Trailing-Edge
-
PDP-10 Archives
-
integ_tools_tops20_v7_30-apr-86_dumper
-
tools/sed-for-vms/sed.rnd
There are 16 other files named sed.rnd in the archive. Click here to see a list.
.; NOTE: This is the TOPS-20 version of the reference manual. The VAX
.; version is not yet ready. Except for the command invocations
.; and a couple of features which VMS lacks, this is accurate.
.;
.ps 56 68 .autop
.VAR OLDTAB A B
.VAR TOPS20 A B
.FLAG MACRO {
.DEFINE DATE August 1983
.DEFINE TERMNL VT52
.DEFINE ENTER PAD-BLUE
.DEFINE CURLFT LEFT
.DEFINE CURRGT RIGHT
.DEFINE CURUP UP
.DEFINE CURDWN DOWN
.DEFINE CURHOM PAD-BLK
.DEFINE CARRET _^M
.DEFINE ROLBKP _^A
.DEFINE ROLBKL _^W
.DEFINE SRCBAK _^E
.DEFINE SRCFWD _^R
.DEFINE ROLFWL _^T
.DEFINE ROLFWP _^Y
.DEFINE INSSPS _^K
.DEFINE DELSPS _^L
.DEFINE INSLIN _^D
.DEFINE DELLIN _^F
.DEFINE PUT _^G
.DEFINE SETFIL _^B
.DEFINE TAB _^I
.DEFINE ENTCCH _^O
.DEFINE PERCGO _^P
.DEFINE CLRLIN (none)
.DEFINE EXECUT _^X
.DEFINE EXIT _^Z
.DEFINE ABORT _^C
.DEFINE RECALL PAD-RED
.DEFINE INSMOD PAD-ENT
.DEFINE DELCHR BKSP
.DEFINE RESET DELETE
.DEFINE REALTB PAD-0
.DEFINE MARK PAD-7
.DEFINE LINE (none)
.DEFINE CASE PAD-9
.DEFINE WINDOW PAD-1
.DEFINE ERSLIN _^J
.DEFINE UPTAB PAD-8
.DEFINE DWNTAB PAD-2
.DEFINE REWRIT PAD-5
.DEFINE HELPER PAD-3
.DEFINE SAVER (none)
.DEFINE BEGLIN PAD-4
.DEFINE ENDLIN PAD-6
.DEFINE ERASWD PAD-.
.DEFINE SUBSTI _^_~
.DEFINE TBSET ESC-S
.DEFINE JUSTI ESC-J
.DEFINE MARKER _^_?
.IFNOT OLDTAB
.DEFINE PICK _^V
.DEFINE BAKTAB _^U
.DEFINE SWITCH _^N
.DEFINE SLDLFT ESC-L
.DEFINE SLDRGT ESC-T
.ELSE OLDTAB
.DEFINE PICK _^K
.DEFINE BAKTAB _^N
.DEFINE SWITCH _^V
.DEFINE SLDLFT _^L
.DEFINE SLDRGT _^U
.ENDIF OLDTAB
.IF TOPS20
.DEFINE PUSHER _^_\
.ENDIF TOPS20
.NONUM
.C;SED: A CRT EDITOR FOR TOPS-10 AND TOPS-20
.C;WRITTEN AND DOCUMENTED BY A Christopher Hall
.C;{DATE
.S1
.IFNOT TOPS20
.C;(Operating system: TOPS-10)
.ELSE TOPS20
.C;(Operating system: TOPS-20)
.ENDIF TOPS20
.C;(Terminal described: {TERMNL)
.S2
.c;HISTORICAL NOTE AND COMMERCIAL
The first successful full-screen text editor was developed in 1967 at the
Institute for Defense Analyses by Edgar T. Irons and Franz M. Djorup for use
on the CDC-6600 computer. It was the first editor to use function keys as
editor commands.
The Yale editor "E" was designed in 1970 by Irons and Peter Weiner to be an
improvement on the IDA editor, running on the PDP-10. It demonstrated the
practicality of screen editors on terminals of speeds as low as 2400 baud.
The Rand editor "re" was conceived in 1974 by Peter Weiner, based on the Yale
editor. It was initially designed and written by Walter Bilofsky, and
modified by both Weiner and Bilofsky as it evolved. The Rand editor is used
on PDP-11 computers running the UNIX operating system. It is extremely
powerful, although limited to a small number of terminals.
The author has worked with the Yale editor and the Rand editor for the past
several years, and has made a number of modifications and improvements to a
version of the former. The editor described in this document was initially
developed by the author in 1978 to be a more tightly-written and better
documented program than the Yale editor, but it has evolved into an entity
in its own right. It incorporates useful features from both the Rand editor
and the Yale editor plus several new ideas. It runs on the PDP-10 under
either the TOPS-10 or TOPS-20 monitor (in native mode).
This editor, SED, has a number of features which distinguish it from its
predecessors. It is the first full-screen editor which is easily adaptable to
a wide range of terminals. Also, it exploits more fully the capabilities of
the terminal by letting the terminal do the work whenever possible (and it
knows whether or not the terminal is equipped to do the job). The editor has
the usual set of commands: cursor movers, display window movers, inserts and
deletes, searches, and line copiers; all of which cause immediate changes on
the CRT screen when they are invoked. In addition, there are a number of
shortcuts by which the sophisticated user can save time and typing.
SED is documented with a tutorial, this user's manual, an installation
guide, copious comments within the program, and software history files and
suchlike for the serious hacker.
.PG .NUM 1
.HEADER BOTTOM
.C;INTRODUCTION TO SED
This document contains complete descriptions of all the SED commands.
It is a reference manual, good for finding out the details about things,
but confusing as a learning tool. If you are a new SED user, the tutorial
guide, DOC:SED.MAN would be a better introduction to the editor. But the
tutorial will only get you started; use this reference manual when you
feel comfortable with the basic SED commands.
Commands to SED are given by holding down the key labeled 'CNTR' or 'CTRL'
and typing a letter. Also, some terminals have special keys which are SED
commands. In this manual editor commands are notated in capital letters and
brackets. For example, the command used to advance 1 page (one screenful) in
a file is given by holding the 'CONTROL' key and typing the letter 'Y'; it is
notated <ROLL-FORWARD-PAGES> or {ROLFWP. Think of the 'CONTROL' key as a
shift key.
.S3
.C;RUNNING SED
There are three ways of running the editor. At monitor level type
.IFNOT TOPS20
.S1 .LIT
.R SED or
.R SED;FILE.EXT or
.R SED;FILE.EXT=
.EL .S1
.ELSE TOPS20
.S1 .LIT
@SED or
@SED FILE.EXT or
@SED FILE.EXT=
.EL .S1
.ENDIF TOPS20
The first way will set up for editing the file (and alternate file*) which
were edited last time. SED will return you to the same position in the file
that you were last. If you are running SED for the first time a cheery
welcome message will appear. You can then use the <SET-FILE> command
(described below) to get a file to edit.
.IF TOPS20
(See Appendix F. for more information on the command format).
.ENDIF TOPS20
.FOOTNOTE 3
.LEFT MARGIN 0
* SED allows you to toggle between any two files. The alternate file is the
one which is not currently being edited. See the <SET-FILE> command.
!
If you type the second form SED will put you at the beginning of FILE.EXT (or
whatever file you give). The file you edited last time will be the alternate
file. If the file given in the run command does not exist SED will tell you
so and give you last time's file or the cheery message.
If you use the third form SED will do as above, except that if the file does
not exist the editor will create it.
.PG
.C;THE CURSOR
Run the editor and get the cheery message (don't give a file name). There
will be a blinking dash in the upper left-hand corner of the screen; that
dash is called the cursor, and it represents your position in the file.
You cannot do much with the cheery message, so why not create a new
file and play around with the editor? Type the following:
.S1 .NOFILL
{ENTER A B C D = {SETFIL
.FILL .S1
one key after another (with no spaces in between). The editor will create
a file named ABCD containing the obvious message
.S1 .LIT
; This file is ABCD.
.EL .S1
Note that the cursor is still at the upper left of the screen.
If you type a character it will appear where the cursor is and the cursor
will move one space to the right. The character you type will replace the one
at the cursor position. If INSERT MODE is in effect (see the <INSERT-MODE>
command) the new character will be inserted in the line, and the rest of the
line moved one position to the right.
A carriage return will put the cursor at the start of the next line on the
screen, but will not affect the contents of the file. There are other keys
that simply move the cursor around: the key off to the right labeled HOME and
the four arrow keys near it (which move the cursor in the indicated
directions), <TAB> ({TAB) and <BACKTAB> ({BAKTAB). There's also a command
(<LINE>) to move to the beginning or the end of a line. Linefeed is NOT
merely a cursor mover, however (see the section on <CLEAR-LINE> or
<ERASE-LINE>).
Type something. If you make a mistake use the <CURSOR-LEFT> key to back up
over the bad character, and type the correct one on top of it. If you put in
one character too many, position the cursor to it and type <DELETE-SPACES>
({DELSPS). There is a whole spectrum of commands for inserting and deleting
characters and lines of text, moving around within the file, and other useful
things. Read on.
.PG
.C;FORMAT OF EDITOR COMMANDS
.INDEX ^^ENTER ENTER PARAMETER {ENTER #\\
.INDEX ^^PARAMETER ENTER PARAMETER {ENTER #\\
SED does not exist by cursor movements alone. There is a whole raft of
commands, which have the following form:
.S1 .LIT
<COMMAND>
or <ENTER><parameter><COMMAND>
or <ENTER><cursor movement><COMMAND>
or <ENTER><COMMAND> (TOKEN FORMAT)
.EL .S1
The <ENTER> key is {ENTER. A parameter is a piece of information that is used
by a command. For instance, if you want to insert blank lines in a file the
parameter is the number of lines you want to insert. Most commands have
default parameters; that is, if you simply issue the command without
specifying parameters, SED will use a pre-set value for each parameter.
<INSERT-LINES> ({INSLIN) will insert 1 line until you tell it otherwise,
which you do by using the second form of command. If you say
.S1 .LIT
<ENTER>5<INSERT-LINES>
.EL .S1
SED will insert 5 lines. From that point on whenever you say
.S1 .LIT
<INSERT-LINES>
.EL .S1
SED will insert 5 lines, until you change it again.
Certain pairs of commands share parameters. If you say
.S1 .LIT
<ENTER>5<INSERT-LINES>
.EL .S1
and later say
.S1 .LIT
<DELETE-LINES>
.EL .S1
SED will delete 5 lines because it shares its parameter with
<INSERT-LINES> (<DELETE-LINES> = {DELLIN).
If the /RESET switch is set (see the section on <SWITCH>), parameters
will be reset to their initial values after each command.
There are two other command formats: cursor movement and tokens. Both are
shortcuts for the advanced user, so if you are just starting out don't worry
about them for now.
.S1 .I0
CURSOR MOVEMENT
To define a parameter by cursor movement, type <ENTER>, move the cursor
around using any of the cursor movement commands to get to a different
row and/or column on the terminal screen, then type the desired command.
The command will take as its parameters the difference in rows or columns
(or both). Often both changes are useful; you can delete three lines and
the first 8 spaces of the next line by typing
.S1 .LIT
<ENTER><RETURN><RETURN><RETURN><TAB><DELETE-LINES>
.EL .S1
Cursor movement is often easier to do than counting lines or characters.
It is not meaningful with every command; an error message will result from
using it at the wrong time. But usually it is used for what you think it
ought to be, and illegal when you can't see any reason for doing it.
.S1 .I0
TOKENS
When you give a token parameter you ask SED to read the parameter from the
file itself, at the cursor position. However, for some commands token format
means that special action should be taken, the nature of which is dependent
on the command. For example, <ENTER><PERCENT> will move to the end of the
file; it's a shorthand way of saying <ENTER>100<PERCENT>. Also, <ENTER><PUT>
will insert text which has previously been deleted into the file; it is the
only way of recovering that information. The special functions of this
command format are described in detail with each command.
A token can be thought of as "the word at the cursor position." Actually the
formal definition is: the string of characters starting at the cursor
location and extending to the first nonalphabetic, nonnumeric character. Thus
the location of the cursor is important, and it need not be at the start of
an actual word. Sometimes SED is interested in the token itself, for instance
for the search command; sometimes it is only interested in the length of the
token, like for <DELETE-SPACES>.
The token for the <SET-FILE> ({SETFIL) command is different.
The parameter for <SET-FILE> (set up a new file for editing), is
a filespec. Since filespecs usually consist of at least a name and an
extension, the usual token will not work, since it will recognize only the
file name (and stop on the "."). So for the <SET-FILE> command a token is
defined as extending from the cursor position to the next space, tab, comma,
or control character. All other characters will be legal token characters.
Tokens, of course, don't give you anything you don't have already;
they just save some typing.
.S1 .I0
EDITING A PARAMETER
Parameters can be edited: an entire parameter can be canceled by typing
<RESET> (which is invoked by {RESET). The latest character of the parameter
can be deleted by typing <CURSOR-LEFT> or <DELETE-CHARACTER> ({DELCHR).
The latest word of the parameter can be erased by <BACKTAB> ({BAKTAB) or
<ERASE-WORD> ({ERASWD), and the whole parameter erased by <ERASE-LINE>
({ERSLIN).
.PG
.C;COMMANDS TO THE EDITOR
This section describes all the commands to SED, telling what they are, how
they are (normally) invoked, what types of parameters they accept, and any
special attributes they have.
Each entry in this section describes one command, or two if they are related
and share a parameter value. The format of the entries in this sections is:
.S1 .LIT
COMMAND NAME(S) CONTROL CHAR(S) WHICH INVOKE
STARTING NOMINAL
TYPE OF PARAMETER EXPECTED
CURSOR MOVEMENT DEFINED (NONE, CHANGE OF ROW, COLUMN, BOTH)
TOKEN (NONE, TOKEN, OR SPECIAL)
DESCRIPTION OF THE ACTION TAKEN BY THE COMMAND
EXAMPLE: A LINE OR TWO BEFORE AND AFTER EDITING
.EL
.PG .LIT
***************************************************************************
.EL
The commands on this page and the next (through <LINE>) are the cursor
movement commands. They move the cursor around on the terminal screen.
but have no effect on the file. These commands can be used in a
parameter to define a region of the screen, between where the cursor was
when you typed <ENTER> to where it was when you typed a non-cursor movement
command.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<CURSOR-RIGHT>
<CURSOR-LEFT>
<CURSOR-UP>
<CURSOR-DOWN>
.FILL
.INDEX ^^CURSOR MOVE CURSOR UP {CURUP #\\
.INDEX ^^UP MOVE CURSOR UP {CURUP #\\
.INDEX ^^MOVE MOVE CURSOR UP {CURUP #\\
.INDEX ^^CURSOR MOVE CURSOR DOWN {CURDWN #\\
.INDEX ^^DOWN MOVE CURSOR DOWN {CURDWN #\\
.INDEX ^^MOVE MOVE CURSOR DOWN {CURDWN #\\
.INDEX ^^CURSOR MOVE CURSOR LEFT {CURLFT #\\
.INDEX ^^LEFT MOVE CURSOR LEFT {CURLFT #\\
.INDEX ^^MOVE MOVE CURSOR LEFT {CURLFT #\\
.INDEX ^^CURSOR MOVE CURSOR RIGHT {CURRGT #\\
.INDEX ^^RIGHT MOVE CURSOR RIGHT {CURRGT #\\
.INDEX ^^MOVE MOVE CURSOR RIGHT {CURRGT #\\
Moves the cursor in the indicated direction. If the cursor goes off the
edge of the screen it will wrap around and appear at the other side.
If /SCROLL is set, <CURSOR-UP> at the top of the screen or <CURSOR-DOWN>
at the bottom will roll the screen one line. See the <SWITCH> command.
.S2 .LIT
***************************************************************************
.EL .NOFILL
<CURSOR-HOME> {CURHOM
.FILL
.INDEX ^^CURSOR MOVE CURSOR HOME {CURHOM #\\
.INDEX ^^HOME MOVE CURSOR HOME {CURHOM #\\
.INDEX ^^MOVE MOVE CURSOR HOME {CURHOM #\\
Moves the cursor to the upper left-hand corner of the screen.
.S2 .LIT
***************************************************************************
.EL .NOFILL
<UP-TAB> {UPTAB
<DOWN-TAB> {DWNTAB
.FILL
.INDEX ^^UP MOVE CURSOR UP 6 LINES {UPTAB #\\
.INDEX ^^TAB MOVE CURSOR UP 6 LINES {UPTAB #\\
.INDEX ^^DOWN MOVE CURSOR DOWN 6 LINES {DWNTAB #\\
.INDEX ^^TAB MOVE CURSOR DOWN 6 LINES {DWNTAB #\\
Moves the cursor up or down six lines. Exactly the same as typing six
<CURSOR-UP> or <CURSOR-DOWN>s.
.S2 .LIT
***************************************************************************
.EL .NOFILL
<CARRIAGE-RETURN> {CARRET
.FILL
.INDEX ^^RETURN CARRIAGE RETURN {CARRET #\\
Moves the cursor to the start of the next line (ie, does a CARRIAGE RETURN,
LINEFEED). If the cursor started on the bottom line of the screen the
screen will roll up one line, a new line will appear at the bottom of the
screen, and the cursor will be positioned at the start of it.
In INSERT mode (see the <INSERT-MODE> command) <CARRIAGE-RETURN> will
actually insert a <CRLF> at the cursor location.
Switches relevant to <CARRIAGE-RETURN>: /ROLL and /ICR. See the <SWITCH>
command.
.PG .LIT
***************************************************************************
.EL .NOFILL
<TAB> {TAB
<BACKTAB> {BAKTAB
.FILL
.INDEX ^^MOVE MOVE TO NEXT TAB STOP {TAB #\\
.INDEX ^^TAB MOVE TO NEXT TAB STOP {TAB #\\
.INDEX ^^NEXT MOVE TO NEXT TAB STOP {TAB #\\
.INDEX ^^MOVE MOVE TO NEXT TAB STOP TAB #\\
.INDEX ^^TAB MOVE TO NEXT TAB STOP TAB #\\
.INDEX ^^NEXT MOVE TO NEXT TAB STOP TAB #\\
.INDEX ^^MOVE MOVE TO PREVIOUS TAB STOP {BAKTAB #\\
.INDEX ^^TAB MOVE TO PREVIOUS TAB STOP {BAKTAB #\\
.INDEX ^^PREVIOUS MOVE TO PREVIOUS TAB STOP {BAKTAB #\\
Moves to the next tab stop in the forward or backward direction. Tab stops
are normally 8 spaces apart.
Tabs can be set to be any constant distance apart, using the <SWITCH> command.
For example, to set tabs every 6 columns, type
.S1 .LIT
<ENTER>T:6<SWITCH>
.EL .S1
Also, you can use the <TAB-SET> command to set additional tabs or clear
tabs you don't want. See the <TAB-SET> section.
There are also wordwise tabs, which you ask for by typing the /T:W or
/NOTABS switch. Then <TAB> and <BACKTAB> function in terms of words of
text rather than screen position. The table below explains what happens:
.S1 .LIT
CURSOR AT: TAB MOVES TO: BACKTAB MOVES TO:
within the line start of next word start of previous word
start of line start of next word end of previous line
last word of line end of line start of previous word
beyond end of line next normal tab stop start of last word of line
.EL
The /DELIM: switch will make wordwise tabs stop after the
given characters, in addition to stopping at the starts of words.
If you want a TAB command which actually inserts a TAB into the file you
are editing, see <REAL-TAB>.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<BEGIN-LINE> {BEGLIN
<END-LINE> {ENDLIN
.FILL
.INDEX ^^BEGINNING MOVE TO THE BEGINNING OF A LINE {BEGLIN #\\
.INDEX ^^LINE MOVE TO THE BEGINNING OF A LINE {BEGLIN #\\
.INDEX ^^END MOVE TO THE END OF A LINE {ENDLIN #\\
.INDEX ^^LINE MOVE TO THE END OF A LINE {ENDLIN #\\
<BEGIN-LINE> moves the cursor to the beginning of the line it is on.
<END-LINE> moves the cursor to the end of the line it is on.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<LINE> {LINE
.FILL
.INDEX ^^LINE MOVE TO BEGINNING OR END OF LINE {LINE #\\
.INDEX ^^BEGINNING MOVE TO BEGINNING OR END OF LINE {LINE #\\
.INDEX ^^END MOVE TO BEGINNING OR END OF LINE {LINE #\\
<LINE> does the functions of <BEGIN-LINE> and <END-LINE>: it moves the
cursor to the beginning of the line, unless it was already there, in which
case it moves to the end of the line. Thus <LINE> will usually move to
the beginning, and <LINE><LINE> will usually move to the end. It's not as
hard to do as it is to describe.
If <LINE> is typed within a parameter it acts as cursor movement, moving
to the beginning or the end of the line as above.
.PG .LIT
***************************************************************************
.EL .NOFILL
<ROLL-BACK-PAGES> {ROLBKP
<ROLL-FORWARD-PAGES> {ROLFWP
.FILL .S1 .LIT
Starting nominal: 1 page
Parameter: number (pages to roll)
Cursor move: illegal
Token: length of token is number of pages to roll
.EL
.INDEX ^^ROLL ROLL SCREEN BACKWARD PAGES {ROLBKP #\\
.INDEX ^^BACKWARD ROLL SCREEN BACKWARD PAGES {ROLBKP #\\
.INDEX ^^PAGES ROLL SCREEN BACKWARD PAGES {ROLBKP #\\
.INDEX ^^ROLL ROLL SCREEN FORWARD PAGES {ROLFWP #\\
.INDEX ^^FORWARD ROLL SCREEN FORWARD PAGES {ROLFWP #\\
.INDEX ^^PAGES ROLL SCREEN FORWARD PAGES {ROLFWP #\\
Advances or backs up the given number of pages (screens-full) of the file.
<ROLL-BACK-PAGES> won't move before the start of the file, and
<ROLL-FORWARD-PAGES> won't move beyond the end of the file. Use <INSERT-LINES>
to extend the file, not one of the rolls. Or just type on the last line on
the screen.
Another way to get back to where you were before the <ROLL-PAGES> command
is to type <ENTER>P<MARKER>.
.S6 .LIT
***************************************************************************
.EL .NOFILL
<ROLL-BACK-LINES> {ROLBKL
<ROLL-FORWARD-LINES> {ROLFWL
.FILL .S1 .LIT
Starting nominal: 1/3 page
Parameter: number (lines to roll)
Cursor move: lines only; spaces ignored
Token: causes scanning to begin
.EL
.INDEX ^^ROLL ROLL SCREEN BACKWARD LINES {ROLBKL #\\
.INDEX ^^BACKWARD ROLL SCREEN BACKWARD LINES {ROLBKL #\\
.INDEX ^^LINES ROLL SCREEN BACKWARD LINES {ROLBKL #\\
.INDEX ^^ROLL ROLL SCREEN FORWARD LINES {ROLFWL #\\
.INDEX ^^FORWARD ROLL SCREEN FORWARD LINES {ROLFWL #\\
.INDEX ^^LINES ROLL SCREEN FORWARD LINES {ROLFWL #\\
.INDEX ^^SCAN SCAN THE FILE FORWARD {ROLFWL #\\
.INDEX ^^SCAN SCAN THE FILE BACKWARD {ROLBKL #\\
Same as above, but movement is in terms of lines.
Scanning is when the editor does a roll (using the current nominal),
delays for a short time, then rolls again. It is started by typing
<ENTER><ROLL-BACK-LINES> (or <ENTER><ROLL-FORWARD-LINES>). Scanning
continues until the start or end of the file is reached, or until you type
any character. When scanning stops the cursor is left in the middle of the
screen to indicate that scanning is over.
.PG .LIT
***************************************************************************
.EL .NOFILL
<PERCENT-GOTO> {PERCGO
.FILL .S1 .LIT
Starting nominal: 0 percent
Parameter: number between 0 and 100 (percent distance into file)
Cursor move: illegal
Token: acts same as 100 percent
.EL
.INDEX ^^GO TO GO TO PERCENT IN FILE {PERCGO #\\
.INDEX ^^PERCENT GO TO PERCENT IN FILE {PERCGO #\\
Displays the file starting the given percent into it. Typing
<ENTER>50<PERCENT-GOTO> will display the file starting mid-way through.
<ENTER>100<PERCENT-GOTO> (or just <ENTER><PERCENT-GOTO>) will display the
last several lines of the file, with the cursor at the end of the last line.
<PERCENT-GOTO> is very much faster than a <ROLL> command, since it does not
count lines of text. It can also be used to speed up searching, by <GOTO>ing
a place near where you expect a match and then doing the search.
To get back to where you were before the <PERCENT-GOTO> command, type
<ENTER>P<MARKER>.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<SLIDE-LEFT> {SLDLFT
<SLIDE-RIGHT> {SLDRGT
.FILL .S1 .LIT
Starting nominal: 8 spaces
Parameter: number (spaces to slide)
Cursor move: spaces only; lines ignored
Token: takes length of token
.EL
.INDEX ^^SLIDE SLIDE WINDOW LEFT {SLDLFT #\\
.INDEX ^^WINDOW SLIDE WINDOW LEFT {SLDLFT #\\
.INDEX ^^LEFT SLIDE WINDOW LEFT {SLDLFT #\\
.INDEX ^^SLIDE SLIDE WINDOW RIGHT {SLDRGT #\\
.INDEX ^^WINDOW SLIDE WINDOW RIGHT {SLDRGT #\\
.INDEX ^^RIGHT SLIDE WINDOW RIGHT {SLDRGT #\\
Moves screen window left or right within the file. For example, after a
nominal <SLIDE-RIGHT>, the first 8 characters of each line will not appear
on the screen (they will be off the left of the window), and each line
displayed will show character positions 9 through 88.
The window can be slid right an unlimited amount, but it cannot be slid
left past the left margin. The cursor will always stay on the line it is
on at the start of the slide; if the character it pointed to is out of the
window the cursor will point to the first character on that line that is
on the screen.
_/SLIDE:n sets the nominal distance to slide to be n. /NOSLIDE disables
<SLIDE-LEFT> and <SLIDE-RIGHT>. See the <SWITCH> command.
.PG .LIT
***************************************************************************
.EL .NOFILL
<INSERT-LINES> {INSLIN
<DELETE-LINES> {DELLIN
.FILL .S1 .LIT
Starting nominal: 1 line
Parameter: number (lines to insert/delete)
Cursor move: lines; spaces counted from left margin
Token: takes size of token, plus 1
.EL
.INDEX ^^LINES INSERT LINES IN FILE {INSLIN #\\
.INDEX ^^INSERT INSERT LINES IN FILE {INSLIN #\\
.INDEX ^^LINES DELETE LINES FROM FILE {DELLIN #\\
.INDEX ^^DELETE DELETE LINES FROM FILE {DELLIN #\\
<INSERT-LINES> adds the given number of lines at the CURSOR POSITION (the
given number of spaces are also added, if they are asked for by a cursor
move. Thus if the cursor is on line 5 column 8, and you type
<ENTER><CURSOR-DOWN><INSERT-LINES>; then one line and 8 spaces will be added
to the file. The result will be to split line 5, moving everything after the
cursor one line down. If, in the same position, you type
<ENTER><CARRIAGE-RETURN><INSERT-LINES>; then one line (and no spaces) will be
added, and the result will be to split the line and put the stuff after the
cursor on the next line, starting at the left margin. That last command could
also be made by typing <ENTER>1<INSERT-LINES>.
<DELETE-LINES> removes lines (and spaces, using cursor movement) analogously.
The deleted material is not thrown away, but is stored in the DELETE BUFFER.
It can be replaced in the file using the <PUT> command. So if you
<DELETE-LINES> accidentally, type <ENTER><PUT> to get the lines back.
Deletes of any size are permitted. However, if more than about 50 lines are
deleted, the delete buffer will overflow and it will be written out on disk.
Everything will work normally, although perhaps a little more slowly.
However, there will be a new file, nnnCLS.TMP (where nnn is your job number)
sitting around in your area.
Overflowing the delete buffer has its advantages. nnnCLS.TMP is just another
file, and can be edited the same as any other file. It will stay around
until it is overwritten by another large delete. Thus suppose you do
a 100-line delete, then a 1-line delete, and then realize that part of those
100 lines should not have been deleted. You cannot type <ENTER><PUT> to
restore those lines, since what you will get is the single latestly-deleted
line. But you can get into the nnnCLS.TMP file (using the <SET-FILE> command),
<PICK> up the desired lines, return to your original file, and <PUT> the lines
back in.
Another use of the delete buffer is to split a file into two pieces (as long
as one piece is large enough to overflow the delete buffer). To create two
files, one containing the first 100 lines and the other the remainder, go to
the beginning of the file and type <ENTER>100<DELETE-LINES>. Then exit the
editor and rename nnnCLS.TMP to whatever you want; it contains the first 100
lines.
.PG .LIT
EXAMPLE: The cursor is on the word "and"; you type <ENTER>2<INSERT-LINES>
BEFORE: This is the thing of the gig
and the gig is where it's at
AFTER: This is the thing of the gig
and the gig is where it's at
.EL .S1
If you then do a <DELETE-LINES> the two blank lines will be removed. Now say
the cursor is on the "f" and you type <INSERT-LINES>:
.S1 .LIT
BEFORE: Pack my box with five dozen liquor jugs.
AFTER: Pack my box with
five dozen liquor jugs.
Two blank lines have been inserted at the cursor position.
.EL
.S1 .LIT
***************************************************************************
.EL .NOFILL
<DELETE-CHARACTER> {DELCHR
.FILL .S1 .LIT
No parameters
.EL
.INDEX ^^DELETE DELETE THE PREVIOUS CHARACTER {DELCHR #\\
.INDEX ^^CHARACTER DELETE THE PREVIOUS CHARACTER {DELCHR #\\
Deletes the character to the left of the cursor. If that character is a
tab, <DELETE-CHARACTER> deletes the entire tab. At the beginning of a line
<DELETE-CHARACTER> has no effect.
If typed in a parameter, <DELETE-CHARACTER> will still delete the character
before the cursor. Thus, in a parameter, <DELETE-CHARACTER> and <CURSOR-LEFT>
have the same effect.
Deleted stuff can be gotten back by typing <ENTER><HELP>.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<ERASE-WORD> {ERASWD
.FILL .S1 .LIT
No parameters
.EL
.INDEX ^^DELETE DELETE THE PREVIOUS WORD {ERASWD #\\
.INDEX ^^WORD DELETE THE PREVIOUS WORD {ERASWD #\\
Deletes the word to the left of the cursor. <ERASE-WORD> at the beginning
of a line deletes the last word on the previous line.
.IFNOT TOPS20
A word is a string of alphanumerics, or a single non-alphanumeric, followed
by zero or more spaces or tabs.
.ELSE TOPS20
A word is a string of alphanumerics followed by zero or one non-alphanumeric,
followed by zero or more spaces or tabs.
.ENDIF TOPS20
<ERASE-WORD> deletes words in parameters, too. Thus, in a parameter,
<ERASE-WORD> and <BACKTAB> have the same effect.
Deleted stuff can be gotten back by typing <ENTER><HELP>.
.PG .LIT
***************************************************************************
.EL .NOFILL
<ERASE-LINE> {ERSLIN
.FILL .S1 .LIT
No Parameters
If typed in a parameter, erases entire parameter
.EL
.INDEX ^^ERASE ERASE FROM CURSOR TO END OF LINE {ERSLIN #\\
.INDEX ^^LINE ERASE FROM CURSOR TO END OF LINE {ERSLIN #\\
The <ERASE-LINE> command deletes all characters from the cursor position
to the end of the line. It is also used to erase an entire parameter.
Deleted stuff can be gotten back by typing <ENTER><HELP>.
.S1 .LIT
EXAMPLES: The cursor is on the "h" of "the" and you type <ERASE-LINE>:
BEFORE: This is the thing of the gig
and the gig is where it's at
AFTER: This is t
and the gig is where it's at
.EL .S1
You type <ENTER>THING and wanted to type <ENTER>12. <ERASE-LINE> will
erase the parameter "THING" and let you start again:
.S1 .LIT
BEFORE: >THING
AFTER: >
.EL
.S2 .LIT
***************************************************************************
.EL .NOFILL
<CLEAR-LINE> {CLRLIN
.FILL .S1 .LIT
No Parameters
.EL
.INDEX ^^CLEAR CLEAR LINE BELOW CURSOR {CLRLIN #\\
.INDEX ^^LINE CLEAR LINE BELOW CURSOR {CLRLIN #\\
.INDEX ^^CURSOR CLEAR LINE BELOW CURSOR {CLRLIN #\\
Clears out the entire line which the cursor moves onto. The contents of the
line is not saved. <CLEAR-LINE> won't work if the cursor starts on the last
line of the screen. This is a dangerous command, but useful for cleaning
out a lot of garbage and leaving blank lines behind.
Note: <CLEAR-LINE> is almost the same as <ERASE-LINE> but not as useful.
It is included for compatibility, but is obsolete.
.S1 .LIT
EXAMPLE: The cursor is on the word "thing"; you type <CLEAR-LINE>
BEFORE: This is the thing of the gig
and the gig is where its at
This is another thing of the gig
AFTER: This is the thing of the gig
This is another thing of the gig
Now the cursor is where the "is" used to be on the second line.
.EL
.PG .LIT
***************************************************************************
.EL .NOFILL
<INSERT-SPACES> {INSSPS
<DELETE-SPACES> {DELSPS
.FILL
.S1 .LIT
Starting nominal: 1 space
Parameter: number (spaces to insert/delete)
Cursor move: spaces;
change in line causes rectangular insert/delete
Token: uses size of token, plus 1
.EL
.INDEX ^^SPACES INSERT SPACES IN FILE {INSSPS #\\
.INDEX ^^INSERT INSERT SPACES IN FILE {INSSPS #\\
.INDEX ^^CHARACTERS DELETE CHARACTERS FROM FILE {DELSPS #\\
.INDEX ^^DELETE DELETE CHARACTERS FROM FILE {DELSPS #\\
Adds spaces (to add a string, see <PUT>) or deletes characters starting at
the cursor position. The deleted stuff can be gotten back by typing
<ENTER><HELP>.
.S1
EXAMPLE: The cursor is at "t" of the word "thing"; you type
<ENTER>4<INSERT-SPACES>:
.S1 .LIT
BEFORE: This is the thing of the gig
AFTER: This is the thing of the gig
.EL .S1
Then you move to the "o" of "of" and type <DELETE-SPACES>:
.S1 .LIT
AFTER: This is the thing he gig
.EL
A rectangular <INSERT-SPACES> works as follows: say you type
<ENTER><DOWN><DOWN><RIGHT><RIGHT><INSERT-SPACES>. Your cursor movement has
defined a rectangle whose opposite corners are the starting and ending
cursor positions (in this case, three rows by two columns). That rectangle
will be filled with spaces. For example, say the cursor starts at the "o"
of "of" and you type <ENTER><DOWN><DOWN><RIGHT><RIGHT><INSERT-SPACES>.
The cursor ends after the "f" of "five".
.S1 .LIT
BEFORE: This is the thing of the gig,
and the gig is where it's at.
Pack my box with five dozen liquor jugs.
Quick brown fox jumped over the lazy dog.
AFTER: This is the thing of the gig,
and the gig is whe re it's at.
Pack my box with f ive dozen liquor jugs.
Quick brown fox jumped over the lazy dog.
.EL
Rectangular <DELETE-SPACES> works in a similar manner, except that all
the characters within the rectangle are removed from the file.
If you just move the cursor up or down, with no change in column position,
you will work with a rectangle whose width is the same as last time
(which is useful if you can remember what last time's parameter was).
.PG .LIT
***************************************************************************
.EL .NOFILL
<SEARCH-FORWARD> {SRCFWD
<SEARCH-BACK> {SRCBAK
.FILL .S1 .LIT
Starting nominal: none
Parameter: string (search key)
Cursor move: spaces (change of lines is illegal)
Token: uses token
.EL
.INDEX ^^SEARCH SEARCH FORWARD {SRCFWD #\\
.INDEX ^^FORWARD SEARCH FORWARD {SRCFWD #\\
.INDEX ^^SEARCH SEARCH BACKWARD {SRCBAK #\\
.INDEX ^^BACKWARD SEARCH BACKWARD {SRCBAK #\\
Searches for the given string, from the cursor position to the end of the
file, or from the cursor position to the beginning of the file. When the
key is found, a page from the file is displayed with the line containing
the match one-third of the way down the screen. If the match is on the
page displayed at the start of the search the cursor simply moves to the
first character of the match.
A search can be aborted at any time by typing RUBOUT. This is useful, for
example, when your key was erroneously typed in lower case and the file is
entirely upper case. The screen will be restored to its pre-search status.
CONTROL-? and CONTROL-LEFT-ARROW are wild search characters; they will match
any single character in their position. Thus the command
.S1 .NOFILL
<ENTER>THE{ENTCCH?E<SEARCH-FORWARD>
.FILL .S1
(see the section on <ENTER-CONTROL-CHARACTER> to find out why {ENTCCH ? sets up
a CONTROL-?) will match on THESE, THERE, THEME, or THE ERROR, among other
things.
A useful feature: <ENTER>{ENTCCH_JGIG<SEARCH-FORWARD> searches for the next
line that begins with GIG.
.S1 .I0
HINT: Searches will go faster if the first character of the key is uncommon.
Thus searching for "XECUTE" will be faster than searching for "EXECUTE", and
will accomplish the same thing.
If the /CASE switch is set searches are case-dependent; if /NOCASE they
are case-independent (ie, "t" will match "t" or "T").
You can use the <RECALL> command to get back your current or previous
search key. See the section on <RECALL>.
To get back to where you were before a <SEARCH> command, type
<ENTER>P<MARKER>.
.PG .LIT
***************************************************************************
.EL .NOFILL
<SUBSTITUTE> {SUBSTI
.FILL .S1 .LIT
Starting nominals: none
Parameters: Sstring, Rstring, or number
Cursor move: illegal
Token: illegal
.EL
.INDEX ^^SEARCH SEARCH AND SUBSTITUTE STRINGS {SUBSTI #\\
.INDEX ^^SUBSTITUTE SEARCH AND SUBSTITUTE STRINGS {SUBSTI #\\
.INDEX ^^STRINGS SEARCH AND SUBSTITUTE STRINGS {SUBSTI #\\
<SUBSTITUTE> searches forward from the cursor position for a string (the
key) and substitutes another string (the substitute string) for it. You can
tell <SUBSTITUTE> how many times to repeat the process.
There are three types of parameters to <SUBSTITUTE>:
.S1 .LIT
<ENTER>Sstring1
<ENTER>Rstring2
<ENTER>number
.EL
The first sets up string1 as the search key. The string must be preceded by
the letter "S" (or "s"). The search key is shared with the <SEARCH-FORWARD>
and <SEARCH-BACKWARD> commands.
The second parameter sets up string2 as the entity to replace occurrences
of string1. The string must be preceded by the letter "R" (or "r"). String2
may be null, meaning that string1 will be deleted wherever it is found.
The third parameter tells how many times to perform the search and substitute.
Searching is done the same as for the <SEARCH> commands, so wild characters,
case independence, and aborting the command with RUBOUT all work.
One, two, or all three of these parameters may be given at once, in any order.
The command
.S1 .LIT
<ENTER>SFOO<ENTER>Rfoo<ENTER>1000<SUBSTITUTE>
.EL .S1
will change all occurrences of "FOO" to "foo" from the cursor to the end of
the file (well, 1000 times, anyway).
Following the above command with
.S1 .LIT
<ENTER>SFUBAR<ENTER>1000<SUBSTITUTE>
.EL .S1
will change all occurrences of "FUBAR" to "foo". Note that the substitute key
was not given, so it defaulted to its latest setting.
Sometimes it is useful to set up the parameters, think about them, and
perform the substitute when everything looks right. If the numeric parameter
is omitted, them <SUBSTITUTE> will set up the key and substitute string,
but will not execute. Thus the commands
.S1 .LIT
1) <ENTER>SGIG<ENTER>RTHING<SUBSTITUTE>
2) <ENTER>1<SUBSTITUTE>
3) <SUBSTITUTE>
4) <ENTER>10<ENTER>STHIS<SUBSTITUTE>
5) <SUBSTITUTE>
work as follows:
1) Only sets up the parameters; makes no substitutions.
2) Does one <SUBSTITUTE> since there was a numeric parameter.
3) Also does one <SUBSTITUTE>.
4) Changes the search key to "THIS" and does 10 iterations.
(the replace key is still the same).
5) Also does 10 iterations.
.EL .S1
Note: <SUBSTITUTE> is the only command for which multiple <ENTER>s are
meaningful. For other commands, all <ENTER>s except the first one are
ignored.
.PG .LIT
***************************************************************************
.EL .NOFILL
<PICK> {PICK
.FILL .S1 .LIT
Starting nominal: 1 line
Parameter: number of lines
Cursor move: lines; spaces counted from left margin
Token: uses token
.EL
.INDEX ^^PICK PICK UP LINES OF FILE {PICK #\\
.INDEX ^^LINES PICK UP LINES OF FILE {PICK #\\
.INDEX ^^APPEND APPENDING LINES TO PICK BUFFER {PICK #\\
Copies lines and/or partial lines from the file into the pick buffer. The
file is not changed. Words (instead of lines) can be <PICK>ed using cursor
movement or token picks.
If a pick of more than 100 lines is done, the pick buffer will overflow and
be written out on disk. Everything will work normally, though. Picks of
any size are permitted. However, there will be a new file, nnnPIK.TMP, sitting
around on disk.
The same discussion about overflow of the delete buffer applies to overflow
of the pick buffer.
.S1 .I0
EXAMPLES: Say the cursor is at the start of the first line of:
.S1 .LIT
Pack my box with five dozen liquor jugs.
Quick brown fox jumped over the lazy dog.
Now is the time for all good people to
come to the aid of their party.
.EL .S1
If you type <ENTER>2<PICK> you will load the pick buffer with
.S1 .LIT
Pack my box with five dozen liquor jugs.
Quick brown fox jumped over the lazy dog.
.EL .S1
There's a carriage return at the end of each line. You can avoid picking the
last carriage return by using cursor movement: typing <ENTER><END-LINE><PICK>
will load
.S1 .LIT
Pack my box with five dozen liquor jugs.
.EL .S1
into the pick buffer, without the trailing carriage return.
.S1 .LIT
If you type <ENTER><RIGHT><RIGHT><RIGHT><RIGHT><RIGHT><RIGHT>
<PICK> (6 RIGHTs) you will load
Pack m
.EL .S1
And if you type <ENTER><DOWN><RIGHT><RIGHT><RIGHT><PICK> you will pick up
.S1 .LIT
Pack my box with five dozen liquor jugs.
Qui
.EL
Normally, <PICK> clears the pick buffer each time before writing it.
However, there is a switch, /APPND, which causes <PICK> to append to the
buffer. It works as follows: Type <ENTER>/APPND<SWITCH> to tell SED to
append to the pick buffer. When you do your first <PICK> after setting the
/APPND switch the pick buffer will be cleared and the text put into it. The
second <PICK> will be added at the end of the first, and so on. You can <PUT>
the buffer and then continue appending to it if you want. To stop appending
type the /NOAPPND switch. To clear the buffer and keep appending type the
/APPND switch again.
.S1 .I0
EXAMPLE: Suppose the text is (again):
.S1 .LIT
Pack my box with five dozen liquor jugs.
Quick brown fox jumped over the lazy dog.
Now is the time for all good people to
come to the aid of their party.
.EL .S1
If you type <ENTER>/AP<SWITCH> and then <PICK> the 4th line, then the 3rd, 1st,
and 2nd lines, then <PUT>, you get:
.S1 .LIT
come to the aid of their party.
Now is the time for all good people to
Pack my box with five dozen liquor jugs.
Quick brown fox jumped over the lazy dog.
.EL .S1
inserted into the file. The lines are sorted, for reason unknown.
Typing more <PICKS> will continue to append. If you then type
<ENTER>/AP<SWITCH> and do two <PICK>s, just the text from those last two
<PICK>s will be in the buffer.
You can append any kind of <PICK> to the buffer; full lines, partial lines,
single words, or a <PICK> using a <MARK> will all work.
The /APPND and /NOAPPND switches do not affect the contents of the pick
buffer directly; they just tell the following <PICK>s what to do.
.PG .LIT
***************************************************************************
.EL .NOFILL
<PUT> {PUT
.FILL .S1 .LIT
Starting nominal: none
Parameter: string (read into pick buffer)
Cursor move: spaces (change of lines is illegal)
Token: causes DELETE buffer to be PUT
.EL
.INDEX ^^LINES RESTORE DELETED LINES {PUT #\\
.INDEX ^^DELETED RESTORE DELETED LINES {PUT #\\
.INDEX ^^RESTORE RESTORE DELETED LINES {PUT #\\
.INDEX ^^INSERT INSERT TEXT IN FILE {PUT #\\
.INDEX ^^TEXT INSERT TEXT IN FILE {PUT #\\
<PUT> writes text from either the pick buffer or the delete buffer into
the file, at the cursor location. The contents of the buffer is not changed.
The normal use of this command is to <PICK> something, move to a new
location, and <PUT> it by simply typing <PUT>. Or <DELETE-LINES> something,
move, and insert it by typing <ENTER><PUT>.
PUT is also used to insert a string directly at the cursor position. The
string is given as the parameter. If you type
.S1 .LIT
<ENTER><string><PUT>
.EL .S1
then <string> will be inserted at the cursor location. The string will also
be stored in the pick buffer, so it can be inserted somewhere else by
simply typing <PUT>.
.S1
EXAMPLE: The cursor is at the "t" of the word "thing"; you type
<ENTER><RIGHT><RIGHT><RIGHT><RIGHT><RIGHT><RIGHT><PICK>, thus loading
"thing#" into the pick buffer using cursor movement. The cursor is still at
"thing". Now type <PUT>:
.S1 .LIT
BEFORE: This is the thing of the gig
and the gig is where it's at
AFTER: This is the thing thing of the gig
and the gig is where it's at
.EL .S1
Now do a <BEGIN-LINE> to get to the beginning of that line and type
<DELETE-LINES> (with the starting nominal of 1 already set up):
.S1 .LIT
AFTER: and the gig is where it's at
.EL .S1
The deleted line has been loaded into the delete buffer. Typing
<DOWN><ENTER><PUT> restores that line beneath the other one:
.S1 .LIT
AFTER: and the gig is where it's at
This is the thing thing of the gig
.EL .S1
Now position the cursor to the "w" of "where" and type
<ENTER>NOT<PUT>:
.S1 .LIT
AFTER: and the gig is NOT where it's at
This is the thing thing of the gig
.EL
.PG .LIT
***************************************************************************
.EL .NOFILL
<MARK> {MARK
.FILL .S1 .LIT
No Parameters
Token: cancels mark
.EL
.INDEX ^^MARK MARK POSITION FOR PICK OR DELETE {MARK #\\
.INDEX ^^POSITION MARK POSITION FOR PICK OR DELETE {MARK #\\
.INDEX ^^PICK MARK POSITION FOR PICK OR DELETE {MARK #\\
.INDEX ^^DELETE MARK POSITION FOR PICK OR DELETE {MARK #\\
This command extends the <PICK> and <DELETE-LINES> commands using cursor
movement to work with arbitrarily large sections of text (ie, not
restricted to within the limits of the screen). When MARK is typed
SED stores the current cursor position. You can then move to any other part
of the file and type <PICK> (or <DELETE-LINES>), and all the text between the
mark and the ending cursor position will be picked up (or deleted). It
doesn't matter whether the ending cursor position is before or after the mark.
Thus the <MARK> command temporarily changes the effect of a <PICK> or
<DELETE-LINES> without a parameter. When the <PICK> or <DELETE> has been
done, the mark is forgotten and those commands perform normally again.
Typing <ENTER><MARK> will cancel the mark and restore the screen to the way
it was when you typed <MARK>.
The mark is also forgotten if you type any command which alters the file
(including typing a character). Every other command retains the mark. You can
get to the ending position any way you want: <ROLLS>, <PERCENT-GOTO>, and
<SEARCHES> are probably the most useful ways.
Say the nominal number of lines to pick is 1. If you type:
.S1 .LIT
<MARK>
<ENTER>10<ROLL-FORWARD-PAGES>
<PICK>
.EL .S1
then 10 screens-full (say around 240 lines) of text will be <PICK>ed. Those
lines can be PUT in the normal way. If you now type another <PICK> you will
pick up 1 line.
To move back to the end of the marked region, type <ENTER>P<MARKER>.
.PG .LIT
***************************************************************************
.EL .NOFILL
<MARKER> {MARKER
.FILL .S1 .LIT
Starting nominal: none
Parameters: none, Sname, Gname, Kname, N, P, F
Cursor move: illegal
Token: Goes to default marker
.EL
.INDEX ^^MARKER MANIPULATE MARKER IN FILE {MARKER #\\
.INDEX ^^MARKER POP MARKER STACK {MARKER #\\
.INDEX ^^STACK POP MARKER STACK {MARKER #\\
<MARKER> sets markers (placeholders) in the file, goes to the named
marker, kills a marker name, and outputs the names of the currently-defined
markers. <MARKER> also is used to pop the marker stack (described below).
Note: <MARKER> is different from the <MARK>, described above.
There are 11 markers. All except one (the "default" marker) must be
named. The default marker is easy to work with but it doesn't have a
name.
.S1 .LIT
Type: To:
<MARKER> Set the default marker at current position
<ENTER><MARKER> Go to the default marker
<ENTER>Sname<MARKER> Set marker named "name" at current position
<ENTER>Gname<MARKER> Go to marker "name"
<ENTER>G<MARKER> Go to the latest-set marker
<ENTER>Kname<MARKER> Kill (delete) marker "name"
<ENTER>K*<MARKER> Kill all markers
<ENTER>N<MARKER> List the currently-defined marker names
<ENTER>P<MARKER> Pop the marker stack; go to popped location
<ENTER>F<MARKER> Pop the Marker Stack forward
.EL
When you set a marker, the position of the upper-left of the screen is saved.
When you go to a marker, SED displays the file at the saved position, with the
cursor home.
If you add or delete text in the file the marker may move some;
if so, it will usually be earlier in the file than you expect.
The marker stack is a list of the 8 most recent "interesting" positions
in the file. Positions are pushed on it automatically when certain commands
are typed (see below). Positions are popped and moved to by typing
<ENTER>P<MARKER> or <ENTER>F<MARKER>.
The screen position is pushed on the marker stack before each of these
commands is executed: <ROLL-PAGES>, <PERCENT-GOTO>, <SEARCHES> (when
the screen position changes). Also, if you use <MARK> with <PICK> or
<DELETE-LINES>, the end of the region is stacked.
Thus <ENTER>P<MARKER> will get you back to where you were before a search,
or to the end of a <MARK>ed region. if you pop too far and want to get
back to a previously-popped marker, type <ENTER>F<MARKER>.
.PG .LIT
***************************************************************************
.EL .NOFILL
<INSERT-MODE> {INSMOD
.FILL .S1 .LIT
No Parameters
.EL
.INDEX ^^INSERT INSERT/REPLACE TOGGLE {INSMOD #\\
.INDEX ^^REPLACE INSERT/REPLACE TOGGLE {INSMOD #\\
This command toggles SED between REPLACE mode and INSERT mode.
In REPLACE mode when you type a character it replaces the one at the cursor
position. In INSERT mode the new character is placed before the one at the
cursor and the rest of the line is moved one position to the right.
If SED is in REPLACE mode, typing <INSERT-MODE> will put it in INSERT
mode. Then typing <INSERT-MODE> will put the editor back in REPLACE mode.
If you type the wrong character you can delete it using the <DELETE-CHARACTER>
command (see above).
Character strings can also be inserted in the file using the PUT command,
by typing <ENTER>string<PUT>.
Switches relevant to <INSERT-MODE>: /BEEP, /ICR, and /IMODE.
See the <SWITCH> command.
.S4 .LIT
***************************************************************************
.EL .NOFILL
<SAVE-FILE> {SAVER
.FILL .S1 .LIT
No parameters
.EL
.INDEX ^^SAVE SAVE THE CURRENT FILE {SAVER #\\
.INDEX ^^FILE SAVE THE CURRENT FILE {SAVER #\\
Saves the current file without exiting.
BEWARE: <SAVE-FILE> is designed to work quickly, so it does not eliminate
nulls or trailing spaces from the file. <SAVE-FILE><ABORT> is NOT
the same as <EXIT>. If you have changed the file be sure to end the session
with an <EXIT>.
.PG .LIT
***************************************************************************
.EL .NOFILL
<SET-FILE> {SETFIL
.FILL .S1 .LIT
Starting nominal: none
Parameter: string (filespecs)
Cursor move: spaces (change of lines is illegal)
Token: uses filespec token
.EL
.INDEX ^^EDIT EDIT A NEW FILE {SETFIL #\\
.INDEX ^^FILE EDIT A NEW FILE {SETFIL #\\
.INDEX ^^NEW EDIT A NEW FILE {SETFIL #\\
Looks up and displays the given file. Saves previous file on disk. Saves
position in previous file. If new file is not found, returns to the previous
file. A backup file is made if /BACKU is set, else not.
The filespec parameter is in the usual form:
.IFNOT TOPS20
DEV:FILE.EXT[P,PN]. DEV: defaults to DSK:, [P,PN] defaults to your own.
.ELSE TOPS20
DEV:<DIR>FILE.EXT. DEV: can be a real or defined device. <DIR>
defaults to your connected directory.
.ENDIF TOPS20
If EXT is missing SED will try some extensions until one succeeds: none,
.IFNOT TOPS20
device (if you typed HLP:FILE, SED will look for HLP:FILE.HLP),
.ENDIF TOPS20
the extension of the previous file, MAC, TXT, RNO, FOR, CBL, DAT
(You can change this list of extensions by using the /EXT:
.IFNOT TOPS20
switch).
.ELSE TOPS20
switch. Default extensions cannot be longer than 5 characters).
.ENDIF TOPS20
All of the switches described for the <SWITCH> command can be given along
with the file specs (for example, DSKD:FOO.BAR/NOTABS). Particularly
relevant are the following:
.S1 .LIT
/AGAIN Set to the current file again. (or /AGAIN:n to set n percent
into the file). <ENTER>/AG<SET-FILE> is the same as typing
<ENTER>current-filespec<SET-FILE>.
/BACKU Create a backup file on exit or save.
/CREATE Create the file if it cannot be found.
/EXT:e,e Define the list of default extensions.
/GOTO:n Set to n percent into the new file.
/GREAD Set all files looked at to be read-only.
/ID:s,s Put an I.D. line at the beginning of the file when saving.
/OUT:fs Output the new file under the filespec "fs".
/QUICK Causes the file not to display after the <SET-FILE>.
/RCUR Causes the alternate file to remain and the current file to be
replaced by the new one.
/READ Declare the file read-only.
/STRIP Strip other editors' line numbers from the file.
/WRITE Declare that the file can be modified.
.EL
<SET-FILE> with no parameter will return to the previous file, to the same
page and cursor position as was last set up. The current file will then
become the new previous file. Thus you can toggle back and forth between two
files, after setting them up, simply by typing <SET-FILE>s.
If /CREATE is set the file will be created if it is not found. If /NOCREATE
is set (which is the default), a "FILE NOT FOUND" error will result. To force
the file to be created, end the filespec parameter with an "=". Then if the
file exists it will be looked up as usual, but if it does not exist a file
with a clever message will be materialized and set up for editing. SED will
not overwrite an existing file.
Exception: if you <SET-FILE> to your current file (so you can look at two
different places in the same file), SED will not save the file. This is so
the change of file pointers can take place as quickly as possible. The file
will be saved when you <SET-FILE> to a different file or <EXIT> SED.
The file is only saved on a <SET-FILE> or an <EXIT>; editing is only done
to the in-core image of the file. During the save all nulls are removed
from the file (they are added in abundance during editing), and trailing
spaces and tabs are removed from each line (unless /TRAIL is set) (NOTE:
the file can also be saved using <SAVE-FILE> or with the incremental save
switches. However, in order to save the file quickly those commands to not
remove nulls and trailing spaces and tabs. Therefore you should end every
editing session with a <SET-FILE> or <EXIT>).
.IF TOPS20
If there is not enough room for SED to save the file it will expunge the
directory before it attempts the save.
.ENDIF TOPS20
To exit SED and forget all changes, use the <ABORT> command. The file will
be the way it was before you edited it.
Appendix#A tells how filespecs default. Appendix#B tells how to move back
and forth through a list of files.
.IF TOPS20
.S4 .LIT
***************************************************************************
.EL .NOFILL
<PUSH> {PUSHER
.FILL .S1 .LIT
No parameters
.EL
.INDEX ^^PUSH PUSH TO EXEC CONTEXT {PUSHER #\\
.INDEX ^^EXEC PUSH TO EXEC CONTEXT {PUSHER #\\
Creates an inferior fork and runs the EXEC. Popping from the EXEC continues
the editing session where it left off. The file is not saved.
.ENDIF TOPS20
.PG .LIT
***************************************************************************
.EL .NOFILL
<EXIT> {EXIT
<ABORT> {ABORT
.FILL .S1 .LIT
Parameters or token causes Exit and Go (EXIT only)
.EL
.INDEX ^^EXIT EXIT AND SAVE CHANGES {EXIT #\\
.INDEX ^^SAVE EXIT AND SAVE CHANGES {EXIT #\\
.INDEX ^^CHANGES EXIT AND SAVE CHANGES {EXIT #\\
.INDEX ^^EXIT EXIT AND FORGET CHANGES {ABORT #\\
.INDEX ^^FORGET EXIT AND FORGET CHANGES {ABORT #\\
.INDEX ^^CHANGES EXIT AND FORGET CHANGES {ABORT #\\
.INDEX ^^ABORT EXIT AND FORGET CHANGES {ABORT #\\
Saves the current file and exits to the monitor.
The editor writes the file nnnSED.TMP (where nnn is your job number)
containing the names and status of the last two files edited. The next
time you run the editor the latest of these files will be automatically
set up, with display and cursor position the same as they were upon exit.
If any parameter is typed (or a token, ie, just <ENTER><EXIT>) the editor
will exit as above and run COMPIL. COMPIL will execute the last compile-class
command. Other programs can be run on exit; see the /PROG: switch.
Unless told otherwise (see the <SWITCH> command) the editor creates a backup
file, with extension .BAK, containing the file the way it was before editing.
<ABORT> does not save the file - the file ends up as it was before the
editing session.
The editor can be re-entered after a mistaken <EXIT> or <ABORT> by typing
REENTER.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<RESET> {RESET
.FILL .S1 .LIT
Starting nominal: none
Parameter: ^ (up arrow) is the only meaningful parameter
Cursor move: undefined
Token: causes rewrite of single line
.EL
.INDEX ^^RESET RESET PARAMETER OR DISPLAY {RESET #\\
.INDEX ^^PARAMETER RESET PARAMETER OR DISPLAY {RESET #\\
.INDEX ^^DISPLAY RESET PARAMETER OR DISPLAY {RESET #\\
<RESET> can cancel an entered parameter, reset the cursor
position, rewrite a single line, or re-do the entire screen. It is useful if
you don't trust what you see. If you ever get a "Timesharing ends" message
while you're editing, you'll appreciate being able to <RESET>.
<RESET> also repairs the bottom line of the screen if it has been overwritten
with status information.
.S1 .LIT
To: Type:
Cancel a parameter <RESET>
Position the cursor properly <RESET>
Repair the bottom line <RESET>
Rewrite the line the cursor is on <ENTER><RESET>
Re-do the entire screen <ENTER>^<RESET>
.EL
.PG .LIT
***************************************************************************
.EL .NOFILL
<TAB-SET> {TBSET
.FILL .S1 .LIT
Starting nominal: none
Parameter: C or D
Cursor move: illegal
Token: clear tab at cursor position
.EL
.INDEX ^^SET#TAB SET OR CLEAR TABS {TBSET #\\
.INDEX ^^CLEAR#TAB SET OR CLEAR TABS {TBSET #\\
.INDEX ^^TAB SET OR CLEAR TABS {TBSET #\\
Sets, clears, and displays tab settings. To set a tab move the cursor to
the desired column and type <TAB-SET>. To clear the tab at the cursor,
type <ENTER><TAB-SET>. To clear all tabs type <ENTER>C<TAB-SET>.
<ENTER>D<TAB-SET> displays a ruler on the bottom line of the screen, with
the tab settings marked by dashes, the left margin (/LMARGN) by a slash,
and the right margin (/RMARGN) by a backslash. The ruler can be erased by
typing <RESET>.
Tabs can also be set using the /TSET:n switch: /TSET:12 sets a tab
at column 12; /NOTSET:12 clears that tab. /TSET:14,52,36,76 sets tabs
at the four given columns; /NOTSET:12,15 clears tabs. /TSET is useful in
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
and execute (.XCT) files.
See also the <TAB> and <SWITCH> commands.
.PG .LIT
***************************************************************************
.EL .NOFILL
<REWRITE> {REWRIT
.FILL .S1 .LIT
No parameters
.EL
.INDEX ^^REWRITE REWRITE THE SCREEN {REWRIT #\\
Rewrites the entire terminal screen without changing anything. <REWRITE>
does the same thing as <ENTER>_^<RESET>, above. <REWRITE> is commonly used
to repair a screen which has been messed up by a system message or
something.
.S3 .LIT
***************************************************************************
.EL .NOFILL
<ENTER-CONTROL-CHARACTER> {ENTCCH
.FILL .S1 .LIT
No parameters
.EL
.INDEX ^^CONTROL ENTER CONTROL CHARACTER {ENTCCH #\\
.INDEX ^^CHARACTER ENTER CONTROL CHARACTER {ENTCCH #\\
.INDEX ^^ENTER ENTER CONTROL CHARACTER {ENTCCH #\\
Causes the next character typed to be a control character (a character,
now, not a command). Has no effect on commands. Thus if you type in your file
.S1 .LIT
ABC<ENTER-C-C>D
.EL .S1
the characters A, B, C, and CONTROL-D will be placed in the file.
<ENTER-C-C> also works within parameters, so
.S1 .LIT
<ENTER>ABC<ENTER-C-C>D<PUT>
.EL .S1
will put A, B, C, CONTROL-D into the file.
Note: this command is the way to get a TAB or a FORMFEED (page mark) into a
file. For a TAB type <ENTER-C-C>I, and for a formfeed type <ENTER-C-C>L.
.S3 .LIT
***************************************************************************
.EL .NOFILL
<REAL-TAB> {REALTB
.FILL .S1 .LIT
No Parameters
.EL
.INDEX ^^TAB PUT A REAL TAB IN THE FILE {REALTB #\\
The <REAL-TAB> command is identical to typing <ENTER-CONTROL-CHARACTER>I.
It replaces the character at the cursor with a tab character (or inserts
it, if the editor is in insert mode). <REAL-TAB> also can be used to put a
tab in a parameter.
See also the <TAB> command, which just moves the cursor around but does
not change the file you are editing.
.PG
.LIT
***************************************************************************
.EL .NOFILL
<SWITCH> {SWITCH
.FILL .S1 .LIT
Starting nominal: none
Parameter: string (switch name or names)
Cursor move: undefined
Token: query of nominal parameter settings
No Parameter: query of position and file status
.EL
.INDEX ^^SWITCHES SET SWITCHES OR QUERY STATUS {SWITCH #\\
.INDEX ^^QUERY SET SWITCHES OR QUERY STATUS {SWITCH #\\
.INDEX ^^STATUS SET SWITCHES OR QUERY STATUS {SWITCH #\\
<SWITCH> is used to tailor the operation of SED to the user and to
get status information. The default setting of each switch is denoted as on
(+) or off (-) where it is meaningful. To turn a switch off type "NONAME"
("NOCASE", "NOUPPER", etc.) as the parameter. You only have to type enough of
the switch name to make it unique (and "T" is good enough for TABS, "G" for
GOTO).
You can set more than one switch in one command by separating them with
slashes, for example <ENTER>TABS:6/CASE/BACK<SWITCH>.
To give a slash as an argument to a switch, type "_^/". Use "_^_^" for
up-arrow.
All of the switches below are mentioned along with the commands they control,
or in the appendices, except /HELP, /FENCE, /ITABS, /MESSAG, /UPPER, and
/ALT. They are described only here.
.LM 8
.P-8
-APPND##causes <PICK>s to append to the PICK buffer, not overwrite it, until
/NOAPPND or another /APPND is typed. See the <PICK> command.
+BACKUP#causes a backup file to be made on <SET-FILE> or <EXIT>. The file
has the extension .BAK. /NOBACKUP causes no backup file to be made.
-BEEP###causes the terminal to beep once when entering INSERT mode and twice
when entering REPLACE mode, instead of displaying a message on the screen.
+CASE###make searches case-dependent, ie, the key "THE" is different from
the key "the". /NOCASE causes the key "THE" to match the first occurrence
of "THE", "the", "tHe", _&c.
-CREATE#if <SET-FILE> cannot find a file, always create it. if /NOCREATE
is set the file will be created if the filespec ends with an "=", otherwise
SED will give an error.
-DELIM:CHARACTER-LIST#specifies a set of delimiters used by wordwise tabs.
Example: With /DELIM:.,- a wordwise tab will stop after a period, comma, or
dash, as well as at the starts of words. /NODELIM clears the list of
characters. See the section on <TAB> and <BACKTAB>.
-DTABS##causes tabs to identify themselves. With /DTABS set, each tab displays
as highlighted "."s. The columns on the screen are still aligned properly.
-EXT:EXTENSION-LIST#sets up the list of default extensions (see
<SET-FILE>). Up to 10
.IFNOT TOPS20
extensions
.ELSE TOPS20
extensions, each 5 characters long or less,
.ENDIF TOPS20
can be given. Example: /EXT:MAC,CTL,CMD.
+FENCE##causes the normal fence to be displayed after the end of the
file. /NOFENCE causes a short, discreet fence to be displayed.
-FILL###causes <JUSTIFY> to align the text against the right margin. /NOFILL
gives a ragged right margin.
-GREAD##causes all files to be set read-only until /NOGREAD is set. Can be
overridden for the current file by /WRITE or /NOREAD.
+HELP###enables on-line help when <ENTER><ENTER> or <HELP> is typed.
+ICR####causes a carriage return typed in INSERT mode to insert a carriage
return. /NOICR causes the cursor to move to the start of the next line.
-ID#####when the file is saved, /ID causes a new first line to be added
to the file. The line contains the filespec, the date and time, and the
name of the user who edited the file. The line starts with a ";". /ID:s1
will make the line start with "s1" (5 characters or fewer). /ID:s1,s2 will
also make the line end with "s2". Use "_^/" to get a "/".
+INVRT##tells the <CASE> command to invert case. /NOINVRT tells <CASE> to
use the setting of the /RAISE switch.
-ISAVE:n#causes an incremental save of the file every n commands. Default
is n=0 (ie, no incremental saves) (_#).
+ITABS##if you type something beyond the end of a line SED will insert spaces
so the character is properly positioned. If /ITABS is set SED will try to
use as many tabs as possible instead of spaces. /NOITABS causes SED to use
only spaces.
-JOURN##starts a journal. See Appendix#D for information on journaling.
#JPRE:STRING#causes <JUSTIFY> to delete STRING from the start of each line,
if it's present, and to begin each line of the justified text with STRING.
#LENG:n#causes SED to think your terminal is n lines long. Good for limiting
the amount of information written on a slow terminal. See Appendix#C.
#LMAR:n#sets the left margin to be column n. The default is 1. See
Appendix#C to find out about margins.
+MESSAG#enables the cheery message on startup and the "this file is" message
in a new file. /NOMESSAG causes those messages not to be given.
#OUT:FILESPEC#changes the name of the file being edited (or being set to,
if the switch is given in a <SET-FILE> or the run command) to be FILESPEC.
-PAGE###causes <SWITCH> to output your position as PAGE-LINE, rather than LINES
from start of file. Pages are delimited by formfeeds (_^L).
-PIND:n#sets the paragraph indentation for the <JUSTIFY> command to be n
spaces.
.IFNOT TOPS20
#PROG:FILE#sets up the name of the system program to be run when you type
<ENTER><EXIT>. The default is COMPIL. The switch /PROG:RUNOFF will cause RUNOFF
to be run instead.
.ELSE TOPS20
#PROG:FILE#sets up the name of the program to be run when you type
<ENTER><EXIT>. The default is the latest compile-class command.
The switch /PROG:SYS:RUNOFF.EXE will cause RUNOFF to be run instead.
.ENDIF TOPS20
+RAISE##used by the <CASE> command if /NOINVRT is set. /RAISE tells <CASE>
to change lower case to upper. /NORAISE tells <CASE> to change upper case
to lower.
-READ###makes the file read-only. /READ is the opposite of /WRITE.
-RESET##causes the starting nominal parameters to be reset after each
command. When set, <ENTER>5<INSERT-LINE><INSERT-LINE> will insert 6 lines.
Under /NORESET those commands will insert 10 lines.
#RMAR:n#sets the right margin to be column n. Default is the width of the
screen. See Appendix#C to find out about margins.
+ROLL###causes <CARRIAGE-RETURN> when the cursor is at the bottom of the
screen to roll forward one line. If /NOROLL is set, <RETURN> causes the
cursor to wrap around to the top line of the screen (see also /SCROLL).
-SAVE:n#causes an incremental save of the file every n characters of typein.
Default is n=0 (ie, no incremental saves) (_#).
-SCROLL#causes a <CURSOR-UP> at the top of the screen, or a <CURSOR-DOWN> at
the bottom, to roll one line (see also /ROLL). Note: to get to the bottom
line when /SCROLL is set, type <HOME><CURSOR-LEFT>.
-SHOW###causes <EXECUTE> to display as it is working. /NOSHOW updates the screen
only when the <EXECUTE> is done.
+SLIDE:n#sets the nominal parameter for the <SLIDE> commands to n. /NOSLIDE
disables the <SLIDE>s (no action is taken when they are typed).
-STRIP##causes
.IFNOT TOPS20
SOS
.ELSE TOPS20
EDIT
.ENDIF TOPS20
line numbers to be stripped from the file when it is read in. /NOSTRIP just
marks the file read-only.
+TABS###makes <TAB> and <BACKTAB> use the usual tab positions. /NOTABS sets up
wordwise TABS. See the TAB/BACKTAB section (*).
-TRAIL##causes trailing spaces and tabs to be preserved in the file when it
is saved. /NOTRAIL causes trailing things to be stripped. Warning: SED adds
trailing spaces sometimes, so set /NOTRAIL only on special occasions.
#TSET:n#sets a tab at column n. /NOTSET:n clears that tab. Also,
/TSET:n1,n2,n3 will set tabs at the given columns. See the <TAB-SET>
command.
-UPPER##causes all alphabetic characters to be converted to upper case.
Useful for a terminal whose shift lock shifts all characters.
#WIDTH:n#causes SED to think the terminal is n characters wide. Good for
terminals which can change their widths, like the VT100. See Appendix#C.
+WRITE##allows the current file to be altered. /NOWRITE causes commands which
change the file become illegal, so the file cannot be modified accidentally.
.LM 0 .S1
(_#) /ISAVE and /SAVE should be used together. When either becomes zero the
file is saved and both counters are reset. See also Appendix#D.
.S1 .I0
(*) You can also type /TABS:n to set up tab stops every n positions. Default
is TABS:8. Also, /TABS:W sets wordwise tabs (same as /NOTABS) and /TABS:R
sets regular tabs (same as /TABS).
.P4
There are four other places you can use these switches: as part of the
monitor-level RUN command, in the <SET-FILE> command, in your
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
file, and in an execute (.XCT - see the <EXECUTE> command) file. You can do
things like
.IFNOT TOPS20
.S1 .LIT
.R SED;FILE.EXT/NOCASE/TABS:5
.EL
.ELSE TOPS20
.S1 .LIT
@SED FILE.EXT/NOCASE/TABS:5
.EL
.ENDIF TOPS20
.LIT
or
<ENTER>FILE.EXT/GOTO:50<SET-FILE>
.EL .S1
.IFNOT TOPS20
and if your SWITCH.INI or SED.INI contains the line
.S1 .LIT
SED/NOCASE/UPPER
.EL .S1
.ELSE TOPS20
and if your SED.INIT contains the line
.S1 .LIT
/NOCASE/UPPER
.EL .S1
.ENDIF TOPS20
SED will set switches for searching independently of case and for upper case
characters each time it is run.
There are several switches which are not meaningful in the <SWITCH> command
but can be used with the other four methods:
.LM 8
.P-8
#AGAIN##used in a <SET-FILE> command. If you are editing FILE.FOO, typing
<ENTER>/AG<SET-FILE> acts the same as if the parameter were "FILE.FOO". Also,
/AG:nn acts the same as /AG/GO:nn - the numeric argument is a percentage of
the way into the file. Do not use explicit filespecs with /AGAIN.
#ALT####used when running SED. Swaps the current and alternate files you
were editing last time. Also, /ALT:nn acts the same as /ALT/GO:nn.
#GOTO:PERC#used when running SED or doing a <SET-FILE>. The file will
be displayed starting PERC percent of the way through. If /GOTO, no argument,
is used SED will use the current percent value.
-IMODE##puts SED in insert mode. /NOIMODE puts it in replace mode. This
switch can also be used in execute buffers ($IM_^SW).
-QUICK##causes the file not to display after the <SET-FILE>. This is useful
on a slow terminal when you are not interested in what is on the first page
of display.
#RCUR###used with <SET-FILE>; causes the current filespecs, rather than the
alternate specs, to be replaced by the given ones. Useful for "keeping your
finger" in one file (the alternate) while looking at several others.
#RECOV##typing SED/REC at monitor level tells SED to recover the previous
session using the journal. See Appendix#D for information on journaling.
#X:NAME:txt#used in
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
to set up an execute buffer named NAME with contents txt. See the
<EXECUTE> command for details.
.LM .S1
.P5
The <SWITCH> command has two other functions: to get information about the
file you are editing and to find out the current parameter settings. You
get the former by typing just <SWITCH>, the latter by typing
<ENTER><SWITCH>. The information is written on the bottom line of the
screen. The bottom line can be cleaned up by typing <RESET>.
The file information consists of the current file name, the line and position
the cursor is on (that's the line and position in the file, not the row and
column on the screen), the percent through the file, and the name of the
alternate file (if any). For example,
.IFNOT TOPS20
.S1 .LIT
FILE: *SED.RNO[401,127] LINE: 534(67%) POS: 1 ALT: FOO.BAR[10,15]
.EL .S1
.ELSE TOPS20
.S1 .LIT
FILE: *SED.RNO LINE: 534(67%) POS: 1 ALT: <CHALL>FOO.BAR
.EL .S1
.ENDIF TOPS20
says that the current file is SED.RNO (the "*" indicates that SED.RNO
has been modified), the cursor is at the start of the
534th line, which is 67% through the file, and the alternate file is FOO.BAR
(The "LINE" field can be made a "PAGE-LINE" field. See the /PAGE switch).
To find out how many lines lie between a given line and the beginning of
the file, position the cursor to the desired line and type <SWITCH>. This
function may not occur immediately, since SED has to go back to the
beginning of the file and count every line.
Typing <ENTER><SWITCH> will give you the current parameter settings. The
bottom line of the screen will look something like:
.S1 .LIT
RL:8 RP:1 PC:2 SL:8 IL:1 IS:8 PK:3,8 CS:1 SU:5 TB:8; KEY:th SUB:FOO
.EL
This message tells you that the following values are set:
.S1 .LIT
LINES TO ROLL: 8
PAGES TO ROLL: 1
PERCENT-GOTO: 2
SLIDE: 8
INSERT/DELETE LINES: 1 LINE AND 0 SPACES
INSERT/DELETE SPACES: 8
PICK: 3 LINES AND 8 SPACES
CHANGE-CASE PARAMETER: 1
NUMBER OF SUBSTITUTES: 5
TAB SIZE: 8
SEARCH KEY: th
SUBSTITUTE STRING: FOO
.EL
So if you type a <PERCENT-GOTO> command with no parameter you will go to the
2% point, a <DELETE-LINES> will delete 1 line (and <INSERT-LINES> will insert
1 line), and a PICK will pick up 3 lines.
Remember that you can use cursor movement to tell <PICK> and
<INSERT/DELETE-LINES> to work with both a number of lines and a number of
spaces; thus their parameters are expressed as two values above.
Tab size is controlled by the /TABS switch. If wordwise tabs are in
effect the tab size is "W". Similarly, settable tabs are indicated by "S".
.PG .LIT
***************************************************************************
.EL .NOFILL
<RECALL> {RECALL
.FILL .S1 .LIT
Starting nominal: none
Parameter: type of thing to recall
Cursor move: ignored
Token: ignored
.EL
.INDEX ^^RECALL RECALL PARAMETER {RECALL #\\
.INDEX ^^PARAMETER RECALL PARAMETER {RECALL #\\
.INDEX ^^SEARCH RECALL CURRENT/FORMER SEARCH KEY {RECALL #\\
.INDEX ^^FILESPECS RECALL CURRENT FILESPECS {RECALL #\\
Simulates an <ENTER> with the latest parameter typed. When <RECALL> is typed
the latest parameter appears at the bottom of the screen and the editor is
placed in <ENTER> mode. You can add to or delete from the parameter and then
pass that parameter to a command by typing that command.
The <RECALL> command is useful in three instances: when you have typed the
wrong parameter to a command and want to correct it with a minimum of
typing, when you have given the right parameter to the wrong command (in
which case you can just type <RECALL><RIGHT-COMMAND>), and when you just
want to see what the latest parameter was.
<RECALL> uses the first character of its parameter to recall some
specific information. Type <ENTER><char><RECALL> to get the following:
.S1 .LIT
char THING RECALLED
F Current Filespecs
A Alternate Filespecs
S Current Search Key
O Previous (Old) Search Key
R Substitute String
.EL .S1
After you recall one of these things you can use it as if you typed it in:
For example, <ENTER>F<RECALL><PUT> will put the current filespecs into
the file.
.S1 .LIT
***************************************************************************
.EL .NOFILL
<HELP> {HELPER
.FILL .S1 .LIT
No explicit parameters
Token: Recovers from a delete command
.EL
.INDEX ^^HELP GIVE ON-LINE HELP {HELPER #\\
Asks for on-line help. When <HELP> is typed SED asks for the command to
give help for and waits for you to type any command. SED then tells you
something about the command and asks for another. Typing "G" or "g" instead
of a command will return you to the file. Try it.
<HELP><HELP> gives a layout of the commands on the terminal.
Typing <ENTER><HELP> will put back the text deleted by the latest
<DELETE-CHARACTER>, <ERASE-WORD>, <DELETE-SPACES> or <ERASE-LINE>
command.
.PG .LIT
***************************************************************************
.EL .NOFILL
<CASE> {CASE
.FILL .S1 .LIT
Starting nominal: 1 character
Parameter: number of characters
Cursor move: characters and/or lines
Token: uses token
.EL
.INDEX ^^CASE CHANGE CASE OF LETTERS {CASE #\\
The <CASE> command changes the case of the letters starting where the cursor
is. Two switches control way <CASE> works: /INVRT and /RAISE.
If /INVRT is set (default) then the case of all the letters is inverted:
upper is changed to lower and lower to upper.
If /NOINVRT is set then the setting of /RAISE is used.
Lower case letters will be changed to upper case if the /RAISE switch
is set; if /NORAISE, upper case letters will be changed to lower case.
Cursor movement can be used to make <CASE> work across more than one line.
.S1 .LIT
EXAMPLE: The cursor is at the start of "BOX" and you type
<ENTER><RIGHT><RIGHT><RIGHT><CASE>:
BEFORE: PACK MY BOX WITH FIVE DOZEN LIQUOR JUGS.
QUICK BROWN FOX JUMPED OVER THE LAZY DOG.
AFTER: PACK MY box WITH FIVE DOZEN LIQUOR JUGS.
QUICK BROWN FOX JUMPED OVER THE LAZY DOG.
You then type <ENTER><RETURN><CASE>:
AFTER: PACK MY box with five dozen liquor jugs.
QUICK BROWN FOX JUMPED OVER THE LAZY DOG.
***************************************************************************
.EL .NOFILL
<WINDOW> {WINDOW
.FILL .S1 .LIT
No Parameters
.EL
.INDEX ^^WINDOW SET OR CLEAR SPLIT-SCREEN WINDOW {WINDOW #\\
.INDEX ^^SPLIT SET OR CLEAR SPLIT-SCREEN WINDOW {WINDOW #\\
The <WINDOW> command allows two files to appear on the terminal screen at
the same time. The files each occupy half of the screen, one on top of the
other.
Type <WINDOW> to divide the screen. The file you are currently editing will
be displayed in the top window. If you then do a <SET-FILE>, the new file
will appear in the bottom window. Another <SET-FILE> will get you back to the
top window, and so on.
To get back to normal, non-windowing operations, type <WINDOW> again. The
split screen will go away and the file you are currently editing will
occupy the entire screen.
.PG .LIT
***************************************************************************
.EL .NOFILL
<JUSTIFY> {JUSTI
.FILL .S1 .LIT
Starting nominal: 1 line
Parameter: number of lines, or "C" to center line
Cursor move: lines (change in spaces is ignored)
Token: justify to end of paragraph
.EL
.INDEX ^^JUSTIFY JUSTIFY TEXT {JUSTI #\\
Justifies or centers text within the left and right margins. The number of
spaces to indent a new paragraph (positive, negative, or zero) may also be
given, as well as a string to be removed from the start of each raw line
and added to each justified line. The text can be made flush against the
right margin ("filled") or left ragged.
<JUSTIFY> works from the start of the line the cursor is on for the number
of lines given explicitly or using cursor movement. If <ENTER><JUSTIFY> is
typed the command works from the line the cursor is on to the end of the
paragraph (The end of a paragraph is signaled by a blank line or a line
which begins with spaces or tabs. The line the cursor is on is always the
start of a paragraph; the last line of the range is always the end of a
paragraph).
A new paragraph in the justified text begins with a blank line and the
indentation given by /PIND:.
.S1 .LIT
Switches used by <JUSTIFY>:
/JPRE:s delete string s from each input line;
begin each output line with string s.
/LMAR:n set the left margin to be column n.
/RMAR:n set the right margin to be column n.
/PIND:n set the paragraph indentation to be n columns.
/FILL make a straight right margin (/NOFILL == ragged).
.EL
To center the line the cursor is on, type <ENTER>C<JUSTIFY>. The line will
be centered between the left and right margins.
.S1 .LIT
EXAMPLE: The cursor is somewhere in the first line. You type
<ENTER>JPRE:;/RMAR:55/PIND:3<SWITCH> and then <ENTER>5<JUSTIFY>.
BEFORE:
;When in the course of human events it becomes necessary for one
;people to dissolve the political bands which have connected them.
Pack my box with five dozen liquor jugs.
AFTER:
; When in the course of human events it becomes
;necessary for one people to dissolve the political
;bands which have connected them.
; Pack my box with five dozen liquor jugs.
EXAMPLE: The cursor is somewhere in the line. You type
<ENTER>/RMAR:68<SWITCH> and then <ENTER>c<JUSTIFY>.
BEFORE:
TABLE OF CONTENTS
AFTER:
TABLE OF CONTENTS
EXAMPLES: The cursor is somewhere in the first line. You type
<ENTER>RMAR:55/PIND:3<SWITCH> and then <ENTER><JUSTIFY>.
BEFORE:
When in the course of human events it becomes necessary for one
people to dissolve the political bands which have connected them.
Pack my box with five dozen liquor jugs.
AFTER:
When in the course of human events it becomes
necessary for one people to dissolve the political
bands which have connected them.
Pack my box with five dozen liquor jugs.
Now you type <ENTER>FILL<SWITCH> and <ENTER><JUSTIFY>.
AFTER:
When in the course of human events it becomes
necessary for one people to dissolve the political
bands which have connected them.
Pack my box with five dozen liquor jugs.
Now you type <ENTER>LM:10/RM:68/PI:-5<SWITCH> and <ENTER>5<JUSTIFY>.
AFTER:
When in the course of human events it becomes necessary for one
people to dissolve the political bands which have
connected them.
Pack my box with five dozen liquor jugs.
.EL .S1
Beware: The lines above all begin with spaces, therefore if you try
to justify them again SED will treat them as four separate
paragraphs. Delete the leading spaces from the lines which are not
the beginning of a paragraph before justifying it a second time.
.PG .LIT
***************************************************************************
.EL .NOFILL
<EXECUTE> {EXECUT
.FILL .S1 .LIT
Starting nominal: 0
Parameter: number (iterations)
No Parameter: closes execute buffer, if open
does nominal iterations, if buffer closed
Cursor move: spaces (change of lines is illegal)
Token: opens execute buffer
.EL
.INDEX ^^EXECUTE EXECUTE COMMAND SEQUENCE {EXECUT #\\
.INDEX ^^COMMAND EXECUTE COMMAND SEQUENCE {EXECUT #\\
This command allows you to store a sequence of commands in a command buffer,
and later execute that sequence as many times as desired. A command buffer
can be made to look like an editor command (where pressing a button on the
terminal will cause the buffer to execute once). It is easy to do things
like removing the first 16 spaces from every line of the file, or causing
each occurrence of JU_&$ to be at the start of a new line.
<EXECUTE> is a complicated command. This section explains the basics of how
to use one <EXECUTE> buffer. The next two sections deal with advanced
<EXECUTE> features.
The procedure for using the <EXECUTE> command is as follows: first open the
execute buffer by typing <ENTER><EXECUTE>. From now until the buffer is closed
everything that you type will have its effect on the file and the screen,
and also be saved in the execute buffer. Do one iteration of the sequence.
Close the execute buffer by typing only <EXECUTE>.
When you want to execute that command sequence, figure out how many times
you want to do it (a large number, if you want as many as possible; it will
stop on an error like "Search failure"). For 10 iterations type
.S1 .LIT
<ENTER>10<EXECUTE>
.EL .S1
The command sequence in the execute buffer will be issued 10 times, just as
if you had typed the commands yourself. Except for one thing: the display
screen will not change until the execute is done. This saves time. However,
every 8 times the entire buffer has been executed, the terminal will beep so
you can tell it is still working. So there should be 2 beeps from the command
above.
When the execute is done the screen will be re-displayed to show what
it looks like after the last iteration.
An <EXECUTE> can be aborted early by typing RUBOUT. The command will stop
immediately and return control to you. The file will have been changed by the
commands which have been issued before you aborted the <EXECUTE>.
Some helpful suggestions: set up your nominals before you open the command
buffer. For example, you might type
.S1 .LIT
<ENTER><EXECUTE>
<ENTER>XYZ<SEARCH-FORWARD>
ABC
<EXECUTE>
.EL .S1
which will set up to change XYZ's to ABC's. Instead type
.S1 .LIT
<ENTER>XYZ<SEARCH-FORWARD>
<CURSOR-UP>
<ENTER><EXECUTE>
<SEARCH-FORWARD>
ABC
<EXECUTE>
.EL .S1
This does the same thing, although it requires more typing. But the benefit
is that the work of setting up the search key is not done every time the
command is iterated. The time savings from this gambit can be enormous for a
complicated sequence or a lot of iterations (and you'll be just twiddling
your thumbs and listening to beeps while all this happens). There is another
advantage: the nominals can be changed, resulting in a similar but different
command sequence at minimum setup cost. For example, suppose the lower
sequence above had been typed. Now change all XYZ's to ABC's:
.S1 .LIT
<ENTER>0<PERCENT>
<ENTER>10000<EXECUTE>
.EL .S1
Fine. But it so happens that all xyz's want to be changed to ABC's, too. Just
type:
.S1 .LIT
<PERCENT>
<ENTER>xyz<SEARCH-FORWARD>
<CURSOR-UP>
<EXECUTE>
.EL
This will search for all occurrences (or the first 10000, whichever comes
first) of xyz and change them to ABC. It iterates 10000 times because that
is the nominal set up by the previous <EXECUTE>. The <CURSOR-UP> is so the
first match will be found by the first iteration; if the cursor were pointing
at it, it would be skipped.
Note: if xyz were wanted to be changed to abc (instead of ABC), you would need
to reload the execute buffer, since the ABC is not a command nominal. For the
record, an absolutely arbitrary "change all occurrences of ?? to ??" command
can be set up by the following:
.PG .LIT
<ENTER><STRING><SEARCH-FORWARD>
<ENTER><length of above string><DELETE-SPACES>
<ENTER><STRING1><PUT>
<ENTER><EXECUTE>
<SEARCH-FORWARD>
<DELETE-SPACES>
<PUT>
<EXECUTE>
.EL
This sets up <EXECUTE> to change all occurrences of STRING to STRING1 (except
the first two, which you have done yourself by hand. The first sets up the
command parameters and the second saves the commands in the execute buffer).
To change all STRING2's to STRING3's, type
.S1 .LIT
<ENTER><STRING2><SEARCH-FORWARD>
<ENTER><length of above string><DELETE-SPACES>
<ENTER><STRING3><PUT>
.EL .S1
and then execute; the rest has already been set up.
Another use of the <EXECUTE> command is to do something to the same part of
every line in the file. For example, suppose you had a batch output file and
you wanted to pretty it up by removing all the time tags - the first 16
characters of every line. Move to the start of the first line of the file and
type
.S1 .LIT
<ENTER>16<DELETE-SPACES>
<ENTER>1<ROLL-FORWARD-LINES>
.EL .S1
That sets up the nominal parameter values. Now load the execute buffer:
.S1 .LIT
<ENTER><EXECUTE>
<DELETE-SPACES>
<CURSOR-DOWN>
<ROLL-FORWARD-LINES>
<EXECUTE>
.EL .S1
Then execute the sequence by typing <ENTER>10000<EXECUTE>. Note that the
<ROLL> is necessary because without it, when the execute got to last line on
the screen the <DOWN> would position the cursor back at the first line on the
screen. Note also that the <DOWN> is not really needed for this particular
sequence since the <ROLL> will kick the cursor down a line (since the cursor
is at the top of the screen). But if you were working with lines in the
middle of the screen, or deleting characters from the middle of a line, you
would want the <CURSOR-DOWN>.
Use the /SHOW switch to tell SED whether to display execution as it happens
(/SHOW) or just to re-display the screen when the entire execute is complete
(/NOSHOW, the default).
.PG
.C;MORE INFORMATION ABOUT <EXECUTE>
.S1 .LIT
Sections:
SETTING UP A BUFFER
WRITING A BUFFER
TWO WAYS OF READING THE CONTENTS OF A BUFFER
FINDING OUT WHAT THE NAMES OF THE BUFFERS ARE
KILLING A BUFFER
.EL
.IFNOT TOPS20
.LIT
SETTING UP EXECUTE BUFFERS IN SWITCH.INI OR SED.INI
.EL
.ELSE TOPS20
.LIT
SETTING UP EXECUTE BUFFERS IN SED.INIT
.EL
.ENDIF TOPS20
.LIT
EXAMPLES OF EXECUTE SEQUENCES
.EL
There are a number of execute buffers (normally 32), so a variety of
different command sequences can be set up and iterated. Each buffer can be
given a name up to five characters long. Names can be defined or deleted as
desired. Buffers can be written either by opening them and allowing
commands to be stored, as described in the previous section, or directly
using the WRITE-BUFFER format described below. One buffer has a default
name: null (it's the buffer that was described in the basic <EXECUTE>
section above).
There are also conditional execute commands (such as IF and DO WHILE), and
a way to connect an execute buffer to a command key or sequence, so that
pressing the key will execute the buffer once (thus creating your own
meta-command). Also, execute buffer definitions can be stored in a disk
file and read in when needed. All these things will be taken up in the next
section.
The <EXECUTE> command can be in one of the following formats:
.S1 .LIT
<ENTER>S<name><EXECUTE> set up buffer with given name
<ENTER>K<name><EXECUTE> kill buffer with given name
<ENTER>N<EXECUTE> get list of buffer names
<ENTER>W<sequence><EXECUTE> write sequence into active buffer
<ENTER>R<name><EXECUTE> read name & contents of buffer
<ENTER>R<EXECUTE> read active buffer
<ENTER>L<name><EXECUTE> list contents of buffer
<ENTER>L<EXECUTE> list active buffer
<ENTER><EXECUTE> open active buffer
<EXECUTE> close or execute active buffer
<ENTER><number><EXECUTE> execute active buffer n times
.EL
Note that the WRITE format, and READ and LIST with no argument, work with
the active execute buffer, meaning the one that was most recently set up
(using the S format). Originally the active buffer is the null buffer.
All characters typed in these formats are meaningful. So don't put in
any spaces unless you really want them.
.PG .LIT
***************************************************************************
SETTING UP A BUFFER
.EL
.INDEX ^^EXECUTE EXECUTE: SETTING UP A BUFFER {EXECUT #\\
The general procedure for working with an execute buffer is to set it
up, write it, then execute it.
To define and write a new buffer: choose a name, say "FOO", and set up
the buffer by typing
.S1 .LIT
<ENTER>SFOO<EXECUTE>
.EL
Names can be up to five characters long and can consist of any
printing characters.
The S format is also used to set to a buffer which has already been
created. A buffer must be set before it is written (the W format), read
(R), listed (L), or executed. The S format will look for a buffer of the
given name and make it and its contents active if it exists; if no buffer
by that name exists, it will create one (which is empty). To set to the
null buffer type <ENTER>S<EXECUTE>.
.S1 .LIT
***************************************************************************
WRITING A BUFFER
.EL
.INDEX ^^EXECUTE EXECUTE: WRITING A BUFFER {EXECUT #\\
After creating the buffer named FOO you can write a command sequence into
it in the usual way, by typing <ENTER><EXECUTE> to open the buffer, the
desired sequence of commands, and <EXECUTE> to close it. Note that setting
up the buffer name with the S format does NOT open the buffer.
The execute buffer can also be written directly, without having the commands
take effect on the screen (using the W WRITE format). The WRITE format works
with the active buffer. For example, following the above set command with
.S1 .LIT
<ENTER>W$THING^SF$5^DS$HACK^PT<EXECUTE>
.EL .S1
will load the buffer names FOO with a sequence of commands which searches for
an occurrence of "THING", deletes it, and inserts "HACK". The buffer is
loaded with the following command sequence:
.S1 .LIT
<ENTER>THING<SEARCH-FORWARD>
<ENTER>5<DELETE-SPACES>
<ENTER>HACK<PUT>
.EL
Writing into the execute buffer requires a language to describe the editor
commands, since typing them directly will cause them to take effect.
Appendix#E is a list of the editor commands and the sequences that <EXECUTE>
uses to describe them. In general they are "_^" followed by two or three
letters which describe the command. <ENTER-PARAMETER> is a dollar sign.
For another example, to load the active buffer with a sequence to put an "X"
at the cursor position, move to the position underneath, and roll the screen
one line, type
.S1 .LIT
<ENTER>WX^CL^CD$1^RFL<EXECUTE>
.EL .S1
This sets up the buffer with the following command sequences:
.S1 .LIT
X
<CURSOR-LEFT>
<CURSOR-DOWN>
<ENTER>1<ROLL-FORWARD-LINES>
.EL
Of course, the execute will run faster if the parameters are set up in
advance, not in the sequence itself.
The execute buffer does not have to be opened in order to write it using
the W format, nor does it have to be closed afterward. Those two formats
apply only when you want to save commands as they are issued, as described
in the previous section.
After the buffer has been written it can be executed in the normal
fashion, by typing <ENTER><number><EXECUTE>.
If you want to do the same series of commands several times in a row, you
can give a repeat count. For example, _^4(_^CR) is the same as
_^CR_^CR_^CR_^CR. Any series of commands can be repeated up to 4000 times.
To put a real up-arrow in a buffer type "_^_^" (two up-arrows). To put in a
real dollar sign type "_^$". Within a repeat, a real close parenthesis must
be given as "_^)". When using the /X: switch, use "_^/" to mean a real
slash (to distinguish it from the start of the next switch).
You can use cursor movement to read a buffer definition from the file you
are editing, so a useful way of experimenting with execute buffers is to
put the buffer definition in your file, preceded by the "W" for the WRITE
format. Position to the "W", move the cursor to the end of the line,
and type <EXECUTE>. If the command sequence is not quite right, change it
in the file and use cursor movement to pick it up again.
.S1 .LIT
***************************************************************************
TWO WAYS OF READING THE CONTENTS OF A BUFFER
.EL
.INDEX ^^EXECUTE EXECUTE: READING A BUFFER {EXECUT #\\
There are two ways you can see what is in an execute buffer:
LISTing and READing. LIST presents the buffer in a manner
compatible with the WRITE format; READ presents it in a style that
can be <PUT> into execute files or
.IFNOT TOPS20
SWITCH.INI or SED.INI.
.ELSE TOPS20
SED.INIT.
.ENDIF TOPS20
To see what is in a buffer, type <ENTER>L<name><EXECUTE>. The active
buffer can be read by typing <ENTER>L<EXECUTE>. The null buffer must be made
active before it is read (using the S format). Typing <ENTER>L<EXECUTE>
will cause a line like
.S1 .LIT
W$THING^SF$5^DS$HACK^PT
.EL .S1
to be put into the parameter buffer and displayed on the screen. This line
can be treated just like a parameter that you have typed: you can edit it
and type <EXECUTE> again to load the edited version into the buffer, or
use <PUT> to insert it into the file (where it can be copied or modified
at will, and later reloaded by typing <ENTER><END-LINE>(EXECUTE>.
Similarly, you can type <ENTER>R<name><EXECUTE> (or <ENTER>R<EXECUTE> to
read the active buffer). This format gives you a little more data:
READing the FOO buffer will cause the line
.S1 .LIT
/X:FOO:$THING^SF$5^DS$HACK^PT
.EL .S1
to be put into the parameter buffer and displayed on the screen. Again, you
can treat this line like a parameter that you have typed, for example
to store the command string in
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
or some other file by doing a READ and a <PUT>. For more about these switch
files, see the section on reading execute files from disk.
.S1 .I0
Note: The output of the L or R formats may be too long to fit in
the parameter buffer. In that case the output will be truncated to
fit in the parameter buffer and SED will beep to tell you it's
taken liberties with your buffer.
.S1 .LIT
***************************************************************************
FINDING OUT WHAT THE NAMES OF THE BUFFERS ARE
.EL
.INDEX ^^EXECUTE EXECUTE: FINDING THE BUFFER NAMES {EXECUT #\\
Typing <ENTER>N<EXECUTE> will fill the bottom line of the screen with the
names of the buffers.
.S1 .LIT
***************************************************************************
KILLING A BUFFER
.EL
.INDEX ^^EXECUTE EXECUTE: KILLING A BUFFER {EXECUT #\\
If all the execute buffers are set up and you try to SET another one, SED
will inform you that all the buffers are in use and suggest you kill one of
them. You do that using the K format: to kill the buffer FOO type
.S1 .LIT
<ENTER>KFOO<EXECUTE>
.EL .S1
That name will go away and the buffer will be cleared, and it will be
available to be set to something new.
The null buffer can be killed by typing <ENTER>K<EXECUTE>.
.PG .LIT
***************************************************************************
.EL
.IFNOT TOPS20
.LIT
SETTING UP EXECUTE BUFFERS IN SWITCH.INI OR SED.INI
.EL
.ELSE TOPS20
.LIT
SETTING UP EXECUTE BUFFERS IN SED.INIT
.EL
.ENDIF TOPS20
.IFNOT TOPS20
.INDEX ^^EXECUTE EXECUTE: BUFFERS IN SWITCH.INI {EXECUT #\\
.ELSE TOPS20
.INDEX ^^EXECUTE EXECUTE: BUFFERS IN SED.INIT {EXECUT #\\
.ENDIF TOPS20
Frequently-used command sequences can be put in
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
so they will be set up automatically when SED is run. An example of an
execute switch is:
.S1 .LIT
/X:NAME:$gig^PT^RT
.EL .S1
which sets up execute buffer "NAME" with the command sequence to do a <PUT> and
a carriage return.
You can copy an execute buffer into
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
(or any file) by typing <ENTER>R<name><EXECUTE> and then <PUT>.
.S1 .LIT
***************************************************************************
EXAMPLES OF EXECUTE SEQUENCES
.EL
.S1 .LIT
^8($5^DS^CD)^RFL
.EL .S1
(where the <ROLL-LINES> parameter is set to 8). This deletes 5 characters from
the same column of each of 8 lines, then rolls the screen 8 lines.
.S1 .LIT
$^ECJ^EC?^EC?^EC?^ECM^SF^DL
.EL .S1
This will delete all lines which are exactly three characters long. It
looks for lines which have exactly three characters between a linefeed (the
end of the previous line) and a carriage return.
.S1 .LIT
^SF^LN^PK^RT^FL^PT^RT^FL
.EL .S1
(where some search key has been set up, the <PICK> parameter is 1, and the
alternate file is the end of the current file). This copies all lines which
contain a match of the search key to the end of the file. It searches for a
match, gets to the start of the line, <PICK>s the line, sets to the
alternate file (which is the end of the current file), <PUT>s the line, and
<SET-FILE>s back. Note that there will never be a search failure. The
alternate file could be a separate file, but then the two files will have
to be read and written on disk for each <SET-FILE>, which will slow the
sequence down considerably. The set of matching lines can easily be
transferred from the end of the file to any other file when the execute has
completed.
.PG
.C;ADVANCED EXECUTE BUFFER FEATURES
.S1 .LIT
Sections:
WRITING AND EXECUTING IN ONE COMMAND
SETTING INITIAL PARAMETERS FOR AN EXECUTE
CONNECTING EXECUTE BUFFERS TO TERMINAL KEYS
READING EXECUTE BUFFERS FROM DISK
CONDITIONAL EXECUTE CONSTRUCTS
THE EXECUTE COUNTER
.EL
The story of <EXECUTE> is not over yet.
.S2
.C;WRITING AND EXECUTING IN ONE COMMAND
.INDEX ^^EXECUTE EXECUTE: X (WRITE/EXECUTE) FORMAT {EXECUT #\\
You can write a sequence of commands into the active buffer and immediately
execute the buffer a given number of times by using the X format:
.S1 .LIT
<ENTER>Xn:<sequence><EXECUTE> write buffer & execute n times
.EL .S1
The X format is useful for doing a one-shot series of commands.
.S1 .LIT
***************************************************************************
.EL
.C;SETTING INITIAL PARAMETERS FOR AN EXECUTE
.INDEX ^^EXECUTE EXECUTE: SETTING INITIAL PARMS {EXECUT #\\
The START-UP construct allows you to define parameters and switches
at the beginning of an execute. The format of START-UP is:
.S1 .LIT
START-UP . . . . . . . . . . . ^ST( ... )
.EL
The contents of the START-UP block is executed only once, at the start of
the execute. The parameters of commands in the START-UP block are set up
but the commands themselves are not executed. Thus the START-UP block is
used to initialize command parameters. Switches can also be set in the
block using the <SWITCH> command.
For example the construct _^ST($1_^IS$gig_^SF) will set the <INSERT-SPACES>
parameter and the search key, but will not perform either command.
Initializing can speed up an execute buffer a lot. For example
.S1 .LIT
^ST($THING^SF$2^DS)^SFGIG^DS
acts the same as
$THING^SFGIG$2^DS
.EL .S1
but is much more efficient since the parameters are set up once in the
first case and once per iteration in the second. Parsing those parameters
takes time.
The START-UP block can also be used to set switches which will be in effect
during the execute:
.S1 .LIT
^ST($NC/T:8/NIM^SW) or
^ST($NC+T:8+NIM^SW)
.EL .S1
initializes for case-independent searches, tab length of 8, and no INSERT
mode during the execute. The second form above must be used in
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
or an indirect (.XCT) execute file, since a "/" marks the start of a new
buffer definition or switch.
.S1 .LIT
***************************************************************************
.EL
.C;CONNECTING EXECUTE BUFFERS TO TERMINAL KEYS
.INDEX ^^EXECUTE EXECUTE: CONNECTING TO KEYS {EXECUT #\\
You can set up any command sequence to call one iteration of an execute
buffer. A command sequence is a string of characters which begins with a
control character. The action of the commands in the buffer will be displayed
on the screen as it happens. You can use this capability, in effect, to
define new commands to SED.
There is one more EXECUTE format:
.S1 .LIT
<ENTER>B<EXECUTE> connect active buffer to button
.EL .S1
After you type this format SED will ask you to press any key followed by the
letter "G" (or "g"). SED will take the sequence output by that key as the
invocation of the currently active buffer. Making some other buffer active
has no effect on this relationship. Changing the contents of the buffer also
has no effect, but remember that if you change the buffer something new will
happen when you press the key.
After you have typed <ENTER>B<EXECUTE> you can type any command sequence you
want. The sequence can be a control character, a special key, or any string
of characters which begins with a control character. If you type
.S1 .LIT
<ENTER>B<EXECUTE>
<ESCAPE>XG
.EL .S1
then the buffer would be executed once when you typed <ESCAPE>X (there is a
way to make <ENTER>n<ESCAPE>X cause the buffer to be executed n times; see
the section about the execute counter, below).
The sequence of characters can even be one which already invokes an editor
command, so you can replace commands you don't use with new ones of your
own design.
You can undefine a key by making its buffer active and typing
<ENTER>B<EXECUTE>G (ie, re-defining its key to nothing).
.IFNOT TOPS20
The key definition can be set up with the buffer in SWITCH.INI or SED.INI.
The expanded format of the SWITCH.INI line is:
.ELSE TOPS20
The key definition can be set up with the buffer in SED.INIT. The expanded
format of the SED.INIT line is:
.ENDIF TOPS20
.S1 .LIT
/X:NAME,^[X:$gig^PT^RT
.EL .S1
which sets up a buffer named NAME with <ENTER>gig<PUT><RETURN>, and assigns
it to the sequence <ESCAPE>X. ("_^[" == CONTROL-[ == <ESCAPE>. All the
stuff between the comma and the colon is taken as the key definition. The
key definition can be omitted.
Note: RUBOUT (or DELETE) can be part of the command sequence. If you read
the execute buffer the RUBOUT will appear as "_^?".
.S1 .LIT
***************************************************************************
.EL
.C;READING EXECUTE BUFFERS FROM DISK
.INDEX ^^EXECUTE EXECUTE: READING FROM DISK {EXECUT #\\
There is yet another <EXECUTE> format:
.S1 .LIT
<ENTER>@FILE.EXT<EXECUTE> read from the given file
.EL .S1
This causes the buffer definitions and switches in FILE.EXT to be set up.
The previous definitions are erased. Thus the disk file can contain the
definition of an entire special user environment.
The disk file looks like a list of switches. SED will skip characters until
it finds a "/", so the first several lines can be used for comments. But be
sure there's not a "/" in the comments.
The default extension of this file is "XCT".
You can set up and write execute buffers, and then when they are working
properly you can put them in this file by Setting each one up, doing a
READ, and then a <PUT>.
Here is an example of an execute file. It sets up four special keys on
the VT52 terminal to do useful things for an assembly language programmer:
.S1 .LIT
MACRO PROGRAMMER'S EXECUTES:
BUTTON 4: TAB TO 4TH STOP AND ";"
BUTTON 5: <CR> AND TAB
BUTTON 6: ":" AND TAB
BUTTON 8: PUSHJ<TAB>P,
BUTTON 9: POPJ<TAB>P,
/X:B4CM,^[?t:^3(^FCL32(^TA));
/X:B5RT,^[?u:^RT^TA
/X:B6RC,^[?v::^TA
/X:B8PU,^[?x:PUSHJ^TAP,
/X:B9PO,^[?y:POPJ^TAP,
.EL .PG .LIT
***************************************************************************
.EL
.C;CONDITIONAL EXECUTE CONSTRUCTS
.INDEX ^^EXECUTE EXECUTE: CONDITIONAL CONSTRUCTS {EXECUT #\\
In addition to the editor commands, the following constructs are defined:
.S1 .LIT
ITERATED-DO . . . . . . . . . . ^n ( ... )
DO-WHILE . . . . . . . . . . . ^DW condition ( ... )
DO-ON-SEARCH-FAILURE . . . . . ^DF ( ... )
IF-CHARACTER . . . . . . . . . ^IF condition ( ... )
IF-ROW . . . . . . . . . . . . ^FR comparison ( ... )
IF-COLUMN . . . . . . . . . . . ^FC comparison ( ... )
IF-COUNTER . . . . . . . . . . ^F. comparison ( ... )
EXIT-BLOCK . . . . . . . . . . ^XB
CONTINUE-BLOCK. . . . . . . . . ^CB
EXIT-BUFFER . . . . . . . . . . ^XX
.EL
The ITERATED-DO construct has already been described: it was called a
"repeat block" in a previous section. It causes the stuff following in
()'s to be repeated the given number of times (up to 4000).
DO-WHILE and IF-CHARACTER perform or skip the stuff in ()'s depending on
the nature of the character at the cursor position. DO-ON-SEARCH-FAILURE
performs the stuff in ()'s only if the preceding search fails.
IF-ROW and IF-COLUMN perform or skip the stuff in ()'s depending on the row
or column position of the cursor on the screen. IF-COUNTER does likewise
depending on the value of the Execute counter (use of the counter is
described below).
EXIT-BLOCK causes control to leave the interior of a conditional.
CONTINUE-BLOCK causes the rest of the conditional to be skipped. These two
behave the same inside an IF block; in a DO block EXIT stops iterating and
CONTINUE ends the current iteration. EXIT-BUFFER ends the current iteration
of the execute buffer.
Now for details. The DO-ON-SEARCH-FAILURE construct must immediately follow
an <SEARCH-FORWARD> or <SEARCH-BACKWARD> (_^SF or _^SB). The commands in
()'s will be skipped over unless the search fails. On search failure SED
will execute the commands in ()'s, then continue execution with the next
command after the _^DF block. Use the _^XX, _^CB, or _^XB constructs in the
_^DF block to control how execution proceeds (SED can be told to continue,
to stop the execute, or to stop the current iteration, depending on how
those constructs are used).
If a _^SF or _^SB is not followed by a _^DF and the search fails, an
"search-failure" error will result and execution will be terminated.
.S1 .LIT
EXAMPLES:
^20(^SF^DF(^GO^PT^XX)^LB^PK^RT)END
.EL .S1
<PICK>s up to 20 lines which contain the search key. If fewer than 20
matches are found, SED moves to the beginning of the file, does a <PUT>,
and terminates the execute. If 20 matches are found SED types "END"
wherever the cursor is.
.S1 .LIT
^20(^SF^DF(^GO^PT^XB)^LB^PK^RT)END
.EL .S1
The _^XX is changed to an _^XB. The action is the same as above, except
when the search fails, after the <PUT>, SED terminates the iterate block
(the _^20). "END" is typed out and execution ends.
.S1 .LIT
^20(^SF^DF(^GO^PT^CB)^LB^PK^RT)END
.EL .S1
Now it's _^CB. When the search fails, after the <PUT>, SED skips the rest
of the current iteration and starts the next one. It does another search,
which fails; these searches continue until all 20 iterations are done.
This is probably not what the writer of the buffer had in mind. But it
illustrates the difference between _^XB and _^CB.
DO-WHILE and IF-CHARACTER can check to see if the character
at the cursor is
.S1 .LIT
Some given character
^L ALPHABETIC (A-Z, a-z)
^U UPPER CASE (A-Z)
^N NUMERIC (0-9)
^A ALPHA-NUMERIC
^C ANY CHARACTER
^S SPACE OR TAB
^E END OF LINE OR BEYOND
.EL .S1
or .NOT. any of those things (by preceding them by "_^X"). "Any character" is
anything but space or TAB (same as "_^X_^S").
You can say "if character is a space, move cursor right" by "_^IF_^S(_^CR)".
"Do while character is not the letter 'A', cursor right" is "_^DW_^XA(_^CR)".
IF-ROW, IF-COLUMN and IF-COUNTER check if the current row or column or the
Execute counter is:
.S1 .LIT
En EQUAL TO n
Gn GREATER THAN n
Ln LESS THAN n
Nn NOT EQUAL TO n
.EL .S1
For example, you can say "If column is less than 64, cursor right"
by "_^FCL64(_^CR)", and "If it's not row 12, cursor down" by "_^FRN12(_^CD)".
These constructs can be nested to a depth of about 10, and intermingled as
much as desired.
.S1 .LIT
Examples:
^DW (^CR) while character at cursor is " ", cursor right
^DW^X5(^CR) while character is not a "5" move right
a.k.a. DO-UNTIL character is "5" ...
^DW^X^N(^CR) until character is numeric move right
^FCE64(^RT) if at column 64 do a RETURN
^FRG10(^CH) if beyond row 10 go HOME
^IF^E(^RT) if at or beyond end of line do a RETURN
^IF^X^E(^CR) if not end of line move right
^1000(^PT) PUT 1000 times
.EL
EXIT- and CONTINUE-BLOCK are usually used inside a conditional, along with an
IF statement. For example
.S1 .LIT
^DW^X^E(^CR^FCG40(^XB))
.EL .S1
will stop DOing when the end of the line is reached or when the column is
greater than 40.
.S1 .LIT
^1(^IF^S(^DS^PT^XB)^TB)
.EL .S1
is an IF-THEN-ELSE statement. If the character is a space or TAB SED deletes
it, does a <PUT>, and skips the rest of the conditional; else it TABs over.
.PG .LIT
***************************************************************************
.EL
.C;THE EXECUTE COUNTER
.INDEX ^^EXECUTE EXECUTE: THE EXECUTE COUNTER {EXECUT #\\
You can keep a count of something and use that count as a command parameter
or the number of iterations of a DO. The counter constructs are:
.S1 .LIT
CLEAR-COUNTER . . . . . . . . . ^C=
INCREMENT-COUNTER . . . . . . . ^C+
DECREMENT-COUNTER . . . . . . . ^C-
USE-COUNTER . . . . . . . . . . ^CT
ITERATE-COUNTER . . . . . . . . ^C.( ... )
IF-COUNTER . . . . . . . . . . ^F.( ... )
SET-COUNTER . . . . . . . . . . ^SC
.EL
The counter can be cleared, incremented, and decremented anywhere in the
execute buffer.
You can use the counter as a parameter: if the counter contains 12, then
"$_^CT_^DS" will delete 12 spaces. Or "$_^CT_^PT" will <PUT> the characters
"12" into the file. You can use the counter with characters in the
parameter; "$123_^CT45_^PT" and "$AB_^CTCD_^CTEF_^PT" are both legal.
To loop <counter> times on a set of commands, type "_^C.(#...#)". For example
if the counter contains 5, "_^C.(_^PT)" will do 5 <PUT>s.
The value of the counter can be used for conditional branching:
"_^F.E10(_^PT)" will do a <PUT> only if the counter contains 10.
SET-COUNTER (_^SC) is used to set the counter to a specific value.
"$12^SC" sets the counter to 12.
SET-COUNTER can also be used to set the execute counter to the (numeric)
parameter the user typed. It is intended for execute buffers which are
assigned to keys on the terminal. If _^SC appears at the beginning of
the buffer and the user types <ENTER>17<KEY>, then 17 will be stored in
the counter. If the user does not type a parameter, the same value as
last time will be used (the default is 1). Non-numeric parameters cause
an error.
.I0
For example the buffer
.S1 .LIT
^SC^C-^ND$^ECL^C.(^SF)^RT
.EL .S1
sets up a command to put the user at the start of some page of the file
(where pages are marked by formfeeds - CONTROL-L's). Suppose the user types
<ENTER>4<KEY>. The counter is set to 3, and CONTROL-L is searched for 3
times [the construct _$_^ECL_^C.(_^SF) is used rather than _^C.($_^ECL_^SF)
because it is faster - the parameter is only processed once). After the last
search a carriage return is done. NO-DISPLAY (_^ND - see next section) is
set, so the action is displayed only after the buffer is done executing.
Without _^ND all the search targets would appear on the screen one by one.
.S3 .LIT
***************************************************************************
.EL
.C;DEFERRING THE DISPLAY OF A BUFFER ATTACHED TO A TERMINAL KEY
.INDEX ^^EXECUTE EXECUTE: DEFERRING THE DISPLAY {EXECUT #\\
.S1 .LIT
NO-DISPLAY . . . . . . . . . . ^ND
.EL
Normally, the action taken by an execute buffer which is connected to a
terminal key is displayed step by step as the buffer is processed. If the
NO-DISPLAY construct appears in a buffer, the screen will not be updated
until the buffer has completed. Then the entire screen will be re-written.
This feature is useful if the buffer will display a lot of parts of the file
but you only care about where it ends up.
A buffer with _^ND in it acts the same as a buffer invoked by
<ENTER>n<EXECUTE> with /NOSHOW set.
.S3 .LIT
***************************************************************************
.EL
.C;OUTPUTTING IMAGE CHARACTERS
.INDEX ^^EXECUTE EXECUTE: OUTPUTTING IMAGE CHARS. {EXECUT #\\
.S1 .LIT
OUTPUT . . . . . . . . . . . . ^OU( ... )
.EL
The OUTPUT construct is used to output an image string of characters. SED
will send to the terminal the ASCII argument to _^OU. This construct can
be used to cause things to happen to the terminal that SED does not know
about (for example, changing colors on a graphics terminal).
NOTE: The text is output to the terminal screen. It is not put in the file.
Control characters are denoted by "_^K". <ESCAPE> is "_$". "_^_$", "_^_^",
and "_^)" denote "_$", "_^", and ")".
.S1 .I0
EXAMPLE: _^OU(_$H_$J(Running_^)) outputs "<ESCAPE>H<ESCAPE>J(Running)".
.PG
As an overall example, here is an execute buffer which lines up every comment
line at the 4th tab position, or if the comment starts after the 4th tab, it
puts one space between the instruction and the comment. Also, comments which
start before column 16 are assumed to be subroutine headers, and are left
alone.
Note that if the search key were "'" the buffer would work with BASIC
programs; if it were "!" it would work with FORTRAN, etc.
One comment is handled by every iteration of this buffer.
.S1 .LIT
^ST($;^SF$1^IS$IM^SW)
^SF
^FCG16(
^CL^C=
^DW^S(^CL^C+)
^CR
$^CT^DS
^4(^FCL32(^TA))
^CL
^IF^X^S(^CR )
)
^RT
.EL .S1
The buffer first sets the search key to be ";", the parameter for
<DELETE-SPACES> to be 1, and INSERT MODE to be in effect.
SED finds a match. If the match is not beyond the 16th column, SED does
a carriage return (one comment per line!) and stops the iteration.
Else SED moves to the left of the comment symbol. It counts up all spaces
to the left of the comment, and then deletes them. Then SED puts in tabs
until the comment starts at column 32.
Then SED looks at the character to the left of the comment symbol. If it is
not a space or tab (meaning that the comment started beyond column 32, thus
no tabs were inserted in the loop above) SED separates the comment from the
instruction with a space. SED then does a <RETURN> to end the iteration.
Other sample execute buffers can be found in SED.XCT and other .XCT files.
Incidentally, the multi-line format above can be used in
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
or .XCT
files. The indentation characters must be TABs. When SED is reading a
buffer definition all control characters are ignored, so CARRIAGE RETURNs,
LINEFEEDs, and TABs can be put in wherever you want.
.PG
.C;APPENDICES
.S1
.IFNOT TOPS20
.C;Appendix A. DEFAULT FILESPECS AND SFDS
.INDEX ^^SFDS DEFAULT FILESPECS AND SFDS Appendix #\\
.INDEX ^^FILESPECS DEFAULT FILESPECS AND SFDS Appendix #\\
When SFD's are in effect SED's default filespecs behave a little
differently than when they are not. Without SFD's, if you do not give a
PPN the default is the number you logged in under. Thus if you are
[100,150],
.S1 .LIT
<ENTER>FILE.EXT[200,25]<SET-FILE>
<ENTER>NEWFIL.MAC<SET-FILE>
.EL .S1
will find NEWFIL.MAC on [100,150].
However, with SFD's in effect the default is the path which SED is following
at the moment. The above commands would find NEWFIL.MAC on [200,25]. Here's
how the defaults work when you have SFD's:
.S1 .LIT
<ENTER>FILE.EXT[200,25]<SET-FILE> gets things started
<ENTER>FOO.MAC[,,THING]<SET-FILE> looks on [200,25,THING]
<ENTER>FOO.FOR<SET-FILE> on [200,25,THING]
<ENTER>FOO.BAR[,]<SET-FILE> on [200,25]
<ENTER>NEWFIL.MAC[100,150]<SET-FILE> only way to get [100,150]
.EL
.PG
.ELSE TOPS20
.C;Appendix A. DEFAULT FILESPECS
.INDEX ^^FILESPECS DEFAULT FILESPECS Appendix #\\
Files are always looked for in your connected directory unless you
give an explicit device and/or directory name.
.S1 .LIT
***************************************************************************
.EL
.ENDIF TOPS20
.C;Appendix B. LOOKING THROUGH A LIST OF FILES
.INDEX ^^INDIRECT LOOKING THROUGH A LIST OF FILES Appendix #\\
.INDEX ^^LIST LOOKING THROUGH A LIST OF FILES Appendix #\\
You can create a file containing a list of files and then look at them
one by one, moving one file forward or backward along the list. The
easiest way to create this list of files is using the DIRECT program
(or command). For example, type
.IFNOT TOPS20
.S1 .LIT
.DIR/F LIST=*.MAC
.EL .S1
.ELSE TOPS20
.S1 .LIT
@DIR *.MAC,
@@OUTPUT (TO FILE) LIST
.EL .S1
.ENDIF TOPS20
This creates LIST.DIR with a list of the MACRO files in your area.
.IF TOPS20
It will probably need a little editing since the full filespec (in particular
the file name) must appear on each line.
.ENDIF TOPS20
Any method which sets up a list of files, one to a line, will do.
If you then run SED and say <ENTER>@LIST.DIR<SET-FILE> (or run with
.IFNOT TOPS20
".R#SED;@LIST.DIR")
.ELSE TOPS20
"@SED#@LIST.DIR")
.ENDIF TOPS20
(the default extension is "DIR") you will be
looking at the first file on the list. Then a lone <SET-FILE> command will
get the next file on the list, and <ENTER><SET-FILE> will get the previous
one. When you reach the beginning or end of the list SED will toggle between
the last two files it found.
You can return to normal by typing <ENTER>filespecs<SET-FILE>.
.S1 .I0
NOTE: The indirect file must reside in
.IFNOT TOPS20
the top level of your logged-in directory.
.ELSE TOPS20
your connected or login directory.
.ENDIF TOPS20
If the indirect file is somewhere else, include the device and/or
directory in the filespec.
The indirect file (such as LIST.DIR above) can contain comments (which start
with "!" or ";"). Spaces and tabs are ignored, so they can be put in anywhere.
Entirely blank lines can appear before the first filespec, but not between
specs (use a lone comment symbol).
Here is an example of an indirect <SET-FILE> file:
.S1 .LIT
;LIST OF EXPERIMENTAL FILES
EXP1.MAC !FILE TO CHANGE ALL THE WORLD
EXP2.MAC ;FILE TO CHANGE ALL THE REST
EXP3A.MAC
!
;THE FOLLOWING ARE NOT VERY USEFUL:
EXP8.MAC
EXP15.MAC
.EL
.PG
.C;Appendix C. USING MARGINS AND SCREEN LIMITS
.INDEX ^^MARGINS USING MARGINS AND SCREEN LIMITS Appendix #\\
.s1
This appendix discusses the action of four switches: /LENGTH:, /WIDTH:,
/LMAR:, and /RMAR:.
/LENGTH: and /WIDTH: set up the dimensions of the terminal display. You can
use them to limit the amount of output. For example, for a lot of searching
on a slow terminal, set /LENGTH:10 and only a few lines around the search
match will be displayed. Also, SED can be adjusted for terminals which change
widths, like the VT100, using the /WIDTH: switch.
/LMAR: and /RMAR: control how typed-in characters are entered on the screen
and in the file. If you set, say, /LMAR:8, then a <CARRIAGE-RETURN> or
<BEGIN-LINE> command will move the cursor, on the screen and in the file, to
the eighth column. The line in the file will begin with eight spaces. /LMAR:
is initially 1.
You can move into that indented region using <CURSOR-LEFT> or
<DELETE-CHARACTER>.
/RMAR: sets up the right margin. If you are typing in text and exceed the
right margin then the word you are typing will be moved to the next line. SED
will put in the carriage returns for you. /RMAR: starts as the width of the
entire screen.
/RMAR: can be disabled by setting it larger than the screen width (like
/RMAR:1000).
.PG
.C;Appendix D. INCREMENTAL FILE SAVES AND JOURNALING
.INDEX ^^JOURNAL INCREMENTAL SAVES AND JOURNALING Appendix #\\
.INDEX ^^SAVES INCREMENTAL SAVES AND JOURNALING Appendix #\\
.s1
There are a number of ways to recover editing sessions in SED. You can save
the file you are editing automatically by setting the /ISAVE: and /SAVE:
switches. Say you set /ISAVE:30/SAVE:150. Then after 30 commands or 150
characters of type-in, whichever comes first, the file will be saved. Then
both counters are reset.
/ISAVE: and /SAVE: both start off at zero, meaning no incremental saves
(which is what these saves are called). You can tune SED's saving
characteristics to your editing style.
Of course, you can also save the file by simply exiting SED and rerunning it,
since SED will put you back where you were in the file. This scheme is good
if you want to save only at specific times (and can remember that it needs
to be done). There's also the <SAVE-FILE> command, which saves the file
without exiting SED.
One thing to beware: incremental saves and the <SAVE-FILE> command do not
clean up the file. SED puts a lot of nulls and trailing spaces into the file;
they are eliminated when the file is saved by an <EXIT> or <SET-FILE> command.
But the other saves are designed to take as little time as possible, so the
nulls and spaces are left in. However, some programs will break if there are
nulls in the file. So end your editing sessions by typing <EXIT>, not
<SAVE-FILE><ABORT>.
Incremental saves can take a long time for large files. So there is another
recovery mechanism: journaling. The journal is the file SEDJRN.TMP, which
holds the stream of commands which you typed during your editing session. If
the computer crashes you can recover your session by telling SED to read its
commands from the journal and apply them to the unchanged file (it is
unchanged because the system crashed before you exited). A few commands at
the end of the session will be lost, but you will get most of your work back.
.S1 .LM5 .I0
ALERT: Command parameters settings are not saved in the journal. Suppose
you edit a file, do a search (say), set to another file, and type
<SEARCH-FORWARD>. When you set to the other file a new journal was started,
and it does not contain the search key. If you try to recover using that
journal SED will give a "No search key" error. The same applies to any
parameters which were changed before the current file was edited. This
problem does not occur very often, but it is a known deficiency of
journaling.
However, the journal is an ASCII file and can be edited, so if you can
remember what the missing parameters are you can insert them in the
journal and make it work.
.LM0
Start journaling by typing the /JOURN switch (in the <SWITCH> command, in the
monitor-level RUN command, or as part of a <SET-FILE>. The journal will be
started immediately, and SED will remember what file you are editing and
where you are in it. The journal is restarted every time the file is saved
(on a <SET-FILE> or incremental save).
You don't have to worry about the journal unless the computer crashes. When
the system comes up, type
.IFNOT TOPS20
".R#SED;/RECOV"
.ELSE TOPS20
"@SED/RECOV"
.ENDIF TOPS20
at monitor level. SED will take its
commands from the journal, update the file, and tell you when it's done.
If you want to see your own commands being performed at super-speed, type
.IFNOT TOPS20
".R#SED;/RECOV/SHOW"
.ELSE TOPS20
"@SED/RECOV/SHOW"
.ENDIF TOPS20
Watching your two-hour session whiz by in a couple of minutes is very
humbling, and has the same entertainment value as watching two shirts fight
in a clothes dryer.
To abort a journal recovery, type RUBOUT anytime during the recovery.
If you recover by typing
.IFNOT TOPS20
".R#SED;/REC/JOURN" (or ".R#SED;/REC/SHOW/JOURN")
.ELSE TOPS20
"@SED/REC/JOURN" (or "@SED/REC/SHOW/JOURN")
.ENDIF TOPS20
then the journal will be appended to after the recovery is finished. This
also happens if /JOURN is in
.IFNOT TOPS20
SWITCH.INI or SED.INI.
.ELSE TOPS20
SED.INIT.
.ENDIF TOPS20
Note: if, after recovery, you type
<ENTER>JOURN<SWITCH> then the journal file will be STARTED, not appended to.
When in doubt save the file (by exiting, for example), and start with a fresh
journal.
If you put the /JOURN switch in your
.IFNOT TOPS20
SWITCH.INI or SED.INI
.ELSE TOPS20
SED.INIT
.ENDIF TOPS20
file then journaling will be done automatically. You can turn it off for an
editing session by typing <ENTER>NOJOURN<SWITCH> (as you might expect).
.IF TOPS20
.PG
.C;Appendix E. VARIATIONS ON THE EXEC-LEVEL RUN COMMAND
.INDEX ^^RUN THE EXEC-LEVEL RUN COMMAND Appendix #\\
The "classical" SED run command is
.S1 .LIT
@SED FILE.EXT/SWITCH/SWITCH or
@SED FILE.EXT=/SWITCH/SWITCH
.EL .S1
where the second format creates the file if it not found; the first gives
an error. If the /CREATE switch is set then both formats create the file.
SED also understands the TOPS-20 Exec's EDIT and CREATE commands. The
general format for them is
.S1 .LIT
@EDIT/SWITCH/SWITCH INFILE.EXT OUTFILE.EXT
@CREATE/SWITCH/SWITCH INFILE.EXT
.EL .S1
These commands act the same as
.S1 .LIT
@SED INFILE.EXT/SWITCH/SWITCH/OUT:OUTFILE.EXT
@SED INFILE.EXT=/SWITCH/SWITCH
Furthermore,
@SED/SWITCH INFILE.EXT/SWITCH/SWITCH
.EL .S1
is also legal.
Note: some versions of the Exec do not allow SED switches to be included in
the command line.
.ENDIF TOPS20
.PG
.PS 60 68
.C;Appendix F. TABLE OF EXECUTE COMMAND NAMES
.INDEX ^^EXECUTE TABLE OF EXECUTE COMMAND NAMES Appendix #\\
.INDEX ^^NAMES TABLE OF EXECUTE COMMAND NAMES Appendix #\\
.S1 .LIT
COMMAND NAME OTHER COMMANDS NAME
------- ---- ----- ---------- ----
ENTER-PARAMETER . . $ REWRITE . . . . . . . . ^RW
SAVE-FILE . . . . . . . ^SV
CURSOR CURSOR-UP . . . . . ^CU SET-FILE . . . . . . . ^FL
CURSOR-DOWN . . . . ^CD SWITCH . . . . . . . . ^SW
CURSOR-LEFT . . . ^CL TAB-SET . . . . . . . . ^TS
CURSOR-RIGHT . . . ^CR WINDOW . . . . . . . . ^WI
.EL
.IFNOT TOPS20
.LIT
CURSOR-HOME . . . ^CH
.EL
.ELSE TOPS20
.LIT
CURSOR-HOME . . . ^CH PUSH . . . . . . . . . ^PU
.EL
.ENDIF TOPS20
.LIT
CARRIAGE-RETURN . . ^RT
TAB . . . . . . . . ^TB OTHER CONSTRUCTS FORMAT
BACKTAB . . . . . . ^BT ----- ---------- ------
UP-TAB . . . . . . ^UT START-UP . . . . . . . ^ST( ... )
DOWN-TAB . . . . . ^DT IMAGE-OUTPUT . . . . . ^OU( ... )
BEGIN-LINE . . . . ^LB NO-DISPLAY . . . . . . ^ND
END-LINE . . . . . ^LE
LINE . . . . . . . ^LN ITERATED-DO . . . . . . ^n( ... )
DO-WHILE . . . . . . . ^DWc( ... )
ROLLS ROLL-BACK-PAGES . . ^RBP DO-ON-SEARCH-FAILURE . ^DF( ... )
ROLL-FORWARD-PAGES. ^RFP IF-CHARACTER . . . . . ^IFc( ... )
ROLL-BACK-LINES . . ^RBL IF-ROW . . . . . . . . ^FRc( ... )
ROLL-FORWARD-LINES. ^RFL IF-COLUMN . . . . . . . ^FCc( ... )
PERCENT-GOTO . . . ^GO EXIT-BLOCK . . . . . . ^XB
SLIDE-LEFT. . . . . ^SL CONTINUE-BLOCK. . . . . ^CB
SLIDE-RIGHT . . . . ^SR EXIT-BUFFER . . . . . . ^XX
SEARCH SEARCH-FORWARD . . ^SF CLEAR-COUNTER . . . . . ^C=
SEARCH-BACKWARD . . ^SB INCREMENT-COUNTER . . . ^C+
SUBSTITUTE . . . . ^SU DECREMENT-COUNTER . . . ^C-
SET-COUNTER . . . . . . ^SC
TEXT INSERT-SPACES . . . ^IS USE-COUNTER . . . . . . ^CT
DELETE-SPACES . . . ^DS ITERATE-COUNTER . . . . ^C.( ... )
INSERT-LINES . . . ^IL IF-COUNTER . . . . . . ^F.c( ... )
DELETE-LINES . . . ^DL
PICK . . . . . . . ^PK For DO-WHILE or IF-CHARACTER:
PUT . . . . . . . ^PT
INSERT-MODE . . . . ^IN SOME GIVEN CHARACTER
DELETE-CHARACTER. . ^DC ^L ALPHABETIC (A-Z, a-z)
CLEAR-LINE . . . . ^LF ^U UPPER CASE (A-Z)
ERASE-LINE . . . . ^EL ^N NUMERIC (0-9)
ERASE-WORD . . . . ^EW ^A ALPHA-NUMERIC
^S SPACE OR TAB
OTHERS ABORT . . . . . . . ^AB ^E END OF LINE OR BEYOND
CASE . . . . . . . ^CS
ENTER-CTRL-CHAR . . ^EC ^X .NOT. ONE OF THE ABOVE
EXECUTE . . . . . . ^EX
EXIT . . . . . . . ^XT For IF-ROW, -COLUMN, -COUNTER:
HELP . . . . . . . ^HL
JUSTIFY . . . . . . ^JU En EQUAL TO n
MARK . . . . . . . ^MK Gn GREATER THAN n
REAL-TAB . . . . . ^TA Ln LESS THAN n
RECALL . . . . . . ^RC Nn NOT EQUAL TO n
.EL
.PG
.PS 56 68
.C;EXAMPLE I
.C;VT52 TERMINAL KEYBOARD LAYOUT
.INDEX ^^VT52 VT52 TERMINAL KEYBOARD LAYOUT Example #\\
.INDEX ^^TERMINAL VT52 TERMINAL KEYBOARD LAYOUT Example #\\
.INDEX ^^KEYBOARD VT52 TERMINAL KEYBOARD LAYOUT Example #\\
The VT52 has a numeric keypad to the right of the normal keyboard. Four
cursor movement keys are at the right of that keypad and there are three
colored buttons above the pad. The leftmost (blue) button is the <ENTER>
command, the rightmost one (BLACK) is <HOME>, and the center one (RED) is
<RECALL>. The number keys in the pad invoke SED commands. Do not use the
red ESC (SEL) key at the upper left of the keyboard.
.S2 .LIT
VT52 EDITOR KEYBOARD
.____.____.____.____.____.____.____.____.____.____.____.____.____.____.
! ! ! ! ! ! ! ! ! ! ! ! ! !ERAS!
! ! ! @ ! # ! $ ! % ! ^ ! & ! * ! ( ! ) ! _ ! + !SUBS!CHAR!
! 1 ! 2 ! 3 ! 4 ! 5 ! 6 ! 7 ! 8 ! 9 ! 0 ! - ! = ! ~` !back!
!____!____!____!____!____!____!____!____!____!____!____!____!____!____!_.
!xon !ROLL!SRCH!SRCH!ROLL!ROLL!BACK! !ENTR!PERC! ! !ERAS!RE- !
! !BK L!BACK!FWD !FW L!FW P!TAB !TAB !CCH !GOTO! ] ! | !LINE!SET !
! q ! w ! e ! r ! t ! y ! u ! i ! o ! p ! [ ! \ !lnfd! del!
!____!____!____!____!____!____!____!____!____!____!____!____!____!____!_.
!ROLL!xoff!INS !DEL ! !CUR !ERAS!INS !DEL ! ! ! ! !
!BK P! !LINE!LINE!PUT !LEFT!LINE!SPAC!SPAC! : ! " ! } ! RETURN !
! a ! s ! d ! f ! g ! h ! j ! k ! l ! ; ! , ! { ! !
.___!____!____!____!____!____!____!____!____!____!____!____!____!_________!
! ! ! ! ! !SET ! !RE- ! ! ! ! ! ! !
!shift!EXIT!EXCT!ABRT!PICK!FILE!SWCH!TURN! < ! > ! ? ! shift !rept!copy!
! ! z ! x ! c ! v ! b ! n ! m ! , ! . ! / ! ! ! !
!_____!____!____!____!____!____!____!____!____!____!____!_______!____!____!
NUMERIC PAD AT RIGHT .____.____.____.____.
OF NORMAL KEYBOARD: ! !RE- ! !CUR !
!ENTR!CALL!HOME! UP !
!blue!red !blck! /\ !
!____!____!____!____!
! !UP- ! !CUR !
!MARK!TAB !CASE!DOWN!
! 7 ! 8 ! 9 ! \/ !
!____!____!____!____!
!BEG-!RE- !END-!CUR !
!LINE!WRIT!LINE!RGHT!
! 4 ! 5 ! 6 ! -> !
!____!____!____!____!
!WIN-!DOWN! !CUR !
!DOW !TAB !HELP!LEFT!
In addition: ! 1 ! 2 ! 3 ! <- !
!____!____!____!____!
ESCAPE-L is <SLIDE-LEFT> ! REAL !ERAS!INS !
ESCAPE-T is <SLIDE-RIGHT> ! TAB !WORD!MODE!
ESCAPE-S is <TAB-SET> ! 0 ! . !entr!
ESCAPE-J is <JUSTIFY> !_________!____!____!
.EL
.PG
.PS 56 68
.C;EXAMPLE II
.C;VT100 TERMINAL KEYBOARD LAYOUT
.INDEX ^^VT100 VT100 TERMINAL KEYBOARD LAYOUT Example #\\
.INDEX ^^TERMINAL VT100 TERMINAL KEYBOARD LAYOUT Example #\\
.INDEX ^^KEYBOARD VT100 TERMINAL KEYBOARD LAYOUT Example #\\
.S2 .LIT
VT100 KEYBOARD LAYOUT
._____. .____.____.____.____.
!set- ! ! ! ! ! !
! up ! ! /\ ! \/ ! <- ! -> !
! ! ! UP !DOWN!LEFT!RGHT!
!___._!__.____.____.____.____.____.____.____.____.____!____!____!____!____!
! ! ! ! ! ! ! ! ! ! ! ! ! ! !ERAS!
!esc! ! ! @ ! # ! $ ! % ! ^ ! & ! * ! ( ! ) ! _ ! + !SUBS!CHAR!
! ! 1 ! 2 ! 3 ! 4 ! 5 ! 6 ! 7 ! 8 ! 9 ! 0 ! - ! = ! ~` !back!
!___!____!____!____!____!____!____!____!____!____!____!____!____!____!____!_.
! !xon !ROLL!SRCH!SRCH!ROLL!ROLL!BACK! !ENTR!PERC! ! ! !RE- !
! TAB ! !BK L!BACK!FWRD!FW L!FW P! TAB!TAB !CCH !GOTO! { ! } ! !SET !
! ! q ! w ! e ! r ! t ! y ! u ! i ! o ! p ! [ ! ] ! ! del!
!_____!____!____!____!____!____!____!____!____!____!____!____!____! !____!
! !ROLL!xoff!INS !DEL ! !CUR !ERAS!INS !DEL ! ! ! ! !
! caps !FW P! !LINE!LINE!PUT !LEFT!LINE!SPAC!SPAC! : ! " ! RETURN! | !
! lock ! a ! s ! d ! f ! g ! h ! j ! k ! l ! ; ! , ! ! \ !
!_______!____!____!____!____!____!____!____!____!____!____!____!_______!____!
! ! ! ! ! !SET-! !RE- ! ! ! ! !ERAS!
! shift !EXIT!EXCT!ABRT!PICK!FILE!SWCH!TURN! < ! > ! ? ! shift !LINE!
! ! z ! x ! c ! v ! b ! n ! m ! , ! . ! / ! ! lf !
!_________!____!____!____!____!____!____!____!____!____!____!_________!____!
NUMERIC PAD AT RIGHT .____.____.____.____.
OF NORMAL KEYBOARD: ! !RE- ! !SAVE!
!ENTR!CALL!HOME!FILE!
! pf1! pf2! pf3! pf4!
!____!____!____!____!
! !UP- ! !SLID!
!MARK!TAB !CASE!LEFT!
! 7 ! 8 ! 9 ! - !
!____!____!____!____!
!BEG-!RE- !END-!SLID!
!LINE!WRIT!LINE!RGHT!
! 4 ! 5 ! 6 ! , !
!____!____!____!____!
!WIN-!DOWN! !INS !
!DOW !TAB !HELP!MODE!
! 1 ! 2 ! 3 ! e !
!____!____!____! n !
! !ERAS! t !
!REAL-TAB !WORD! e !
! 0 ! . ! r !
!_________!____!____!
In addition: ESCAPE-S is <TAB-SET>
ESCAPE-J is <JUSTIFY>
.EL
.PG
.C;INDEX OF EDITOR COMMANDS, BY COMMAND
.PS 56 100
.S1 .NOFILL .I0
PAGE COMMAND NAME EXPLANATION
.S1
13 {INSSPS INSERT-SPACES INSERT SPACES TO FILE
23 {SETFIL SET-FILE SET UP A FILE FOR EDITING
25 {ABORT ABORT EXIT AND FORGET CHANGES
10 {INSLIN INSERT-LINES INSERT BLANK LINES IN FILE
14 {SRCBAK SEARCH-BACKWARD SEARCH FROM CURSOR TO START OF FILE
10 {DELLIN DELETE-LINES DELETE LINES FROM FILE
19 {PUT PUT PUT TEXT INTO FILE
6 {CURLFT CURSOR-LEFT MOVE CURSOR LEFT
7 {TAB TAB MOVE TO NEXT TAB STOP
12 {CLRLIN CLEAR-LINE CLEAR LINE BELOW CURSOR
12 {ERSLIN ERASE-LINE ERASE FROM CURSOR TO END OF LINE
.IFNOT OLDTAB
9 {SLDLFT SLIDE-LEFT SLIDE VIEWING WINDOW LEFT
9 {SLDRGT SLIDE-RIGHT SLIDE VIEWING WINDOW RIGHT
.ELSE OLDTAB
17 {PICK PICK PICK TEXT FROM FILE
9 {SLDLFT SLIDE-LEFT SLIDE VIEWING WINDOW LEFT
.ENDIF OLDTAB
6 _^M CARRIAGE-RETURN GOOD OL' ASCII CARRIAGE RETURN
.IFNOT OLDTAB
28 {SWITCH SWITCH SET SWITCHES OR QUERY STATUS
.ELSE OLDTAB
7 {BAKTAB BACKTAB MOVE TO PREVIOUS TAB STOP
.ENDIF OLDTAB
27 {ENTCCH ENTER-CTRL-CHAR MAKE NEXT CHARACTER A CONTROL CHAR.
9 {PERCGO PERCENT-GOTO GO TO SOME PERCENT THROUGH THE FILE
8 {ROLBKP ROLL-BACK-PAGES MOVE VIEWING WINDOW BACK PAGES
14 {SRCFWD SEARCH-FORWARD SEARCH FROM CURSOR TO END
13 {DELSPS DELETE-SPACES DELETE CHARACTERS FROM FILE
8 {ROLFWL ROLL-FORWARD-LINES MOVE VIEWING WINDOW FORWARD LINES
.IFNOT OLDTAB
7 {BAKTAB BACKTAB MOVE TO PREVIOUS TAB STOP
17 {PICK PICK PICK TEXT FROM FILE
.ELSE OLDTAB
9 {SLDRGT SLIDE-RIGHT SLIDE VIEWING WINDOW RIGHT
28 {SWITCH SWITCH SET SWITCHES OR QUERY STATUS
.ENDIF OLDTAB
8 {ROLBKL ROLL-BACKWARD-LINES MOVE VIEWING WINDOW BACK LINES
38 {EXECUT EXECUTE PERFORM SEQUENCE OF EDITOR COMMANDS
8 {ROLFWP ROLL-FORWARD-PAGES MOVE VIEWING WINDOW FORWARD PAGES
25 {EXIT EXIT EXIT AND SAVE CHANGES
3 {ENTER ENTER PASS A PARAMETER TO A COMMAND
25 {RESET RESET RESET SCREEN OR PARAMETER
34 {RECALL RECALL RECALL PREVIOUS PARAMETER
22 {INSMOD INSERT-MODE CHARACTER INSERT/REPLACE TOGGLE
11 {DELCHR DELETE-CHARACTER ERASE CHARACTER TO LEFT OF CURSOR
27 {REALTB REAL-TAB TYPE A REAL TAB IN THE FILE
20 {MARK MARK MARK POSITION FOR PICK OR DELETE
35 {CASE CASE CHANGE CASE OF LETTER AT CURSOR
35 {WINDOW WINDOW SET OR CLEAR SPLIT-SCREEN WINDOWING
6 {UPTAB UP-TAB MOVE THE CURSOR UP 6 LINES
6 {DWNTAB DOWN-TAB MOVE THE CURSOR DOWN 6 LINES
27 {REWRIT RE-WRITE REWRITE THE SCREEN
22 {SAVER SAVE-FILE SAVE THE CURRENT FILE
34 {HELPER HELP GIVE ON-LINE HELP
7 {BEGLIN BEGIN-LINE MOVE THE THE BEGINNING OF THE LINE
7 {ENDLIN END-LINE MOVE THE THE END OF THE LINE
11 {ERASWD ERASE-WORD ERASE THE WORD AT THE CURSOR
.IF TOPS20
24 {PUSHER PUSH PUSH TO EXEC CONTEXT
.ENDIF TOPS20
15 {SUBSTI SUBSTITUTE SEARCH AND SUBSTITUTE
26 {TBSET TAB-SET SET AND CLEAR TABS
36 {JUSTI JUSTIFY JUSTIFY TEXT
21 {MARKER MARKER MANIPULATE MARKER IN FILE
.PS 56 68 .PG .FILL
.C;ARTICULATED INDEX OF EDITOR COMMANDS
The COMMAND column refers to the control character which invokes the desired
command. UP, DOWN, LEFT, RIGHT, and HOME are the five cursor movement keys,
usually denoted by arrows on the key caps RUB is the RUBOUT key, sometimes
yclept DEL, for DELETE. ESC is the ESCAPE or ALTMODE key. "PAD" means that
the key is on a keypad which is separate from the normal typewriter keyboard.
.NOAUTOP .NOFIL
.S1 .LIT
KEY EXPLANATION COMMAND PAGE
.EL
.TITLE KEY EXPLANATION COMMAND PAGE
.PRINT INDEX