS.  THEY ARE
	LIMITED TO *_ANSI_* TERMINALS.  They should, for example, work with a
	PC set up to behave like a generic ANSI terminal.  Now, the RESET I
	don't know about.

We're getting into some deep and muddy water here.  Everyone out on the list:
Have you ever *read* ANSI X3.64?  Hmm, don't see too many hands.  Have you
ever *seen a copy*?  Not too many more.  Is your hand up, Mr. Lydick?

ANSI X3.64 is a big, complicated spec - 800 pages or more, as I recall.  It's
also the ultimate vendor's spec, since it is quite possible - extremely easy,
in fact - to build a terminal that can be legally called "ANSI X3.64 compli-
ant" even though it does nothing at all useful.  Alternatively, it's possible
to build two legally "ANSI X3.64 compliant" terminals that are not in any
practical sense of the word interoperable or interchangeable.

ANSI X3.64 starts off with purely syntactic definitions.  (Actually, it may
inherit some of this stuff from yet another ANSI spec, X3.41.)  It defines the
syntaxes of control sequences, escape sequences (yes, technically there is a
*significant* distinction), and several related concepts.  The syntactic
definitions make it possible to recognize when a legal X3.64 "command" begins
in the datastream and when it ends, even though you have no idea what it
means.  You can tell purely by the syntax:  The choice and arrangement of the
constituent characters.  Thus, <ESC>c is a complete escape sequence.  So is
<ESC>(B.  So, for that matter, is <ESC>((((((((c.  <ESC>[2J, on the other
hand, is a complete control sequence.  So is <ESC>[((;;;?????112==??;J.
Even more simply, the single character HT (CTRL/I - Horizontal Tab) is itself
a command.

The rest of ANSI X3.64 defines the meaning of some control and escape
sequences.  As I recall - and I never had reason to look at the spec too
closely; it runs to hundreds of pages - ANSI will define:  (a) a specific
standard name for a function; (b) assign a specific command sequence to that
name; (c) define the semantics of that function.  Thus, the ANSI function name
RI, escape sequence ESC M, is Reverse Index:  The cursor moves up on line in
the current column; if it was in the top column, the screen scrolls down.  In
addition, I believe ANSI X3.64 assigns official names to certain subsets of
functions.

ANSI X3.64 defines many different functions.  It divides the undefined but
syntactically correct sequences into two classes:  Those reserved for future
ANSI standardization, and those available to manufacturers as "private
extensions".  (There is a technical point here:  Control sequences can have
parameters.  It is possible to define private parameters for standard control
sequences.  Thus, ESC [ xxx h is a standard "Set Mode" control sequence, with
the xxx specifying what to set.  ESC [ 1 2 h sets the standard-defined Send-
Receive Mode (which has the effect of turning local echo off).  ESC [ ? 8 h
sets the DEC-private Auto Repeat Mode.  The "?" marks a private parameter.)

As I recall the definition of ANSI X3.64 compliance, a complying device:

	a)  Correctly parse any valid ANSI command sequence.  (If it doesn't
		understand the sequence, it may discard it as a whole or
		report an error - but it can't act on parts of it as though
		they were "just characters".  All DEC terminals ignore any
		syntactically valid sequences they don't understand.)

	b)  Treat any characters that are not part of a command sequence as
		individual displayable units (that's the meaning ANSI assigns
		to them).

	c)  If it chooses to implement any of the ANSI-standardized commands,
		it must implement them in exactly the way specified in the
		standard;

	d)  It may not provide a private meaning to any commands in the class
		of commands reserved for future standardization;

	e)  (This one I'm not certain about):  If it claims to implement one
		of the officially-defined subsets, it must implement all the
		commands that make up that subset.

I do not believe the ANSI spec actually *requires* that any particular
commands be implemented.  In practice, as far as I know, *no one* has ever
implemented a "full" ANSI terminal - i.e., one that implements *all* the
commands the spec defines.  Some of the defined commands have probably never
been implemented anywhere.  On the other side, as far as I know, no one has
every sold a terminal that does not implement at least some private functions.

In practice, very few "pure ANSI" terminals have ever been made, and almost no
software that can deal with a "pure ANSI" terminal has ever been written.  The
private extensions defined in the VT100 have come to be the de facto defini-
tion of "ANSI compatible", whatever the letter of the spec may say.

By the way, if we ignore the problem that an ANSI terminal need not actually
implement the functions involved, there is an easy way to tell whether a given
command accepted by a DEC terminal (or printer) is standardized by ANSI or a
DEC-private extension:  Just check the official name for the command in the
DEC documentation.  All DEC extensions have a name that starts with "DEC".
(Yes, theoretically an ANSI command name might start with those 3 letters, but
I know of no examples.)  Thus:

	ESC [ 2 J	ED (Erase in Display)  Standard command; standard
			parameter 2 requests that the entire display be
			erased.

	ESC [ 1 ; 1 H	CUP (Cursor Position).  Standard command; standard
			parameters specify the cursor position to move to,
			in this case, the home position.  Note:  ESC [ H
			is, by definition, equivalent.  Also note that the
			definition of "home" depends on whether origin mode
			is set and, if so, whether a scrolling region has been
			define.  These are DEC-private, by the way.

	ESC c		RIS (Reset to Initial State).  Surprise!  This one is
			ANSI-standard, too.

If you wanted to clear the scrolling region, you'd use:

	ESC [ r		DECSTBM (Set Top and Bottom Margins).  DEC-private.
			(When issued with no parameters, this command sets
			the top and bottom margins to the whole screen).

			Note the "DEC" at the front of the name.

			This would have no defined meaning on a generic
			ANSI-compliant terminal.  There's no mechanism for
			reserving private extensions, so some other manufac-
			turer could define it to do something quite different.

							-- Jerry
