Digital PDFs

XX-4E589-32

May 1985

24 pages

Original

1.5MB

Document:	TSX-Plus Users Group Newsletter
Order Number:	XX-4E589-32
Revision:	0
Pages:	24
Original Filename:	http://bitsavers.org/pdf/dec/pdp11/tsxPlus/users_group/TSX-Plus_Users_Group_Newsletter_May85.pdf

OCR Text

TSX-PLUS USER'S GROUP NEWSLETTER

TSX-PLUS USERS' GROUP (TSXPUG) NEWSLETTER
Newsletter
TSXPUG
c/o Jack Peterson
Horizon Data Systems
1901 Wildflower Terr.
Richmond, VA 23233
IMPORTANT NOTICE
The TSX-Plus Users' Group Newsletter is a
quarterly publication of the
TSX-Plus
Users' Group (TSXPUG), a volunteer organization dedicated to the exchange of information among users of TSX-Plus. TSXPUG is
sanctioned by, but is otherwise independent of and receives no financial support
from S & H Computer Systems, Inc., developers of TSX-Plus. All expenses are paid
from membership fees.
Jack and Glenn, whose addresses are above,
run TSXPUG.
Jack writes and edits the
newsletter; Glenn handles administration,
collecting money, maintaining the membership list, and mailing the newsletter. If
you have a contribution for or a question
about the newsletter, write to Jack; direct all other correspondence to Glenn.

Administration
TSXPUG
c/o Glenn Burnett
Zia Corporation
P.O. Box 351
Morris Plains, NJ

07950

Reader contributions to the newsletter are
solicited. Brief articles may be submitted in hardcopy form, but we prefer lengthy articles on magnetic media.
We can
read RX01, RX02, and 1600 bpi magnetic
tape; in a pinch, RX50 and RL02; in a real
pinch, RL01. Our preferred format is .RNO
or .DOC. We will return media to you if
you ask us to do so. All submissions
become the property of the TSX-Plus Users'
Group.
Thanks for your cooperation and
your contributions.
All material contained herein is presumed
to be correct, but TSXPUG, its staff, and
newsletter contributors assume no responsibility for any errors that may appear.
Readers use any and all published materi~
at their own risk.
COPYRIGHT NOTICE

Membership fees are $15.00 (US) per year
in the USA, Canada, and Mexico, $20.00
(US) per year elsewhere. Memberships must
be paid in US dollars with an instrument
payable at a bank in the United States.
All memberships terminate on December 31
each calendar year. Partial year memberships are NOT pro-rated. Current members
will receive a renewal notice in January
for the coming year; payment must be returned before February 25th to assure
receipt of the first newsletter of the
year. New members may join at any time by
completing the application on the inside
back cover, and including the appropriate
membership fee with the application.

This newsletter is published by the TSXPlus Users' Group for distribution to and
use of its members only.
TSXPUG members
are hereby granted permission to copy,
modify, and use any information published
herein within their own computer installations.
Otherwise, duplication of this
newsletter or any part thereof without the
express written permission of Horizon Data
Systems is prohibited.
UNLESS NOTED TO THE CONTRARY, THE ENTIRE
CONTENTS OF THIS NEWSLETTER ARE COPYRIGHT
(C) 1985 BY HORIZON DATA SYSTEMS. ALL
RIGHTS RESERVED.

TSX-PLUS USERS' GROUP (TSXPUG) NEWSLETTER
May 1985

CONSULTANT HOTLINE

Tom Smith
Data Processing Services, Inc.
P.O. Box 40128
San Antonio TX 78240
(512) 692-0953

Since starting this column in November, we
have published the names, addresses, and
telephone
numbers
of firms providing
TSX-Plus consulting services as they have
made themselves known to us. This information is now spread out over several
newsletter issues, and a few more names
have arrived. What follows is a complete
list, in zip code order, of all TSX-Plus
consultants known to us.
we
neither
endorse nor recommend any of these firms
(except ourselves, of course), but publish
this
information
for the benefit of
readers who may need special help with
their TSX-Plus systems.

David C. Ketchum
STW Software, Inc.
7860 W. 16th Avenue
Suite 115
Denver CO 80215

(303) 232-9164
Nick A. Bourgeois, Jr.
NAB Software Services, Inc.
3236 Candlelight N.E.
P.O. Box 20009
Albuquerque NM 87154
(505) 298-2346

Please save this list for future reference. We now plan to publish a complete
list once each year, probably in the May
issue. If you provide TSX-Plus consulting
services and are not on this list, send us
the necessary information.
We will see
that you are included in the next list.

Milton D. Campbell
Talisman Systems
Drawer CP-255
Manhattan Beach CA 90266
(213) 318-2206

Ned Rhodes
Systems Group
1684 E. Gude Drive
Rockville MD 20850
(301) 340-2773

.::;na.l Farley
Cheshire Engineering Corporation
650 Sierra Madre Villa
Suite 201
Pasadena CA 91107

Soft~are

(818) 351-5493
Jack Peterson
Horizon Data Systems
1901 Wildflower Terrace
Richmond VA 23233
(804) 740-9244

Jim Crapuchettes
Omnex Corporation
2483 Old Middlefield Way
Mountain View CA 94043
(415) 966-8400

Lorin S. Jordan
ABCCI
7400 Metro Boulevard
Suite 109
Edina MN 55435
(612) 893-1354

Gregory Merrick
IRIS Registration, Inc.
75 The Donway West
Suite 1014
Don Mills, Ontario M3C 2E9
Canada
(416) 446-1066

Peter Heinicke
Precision Computer Methods, Inc.
No. 3 2S779 Winchester Circle
Warrenville IL 60555
(312) 393-6478

David Clements
AccountAbility Computer Services
P.O. Box 6043, Stn 'A'
Calgary, Alberta T2H 2L3
Canada
(403) 236-0740

-i-

TSX-Plus Users' Group Newsletter
May 1985
you already use. Simply place the SETSYS
commands at the beginning of one of your
other automatically
execute~
detached
files.

A GOOD USE FOR A DETACHED JOB
Detached jobs are one of the
useful
features of TSX-Plus, but many people do
not know how to use them. At Horizon, we
have developed the concept of a system
configuration detached job we call SETSYS
(Set System).

SENDING MESSAGES TO INACTIVE TERMINALS
One of the items mentioned in our Christmas wish list was the ability to send
messages to inactive terminals.
We have
occasionally found this useful, particularly after we have brought the system
back up after a brief outage. Inspired by
Milton Campbell's creative use of CL in
the
last
newsletter, and ideas from
others, Horizon has developed a program,
ONLINE, reproduced at the end of this
newsletter, which sends a user-defined
message to all inactive terminals. The
comments on the program are self-explanatory and should allow you to customize the
program as needed for your own installation.
You must have at least one free CL
unit on your system in order to use
ONLINE.

With the advent of Version 5.1, TSX-Plus
has provided users a high level of flexibility in altering the characteristics of
time sharing and communication (CL) lines.
Parameters such as line speed and phone
operation previously required modifications to TSGEN.MAC, reassembly, relinking,
and rebooting. But now, these parameters
and others can be changed by commands.
In our installation (and those of our
customers), we define a detached job file
called SY:SETSYS.TSX and specify, via the
DETACH macro in TSGEN.MAC, that it be
executed automatically when the system
st:lrts. TSGEN.MAC is built to reflect our
normal configuration, and SETSYS is used
to alter that configuration on an as
needed basis.
For example, if we later
wish to use a spare port (always genned as
a time-sharing line) as a CL, we place the
appropriate SET CLn LINE=xx command in
SETSYS.
Or, if we later wish to use that
port as a dial-in line, we SET the line as
PHONE and AUTOBAUD in SETSYS. In this
way, we eliminate many SYSGENs and/or
versions of TSX-Plus, and are confident
that the system will always come up with
our latest configuration because SETSYS is
executed whenever TSX-Plus is restarted.

To use ONLINE, first enter the program as
file ONLINE.MAC. Then assemble and link
it
using MAC ONLINE and LINK ONLINE
commands.
Finally, copy the resultant
ONLINE.SAV to your system disk. ONLINE is
run by the following commands:
R ONLINE
<Put Message To Be Sent Here>
In our installation, we put this command
(the message reads *** SYSTEM IS OPERATIONAL ••• YOU
MAY NOW LOG ON ***) in
SY:SETSYS.TSX
(see
previous
column).
Whenever
TSX-Plus
is
restarted, the
message is displayed on all
inactive
terminals.

SETSYS may contain any commands
that
affect
the system on a global basis
because it runs with operator privilege.
For example, handler SET commands can be
issued to ensure that handlers have a
known
initial
configuration.
SETSYS
cannot be used to make global logical name
assignments, however, because such assignments are local to each line.

We are indebted to S & H for this idea.

To use SETSYS, you must specify at least
one detached line in the TBLDEF macro of
TSGEN.MAC, but, because SETSYS runs for
only a short period, you need not otherwise increase the number of detached lines

Many TSX-Plus installations, especially
those with dial-in lines, use account
codes and passwords to regulate access to
system features and files, and to perform
automatic mounts
and
logical
device

CUSTOMIZING STARTUP COMMAND FILES

-2-

TSX-Plus Users' Group Newsletter
May 1985
1ssignments for users. In TSGEN.MAC, each
line definition may include a CMDFIL macro
naming a startup command file executed
when the line becomes active. Often, all
(or most) lines specify a common command
file that ends with R/LOCK LOGON to start
the logon process.
After the user is
identified, a command file unique to that
account, containing ACCESS, MOUNT, ASSIGN,
and other commands, is executed.

1$:
DBLK:
BUF:

.TTINR
BCC
START
.GTLIN flBUF,IBUF,TTINP
•LOOKUP flBUF , 110 ,IJDBLK
BCS
1$
fl4000,@fl44
BIS
RO
CLR
.EXIT
.RAD50 "DK LOGON COM"
41 •
•BLKW

.=BUF
Many users would like to put
ACCESS
commands in one command file which the
user cannot change (.TSX extension on the
system disk), then chain to a user-modifiable command file so each user can control
default
assignments,
mounts, and UCL
definitions.
Unfortunately,
once
a
startup command file has been started
TSX-Plus interprets all input (terminal or
command file) as part of the startup
process, so a user could defeat file
access
permissions
by putting ACCESS
commands in his command file.

.NLIST
.. ASCII
.END

LOGCHK first drains the terminal input
ring buffer, discarding all typed-ahead
input. This has two benefits.
First,
typed-ahead input is not interpreted as
part of the startup process, enhancing
security. Second, it forces the following
terminal-mode .GTLIN request to hang for
lack of input data, ending startup privilege. The .GTLIN request displays the
message
"Press
Return
to Continue:"
(abbreviated "Prompt:" in the
program
above due to column width limitations) •

startup

privilege (during which ACCESS
are allowed) ends whenever a
•GTLIN request (usually issued by KMON)
finds no input available.
If we could
arrange to have this event occur, we could
then execute a user-modifiable command
file during
startup
without
startup
privilege. The following program, LOGCHK,
does exactly that, and a little more.
We
present the program, explain how it works,
and how we use it.
Feel free to copy
and/or modify this program for your own
installation.
~ommands

.TITLE

After the user responds, type-ahead is
permitted, and LOGCHK determines whether
file DK:LOGON.COM exists.
If so, it is
executed, terminating the original startup
command
file;
otherwise,
control is
returned to the startup file permitting
default assignments, etc., to be made.
Neither commands in LOGON.COM nor those
remaining in the startup command file are
executed with startup privilege~

LOGCHK

To use LOGCHK, enter the program
as
LOGCHK.MAC, assemble and link it using MAC
LOGCHK and LINK LOGCHK commands, and copy
the resultant LOGCHK.SAV file to your
system disk •

HORIZON DATA SYSTEMS, RICHMOND VA
.ENABL
•MCALL
•MCALL
• ASECT

LC
.EXIT, .GTLIN
•LOOKUP, •TTINR

In our installation, all lines are directed (via the CMDFIL macro) to command file
SY:TSXGO.TSX containing assignments (ASS
CL2 LP) common to all users and terminated
by R/LOCK LOGON.
All lines have the
$QTSET flag set so startup files are not
displayed. Each logon account names a
unique command file that defines access
permissions as appropriate,
mounts
a
subdevice unique to that account, and

.=44
•WORD

10100

.=510
.WORD
1$:
2$:

~ASCIZ

BEX
<15><12>"Prompt:"<200)
START

2$-1$
"@LOGON"

.PSECT
START: :

-3-

TSX-Plus Users' Group Newsletter
May 1985
priority of IIO initiation (12.) is lowe~
than the default priority of an interrupt
.FORK (50.).
Apparently, the following
sequence of events occurs:

assigns it logical name DK:.
A R/SINGLE
LOGCHK is the last command in the file
(/SINGLE is needed to prevent LOGCHK from
hanging while purging the input stream).
Users are invited to place MOUNT, ASSIGN,
and other commands, as well as UCL definitions in their own LOGON.COM files, and to
make changes as desired, subject only to
privilege and access permissions.

1. The DY handler initiates IIO on the
emulated RX02, with the initiation routine
completing normally.
2. The system sets the .FORK priority to
FP$IOS (12.) and calls the DL handler to
initiate a new IIO request.

Several modifications to LOGCHK could be
useful. For example, .GTLIN could request
entry of a password to be compared against
a key stored in LOGCHK or in a file
accessible only to the system manager.
This places password protection in the
hands of the system manager because users,
who can change their logon passwords to
strings easily guessed
by
intruders,
cannot change this value. Another modification could interpret the .GTLIN response
as the name of a command file to execute.

3. The DL handler starts to initiate 1/0
on the emulated RL02.
4. Upon completion of
the
previously
initiated DY IIO, the RX02 signals an
interrupt. The DL handler is interrupted
and program control is passed to the DY
handler interrupt entry point, which does
a .FORK call with an implied priority of
FP$IOF (50.).
5. The CPU executes the rest of the DY
interrupt routine, since its priority Di~
FP$IOF (50.), which is higher than the
~
initiation priority of FP$IOS (12.).

TSX-PLUS .FORK PROBLEM
Jim Crapuchettes and John Rose of Omnex
Corporation have forwarded an analysis af
and a solution to a .FORK problem in
TSX-Plus that affects users of DSD 880
Winchester/floppy drives, and those who
write
handlers
that perform internal
queueing. Their article follows.

6. Apparently the DSD 880 interface hangs
because of a protocol problem (some, but
not all, of the steps involved in starting
a DL IIO have been done, so the DSD 880
thinks it is a DL now), and does not
recognize the DYCSR address.

A. DSD 880 PROBLEM
A DSD 880 contains a single controller
which services both a Winchester drive
(emulating RL02s) and a floppy disk drive
(emulating
an
RX02).
Under TSX-Plus
Version 5.1, simultaneous Winchester/floppy access on a DSD 880 can cause a system
crash: a kernel mode trap (KTP) from the
DY
handler
may occur upon attempted
accesses to the floppy Control and Status
Register (DYCSR).
The problem is most
apparent when using an LSI-11/73 processor, but it can also occur on other LSI-11
and PDP-11 systems.

7. After a few microseconds a bus timeout
occurs, causing a kernel mode trap.
C. SOLUTION
The solution is to change the TSX-Plus
.FORK priority for IIO initiation from 12
(decimal)
to 50 (decimal).
Find the
following line (in the non-user-modifiable
section of TSGEN.MAC:
FP$IOS = FP$FLG+12. ;I/O initiation
Change "12." to "50.", as follows:

B. ANALYSIS
FP$IOS = FP$FLG+50. ;I/O initiation
Under TSX-Plus, both the DL and DY handlers execute a .FORK in their interrupt
handling routines. However, the default

After making this change, regenerate your
TSX-Plus system.

-4-

TSX-Plus Users' Group Newsletter
May 1985
I. INTERNAL QUEUEING HANDLERS

that I identify, send it to me and I will
pass it on in the next newsletter.

This same problem might affect internally
queued handlers, such as the KERMIT KM
handler. Internally queued handlers often
assume that a .FORK in the interrupt code
will eliminate potential interlock problems during internal queue manipulation.
This is not necessarily the case under
Version 5.1 of TSX-Plus, because of the
difference between initiation and interrupt .FORK priorities.

DY Revisited
In our last adventure, a system manager
was having problems with floppy drives,
experiencing "Trap to 4"
errors.
I
suggested you lock PIP, DUP and DIR into
low memory to solve that problem.
You
will notice in a column above that Jim
Crapuchettes gave additional information
about the DSD 880 disk.
I intend to
unlock PIP, DUP, and DIR and try his
suggestion.
I will report the results in
the next issue.

Without proper caution, internally queued
handlers
could
experience
interlock
problems similar to the DSD 880 problem
described
above.
Programmers writing
handlers for Version 5.1 should acquaint
themselves with TSX-Plus =FORK priorities
(Chapter 4, TSX-Plus System
Manager's
Guide), and take them into account when
using .FORK requests.

FORTRAN Question
-A --._"----

-~---

The following was contributed by
Bob
Walraven. Care to have a crack at it?
One of our students has a FORTRAN program
that he ran under TSX-Plus V5.0.
The
program opens a couple of logical units
and computes for a while before trying to
open a third logical unit. At this point
the program aborted indicating insufficient memory for the new 1/0 buffer. That
made sense, but here is the interesting
part.
After thinking about the problem a
bit, he had a flash of brilliance and
ADDED another 2Kword array to his program.
When he ran it again, it completed successfully. Do you know why?

If you have further questions or difficulties concerning .FORK priority problems
under TSX-Plus V5.1, contact:
Jim Crapuchettes
Omnex Corporation
2483 Old Middlefield Way
Mountain View CA 94043
(415) 966-8400
AN IND HINT

The answer might appear a little later on •

If you use timeouts in IND .ASK, .ASKN, or
• ASKS directives, be sure to SET TT NOWAIT
before running IND or the timeout will be
ineffective (the system will hang waiting
for terminal input).

MT/MS Revisited

In the last issue, we mentioned some MT/MS
problems. Tom Shinal has reported the
following:

HINTS AND KINKS
PROBLEM: If a tape operation is aborted
using the MS Handler while under TSX-Plus,
the system hangs and a reboot is required.
An abort can be a Control-C or any other
anomoly (no, an anomoly is not a sea
urchin) •

Ned W. Rhodes
Software Systems Group
1684 East Gude Drive
Rockville, MD 20850
(301) 340-2773
The response to my first column was so
underwhelming that I decided to contribute
another.
As always, if you have
an
unusual problem or a workaround to one

HARDWARE: Cipher
TC-02 controller.
SOLUTION:

-5-

Cachetape

with

Emulex

Check the Emulex controller for

TSX-Plus Users' Group Newsletter
May 1985
The Dreaded TO State

the reV1Slon level of the PROMS.
We had
trouble with REV D. REV H is the current
revision and solves the problem.
Emulex
replaced them at no charge. I won't speak
for them, but if you have the same problem, give them a call.

To finish up, I would like to share a
problem concerning the terminal output
(TO) state.
The WHO
system
command
produces a snapshot of the current status
of each line, including its execution
state: HI means high priority, TI means
waiting for terminal input, and TO means
terminal
output
is pending, but the
terminal has sent an
XOFF
character
suspending further output operations to
it.

As Jack Peterson has pointed out before,
there are cases in which the mag tape
handler can get corrupted, usually when
aborts or other unexpected things happen.
Under RT-11, the solution is simple -unload and reload the handler.
Under
TSX-Plus we can't reload handlers .•• yet.

A problem occurs when a job in TO state is
killed either by the KILL keyboard command
or the user hanging up the phone without
logging
off.
Because the system is
waiting for an XON from the terminal, It
cannot output the logoff message so the
line is left active and in the TO state.

FORTRAN Solution
His original program required about 55
Kbytes, so TSX-Plus gave him 56K bytes
(even though MEM 64 had been issued).
When the program started, it did a .SETTOP
to get FORTRAN buffer space, recelvlng
about 1Kbyte, enough for two buffers, but
not three.
By adding the 2Kword array,
the job size passed the 56Kbyte boundary,
so TSX-Plus allocated a full 64Kbytes for
the job, ensuring adequate room for the
1/0 buffers.
The problem does not occur
with small jobs because TSX-Plus allocates
the 56Kbyte default.

This problem occurs frequently on our
somewhat noisy dial-in lines. A user may
be getting a directory display when a
burst of noise causes the modems to dro.
carrier and disconnect. If the line wa~
in the TO state at that time, the line
hangs in this state and is not logged off.
TSX-Plus drops Data Terminal Ready (DTR)
(as it should), but because the line is
not inactive, subsequent calls to the port
modem go unanswered, and a dial-in line
has been lost.

Moral of story:
If your FORTRAN job is
about 54 or 55Kbytes, you may want to make
it a little over 56 Kbytes in size.
Running SETSIZ to add additional space to
the original program will do it.

One workaround is to force DTR high on the
modem at all times permitting the modem to
answer incoming calls without intervention
by TSX-Plus. Then a user calling in can
enter a Control-Q (XON) to clear the TO
state, and use the line. But this causes
security problems because the new caller
inherits the old user's output buffer. Do
you really want everyone looking at your
resume?

A Small Problem
John Davies at the Naval Ship Research and
Development Center shared this "feature"
with me.
If you turn your terminals off
at night then on again in the morning,
TSX-Plus drops the first character of your
logon. I didn't believe him until I tried
it out.
The problem occurs only during
the first logon; for the rest of the day,
no problem. Have you seen this?

Version 5.1C was supposed to fix the
problem by waiting a while before automatically flushing the output buffer.
My
experience says this is not happening.
Has anyone else experienced the
sa~i
problem? I have also had the problem wi~.
hardwired (non-modem) lines.
Drop me a
note if you have had similar problems.

[Editor's note: Yes, we have. Anytime we
power up a terminal AFTER TSX-Plus has
started running, the carriage return typed
to activate the line gets through okay,
but the first character of the account
name is invariably discarded. Very, very
strange.]

-6-

TSX-Plus Users' Group Newsletter
May 1985
forward searches and a
using DBL's excellent
interface.

THE DIBOL/DBL WIZARD RETURNS
Due to circumstances beyond our control
(as they say), John Stewart's column was
noticeably missing from the last newsletter.
He is now back with a vengeance.
Even if you don't know a DIBOL from a
SNOBOL or a COBOL or a BASE-BOL (go
Atlanta!), you might find his column quite
interesting. Read on •••

Both routines find the key and display the
text line or record where found, both on
the terminal and in a file, along with the
name of the file. You can modify these
routines to
search
files
containing
non-printing characters; only the file
name is displayed. This modification is
quite easily done.

SYSTEM-WIDE SEARCHING
A system manager often needs to search
many files efficiently for a given keyword
or phrase. A change affecting the entire
system may require it, or a troublesome
glitch may be causing problems and it is
not where you thought it was.
Your need
for this will be especially acute if you
do not have a nice documentation package.

This column is dealing with global searches where the key is matched regardless of
where it appears in a record. There are
many methods to limit a search to certain
positions within a record or line. One
example is using RTSORT for positional
searching. It is fast. The sort key must
be stated, but it is really a dummy if
only one match is found. Use the IINCLUDE
selection key for your search. It is also
possible to modify the methods here to
test the distance from the keyword to the
beginning of the containing line and to
accept matches only if this distance is
within a defined range. Other than these
brief comments, this section will focus
exclusively on global searches.

Searching efficiently is the big problem
for minis doing global searches.
You
don't have all day to get the answer.
Your system has 500 source files and you
want to find a keyword in less than 5 or
iO minutes.
The DBL/DIBOL programmer has seen all
kinds of line- and
character-oriented
search routines that were fine if you
didn't

mind

taking

all

day

ru~ning

backward version
a.ssembly language

The TECO Global Search

program that slowed everybody down for the
whole time. Several hours is typical for
this type of global search routine.

At the system prompt, enter:

The routines provided in this issue do
global searches of several hundred sources
in 5-10 minutes on an idle system. The
DBL version shown below searched TSGEN.MAC
for the word "SPOOL" and output all lines
on which it was found on the terminal and
in a disk file in less than six seconds.
We can live with that speed.

• and % may be used as wildcards in the
FILENAM parameter.

SRCH KEY FILENAM

Enter the following program into file
SRCH.COM.
A is a real caret; $ is a real
dollar sign. Structuring and documentation are omitted to conserve space:
A>R/N TECO
6ETA$A$O,128ETA$_1A~A$

There are two solutions, each with its own
advantages:
1. A TECO routine that
quickly and get running.

you

EWSRCH.TMPA$I
Search for A1 in A2.
Results stored in SRCH.THP.
A$HTHPWHKEN A2A$

can key in

<HK:EN $;IER $G·27I $HIAHKG*
A

14+(OAQ)<I A$>HT13 THKHA
<:E A1A$;OAQ+.,AQ-2+.FBANAESA$

2. A DBL system that is more than twice as
fast as the TEeO routine. It uses DBL's
assembly-language search subroutine for

._1~.:FB;A$"LLF<'._1,AQ+.XAL

-7-

TSX-Plus Users' Group Newsletter
May 1985
.UAZJ.UZGAQZJG*QZ+15-.(I A$>I>
VQZ , ZPWQZ , ZKQAJ»
HKISearch for A1 in A2.
Results stored in SRCH.TMP.
A$HTEXA$"$

The Faster DBL Method

First, this method can be done only in
DBL, not DIBOL, because DIBOL requires
files
to
have pretty "straight" DDF
format, while DBL can be used to process
any file, including those with non-printing characters. DIBOL assembly language
routines could be used, but with much more
risk and difficulty.
More importantly,
DIBOL won't let you read binary files.
Assembly language routines are absolutely
necessary for speed and DBL's assembler
interface is well documented.

As each file is scanned its name appears
on the screen. If a match is found, the
containing line is displayed after the
file name. The same information is output
to SRCH.TMP for printing, etc. For its
size, this is a very powerful routine,
proving there will always be a need for
TECO in the commercial software area.

This program searches considerably faster
than TECO, although TECO runs rings around
most other high-level search routines.
VFSRCH is a DBL program that does forward
and backward searches, mostly in assembly
language, at 15K to 20K bytes per second
on our 11/34 TSX-Plus system with or
without
the
generalized
data cache.
Because of Lts speed, it is used very
often and is probably my most valuable
productivity tool.

If you are using TSX-Plus 5.1B, you may
find your TECO command files no longer
properly process TSX-Plus command file
control characters. Here is a fix to the
5.1B TSGEN.MAC file from S & H that allows
TECO programs to operate without echoing:
Locate these lines in the non-user-modifiable part of TSGEN.MAC:
.RAD50
.WORD

ITECO

AF$SCA!AF$HIE!AF$NOW
Included here is the core of the program
only, vastly simplified from what I use,
but you can expand on what is here.
A
great deal of what I do works under
T3X-Plus as layers, so the whole system
listing would be larger than we could
print in this newsletter.

Replace the last line with:
.WORD

AF$SCA!AF$NOW

This
change
disables high efficiency
terminal mode, and allows TECO to operate
without echoing while maintaining compatibility with RT-11. According to S & H,
TSX-Plus Version 5.1C will include this
change [Editor's note: it does]. It still
does not allow all command file control
characters (Chapter 3, TSX-Plus Reference
Manual) to work the way they did before.
A! does turn echoing off and A) turns it
on, but A( seems to be ignored, so you get
the TECO commands along with the desired
terminal output.
Perhaps S & H can work
this out in the future.

Don't be afraid of BSTR.MAC.
Very easy
instructions are given, so if you have
never programmed in anything but DIBOL or
DBL, you'll do just fine with BSTR.MAC.
It is included because
an
efficient
backward search is needed to isolate the
text line where a match is found so the
full line may be displayed.
The fast DBL forward search routine INSTR
comes with DBL and is one of my favorites.
I used to wonder why DISC didn't just
change the direction of the search, but,
once I got into it, I found that the
fuctionality was quite different. It did
set up similarly, though • IncidentallY}'
include BSTR.OBJ
in
your
subroutin~
library if you find it useful.
BSTRdocumentation would be similar to INSTR
except that it goes backward.

S & H had some nice ways TECO could be
used in a command file. Many of us have
dozens of these running on our systems.
Contrary to the opinions of some, TEeO is
not dead. It is hard to use, granted, but
some of us went over the hump years ago
and haven't found anything since with
quite so much power on TSX-Plus.

-8-

TSX-Plus Users' Group Newsletter
May 1985
which may contain any valid DIRECTORY
constructs,
including
wildcards
and
factoring, e.g., PRG:*.(DBL,SUB). If the
first valid character (not necessarily the
first character) in a file line is a
semicolon (a comment), any match on that
line is ignored Q~less a IC switch is
appended to the key value.

Jme of the things you can do to make
vFSRCH more productive (very much so) is
front-ending it with a directory listing
obtained from a wildcarded filespec, as
you can do with the TECO routine described
earlier.
Again, I am not incl '-1ding our
complete implementation because of space,
but it is available if you don't mind
developing a complete system.
The basic
key is a subroutine that joes keyboard
commands from within a DBL program -- all
done in DBL.

;Very fast search
.TITLE VFSRCH
;Very fast search.
;Scans a device or file up to EOF or NUL.
;

;Stores results in VFSRCH.TMP or optior-al
file name.

In our version, we ask DIRECTORY to create
a file of names that satisfy our wildcarded and parenthetical (~ot supported by
the TECO routine) file specifications,
then our program reads the file DIRECTORY
produced.
[Editor's note: Borizon has
developed an IND program that does the
same thing; we will try to publish some of
it in the next newsletter]. Our implementation can thus look through multiple
files at a rate of 500 typical source
files in les3 than 10 minutes on an 11/34.
We don't usually ask for a
.'ull scan, so most of what we run takes
only a few seconds, not minutes.

;

;Entire line Nhere match found is
displayed and output except:
(optionally) leading space and tabs
(optionally) if a semicolon (;) is
the first non-tab/space on a line,
any match on the line is ignored.
RECORD
FN,
OFN,
ALL,

J•

tTt:'vC' '7

,.... .L.J J. .J£... ,

COMM,
ENTRY,

Another variation outputs only the file
name whenever a match is found, so we can
scan binary files, too.
Just drop the
parts that output the text line and look
for nulls and EOF characters, and use a
general error trap to detect when processing is done.
BSTR is not needed since
this variation does not output text lines
(although you could have the routine find
and display a limited number of nearby
printable characters if you wished, but
this should be done in high-level DBL).

114
A10
D1
D2
D1
A14

; ; ; ; ; ; ; ;; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;; ; ; ; ; ; ;

;Adjust these sizes upward or downward so
it still fits. Use 512-byte increments.
Adjust blocking factor (see DBL notes
for TSX implementations) like this:
PAGE/512 = Blocking factor.
PAGE,
A14848
HALF, 2A7424 @PAGE
HFPGSZ, D4,
7424
;;;;; ;;;;; ;;;;; ;;;;;;;;; ;;; ;;;;; ;;;;

DBL als~ all~ws a device name as a filespec.
Our 65,535-block system disk was
searched in 28 minutes, about 20K bytes
?er second!
~ajor
changes would not be
required to report the block and byte
where
matches occur, or to test for
printable characters in the neighborhood
of the match.
Such a program could be
handy the next time you have to rebuild a
lir-ectory!

CRLF,
A2
CR,
A1 @CRLF
LF,
A1 @CRLF+1
CTRLZ, 11
NULL,
A1
NUPAG, D1
A1
TAB,
KEY,
A50
SPECSZ, D2
T4'()k'k'~~
...... ..., ............. , ... ,

SIZ,
BEG,
EOL,

When run, the program prompts the user for
both the key and the filespec expression,

-9-

, D1
D5
D5
D5

TSX-Plus Users' Group Newsletter
May 1985

& characters for key.',CRLF)
XCALL DAESC (13)
STOP
END
XCALL DAESC (13)
DISPLAY (15,'Strip leading spaces and
& tabs [default=Y]? ')
READS (15,ENTRY)
IF ENTRY.EQ.'N' THEN ALL=O ELSE ALL=1
FILE,
DISPLAY (15,'File to search:',CRLF)
XCALL ACESC (13)
READS (15,FN)
DISPLAY (15,CR)
IF • NOT. $RDLEN THEN GO TO FI~~E
SPECSZ = $RDLEN
DISPLAY (3,FN(1,$RDLEN),CRLF)
OPEN (2,T,FN,30)
DISPLAY (15,CR,FN,' > ',CR)
GETS (2,PAGE,PX)
IF $FALSE THEN
BEGIN
PX,
EOFERR =
END
CALL EOFCK
NXSRCH,
XCALL INSTR (BEG,PAGE,KEY(1,KEYS2),
&: KEYFND)
IF .NOT.KEYFND.OR.EOF.AND.
& EOF.LT.KEYFND THEN
BEGIN
IF EOF THEN GO TO EOFILE
CALL HALF
IF .NOT.EOF.AND.EOFERR
& THEN GO TO EOFI~E
BEG = HFPGSZ+1
GO TO NXSRCH
END

KEYFND, D5
BOL,
D5
EOF,
D5
NULFND, D5
PROC
OPEN (15,I,'TT:')
XCALL DFLAG (111,1)
XCALL FLAGS (01001010)
XCALL ASCII (O,NULL)
XCALL ASCII (26,CTRLZ)
XCALL ASCII (9,TAB)
XCALL ASCII (13,CRLF)
XCALL ASCII (10,LF)
DISPLAY (15,
& 'Ignores comments unle~s IC appended to
& key.',CRLF,'Wildcards, Ate. accepted.',
& CRLF, 'Key::: ')
KEYENT,
READS (15,KEY)
IF .NOT.$RDLEN THEN
BEGIN
DISPLAY (15,'A valid key must be
& entered:',CRLF)
GO TO KEYENT
END
KEYSZ = $RDLEN
OUTDEV,
DISPLAY (15,10,27,'M',
& 'Output file or devIce name
& [default:::VFSRCH.TMP]:
& 27,'7')
READS (15,OFN)
IF .NOT.$RDLEN THEN
BEGIN
OFN = 'VFSRCH.TMP'
DISPLAY (15,27,'8',OFN,CRLF)
END
ON ERROR OUTDEV
OPEN (3,0,OFN)
OFF ERROR
IF OFN.NE.'TT:' THEN
DISPLAY (3,CRLF,CRLF,CRLF,'«<Search
& for ',KEY(1,KEYSZ),
& ' in filespec ')
IF KEY(KEYSZ-1,KEYSZ).EQ.'/C' THEN
BEGIN
COMM = 1
KEYSZ = KEYSZ-2
KEY(KEYSZ+1,KEYSZ+2) =
END
IF .NOT.KEYSZ THEN
BEGIN
DISPLAY (15,'«<VFSRCH-F-No valid

;Key is found
NUPAG = 0
EOLCK,
XCALL INSTR (KEYFND,PAGE,CRLF,EOL)
IF .NOT.EOL THEN
IF .NOT.NUPAG THEN
BEGIN
NUPAG = 1
IF KEYFND.GT.HFPGSZ THEN
BEGIN
CALL HALF
KEYFND = KEYFND -HFPGSZ
GO TO EOLCK
END
-10-

TSX-Plus Users' Group Newsletter
May 1985

& THEN EOF ~ NULFND
IF EOF THEN EOF = EOF -1
END
ELSE EOF = 0
RETURN
END

ELSE EOL = EOF
END
IF EOL THEN EOL = EOL-1
XCALL BSTR (KEYFND,PAGE,CRLF,BOL)
IF BOL THEN BOL = BOL+2 ELSE BOL =
IF BOL.LT.EOL THEN
BEGIN
WHILE ALL.AND.(PAGE(BOL,EOL).EQ.' ,
& .OR.PAGE(BOL,EOL).EQ.TAB)
& DO PAGE(BOL,EOL) =
& PAGE(BOL+1,EOL)
IF PAGE(BOL,EOL).NE.';'.OR.
& COMM.OR.ALL.EQ.O THEN
BEGIN
IF .NOT.ALL THEN SI2=EOL+1-BOL
& ELSE XCALL TRIM
& (PAGE(BOL,EOL),SIZ)
IF SIZ THEN
BEGIN
DISPLAY (3,FN,' >
,
& PAGE(BOL,BOL-1+SIZ),
& CRLF)
DISPLAY (15,FN,' > ,
& PAGE(BOL,BOL-1+SIZ),
& CRLF)

Next, BSTR.MAC:

PTR:
LNG:

.TITLE
.ENABL
.MCALL

BSTR
LC
.TTYOUT

;Where string is.
;How long string is.

°.GLOBL BSTR
.GLOBL
.GLOBL

BSTR:

END

$CKARG,$CLRLT,$POPNM
$FMTNM,$CHKLT

JSR
RO,$CKARG
.WORD
4
CALL SETUP
CALL MATCH
CALL DONE
RETURN

END
END
FN =
BEG = EOL+1
GO TO NXSRCH

SETUP:

MOV

MOV
CALL
MOV
MOV
CALL
MOV
MOV
CALL
CALL
MOV
MOV
RETURN

EOFIi..E,

CLOSE 2
CLOSE 3
DISPLAY (15,'Output goes to ',OFN,CRLF)
XCALL DAESC (13)
STOP
HALF,
HALF = HALF(2)
GETS (2,HALF(2),CTRLZ)
CALL EOFCK
RETURN

MATCH:

STELZ,
EOFERR =
EOFCK,
IF $RDTRM.EQ.O.OR.$RDTRM.EQ.26
& .OR.EOFERR THEN
BEGIN
XCALL INSTR (O,PAGE,CTRLZ,EOF)
XCALL INSTR (0 ,PAGE,NULL,NULFND)
IF .NOT.EOF.OR.NULFND.LT.EOF

9$:
10$:

-11-

DEC
MOV
TST
BEQ
CMP
BLT
CMP
BLT
ADD
MOV

BR
DEC
MOV
BEQ

-(R5),-(SP)
-(R5),-(SP)
$CLRLT
-(R5),LNG
-(R5),PTR
$CLRLT
-(R5),R1
-(R5),RO
$POPNM
$CHKLT
(SP)+, (R5)+
(SP)+, (R5)+
RO
RO,-(SP)
LNG
40$
R1,R2
40$
R2,LNG
40$
R2,RO
R2,Rl
iO$
RO

PTR,R1
20$

TSX-Plus Users' Group Newsletter
May 1985
15$:

20$:
21$:

40$:
45$:
DONE:

DEC
CMP
BLT
BR
MOV
MOV
INC
DEC
BEQ
CMPB
BNE
BR
MOV
SUB
RETURN
MOV
JSR
RETURN
.END

A)"(DIR A1
A)R SETSIZ
SY:DIR/D:O
AC

RO
RO, (SP)
40$
10$
LNG,R4
RO,R3
R3
R4
45$
(Rl)+,(R3)+
9$
21$
(SP) ,RO
(SP)+,RO

For non-sorted directory operations, use
the DIR command as usual.
For sorted
directories, use DIRSRT, which temporarily
allocates
maximum
memory for DIR.SAV
before running it, then restores
its
original, smaller size. For example:
DIRSRT SY:*.SAV/SORT:DATE/REVERSE
If

are running DIR from a program or
file and do not know whether sorts
will be required, use DIRSRT.

RO,R4
R4,$FMTNM

you

co~and

COMMENTS ON 5.1B

----------~

Now compile:

Everyone
else
must have gotten this
version long before I did, so you m~y have
all the glitches worked out. [Editor's
note: Actually, .John, we've about worn 'Jut
5.1CJ.
Probably the most devastatLng
change is the lack of terminal support for
TECO in command files. We have 33 quick
TECO routines (literally progra~s)
in
TSX-Plus command files, many of which run
from startup command files.

R DBL
VFSRCH=VFRCH

R MACRO
BSTR=BSTR

R LINK
SY:VFSRCH=VFSRCH,BSTR,DLIB

No matter how big our silos, i~put characters get dropped and the programs have to
be rerun. Maybe help will come with the
next
version???
There could be some
better information on how to set up silos.

Now run it:
R VFSRCH
Several people have indicated a willingness to share what they have on global
searches, but as of this writing nothing
has arrived. If I get something later,
I'll publish it in a future newsletter.
Send me yours.

We're still trying to find a way to end a
CL program's input when no more comes in.
The program is halted in 10 and presently
requires a reboot to flush. Some of our
problems may be due to the lack of DBL
support for some of the CL handler special
functions. If it can be rigidly controlled, though, there are some very nice
features with CL. One struggling point:
Ever notice how many places you need to
find to assemble the full set of CL
documentation?

MEMORY SAVER
You need lots of main memory to run
DIRECTORY if you sort the output (!ALP,
ISORT:category). Otherwise, only about 14
Kbytes is required. Enter the following
as file DIRSRT.COM:

TSX-Plus has done some very nice things in
5.1. We'll have to give them time while
we both work out some of the glitches. It
should be well worth the effort.

"! ")R SETSIZ _.

SY:DIR/D:64.

-12-

TSX-Plus Users' Group Newsletter
May 1985
Also,

TT: is TT: even when it should be
if you know what I mean~ The differences between systems when using DBL are
minor.

DBL VERSION 4

KB~7

DBL v4 will begin shipping around .July. I
had been assured by some that it would be
out in January 1985, but haven't heard
much of that nonsense since about March.

Sort: This is a fast internal sort.
Some
of DBL's implementations do 1,000 80character record sorts in 15 seconds,
while the slowest computer took 25 seconds; this is RTSORT-level performance.
DISC says there will not be a lot of
difference in speed between RTSORT and
their internal sort. They account for the
difference by optimizing for the kinds of
sorts that DBL is likely to be doing.
There may be some functional differences,
however, such as mapping.

It is almost embarrassing to get excited
over a new compiler. We used to become
depressed about it -- those of us who had
DIBOL v3b. I imagine there are still some
who are running 3b rather than a new
version because of the workarounds each
version required in those days. But there
were some of us that knew things had to
get better.
DIBOL has since become a
solid product and is a pleasure to use.
It is concise, programs simply, compiles
quickly, is not as wordy as COBOL, and
runs at a respectable speed if properly
implemented.

Virtual environment: Like VAX, write your
programs in a single dimension
long~
If you move your program to a VAX, the
hardware provides virtual code and data
management; while running under TSX-Plus,
DBL does the same thing with software.
DBL includes a compiler AND a linker that
manages the amount of code and Qa~a ~na~
is really needed in memory at any time.
It sets flags to control use, re-use, or
freeing of memory-resident code and data.
Its paging system uses a least recently
used algorithm for maximum efficiency and
minimum 1/0 overhead.

I understand DIBOL is so solid it may
become an ANSI standard. DEC used to say
it outsold FORTRAN on their systems, but
that may not be the whole S~Qry because
few DIBOL systems are developmental.
Many of us went on to DBL, seeing it as a
superset of DIBOL. Programming started to
be more fun. And DBL has a lot of things
that are worthless until you have to
maintain a program (which will probably be
60% of the time spent on that program).
It carries the principles of structured
programming beyond the basics. DBL v1 was
just DIBOL compatible.
Version 2 added
structured programming constructs.
Skip
v3
it went to VAX. That brings us to
v4.

Assembly language interface: This excellent feature of DBL will be as well
documented as before, enabling us to use
high
level DBL for IIO and assembly
language for crunching. How it is done
will change, though. For example, it now
requires PIC, not object code.

So, those of us using DBL under TSX-Plus
are waiting for July and will be talking
about many things that will be new.
Here
are some of them:

Multi-key ISAM: The really nice thing
about DBL's ISAM is that it is self-reorganizing. There was a time when I spent
14 hours a week reorganizing DIBOL ISAM
f~les, believe it or not.
Under DB~'s v2
ISAM you could occasionally run out of
stalls for data even though there was
enough file space. This should no longer
happen in v4 due to roomier index blocks
and better insertion and removal algorithms.

"Floating" environment: DBL will now run
essentially the same wherever it goes.
The environment "floats" as you move a DBL
program around. We moved our VT -52 DIBOL
programs to VT-100s under DBL without
changing the code. Now we can move to tl~A
and our programs delete the old version of
a file when we CLOSE, just as they do
under TSX, but not as RSX usually does it.

SET: Initializes a whole batch of variables so your listing doesn't need two
inches of =03. Isn't that nice?
-i3-

TSI-Plus Users' Group Newsletter
May 1985
built to make DBL reasonable in size an~
performance; it would limit systems to 2-3
users under one alternative. Since the
product could not be introduced that way,
developing the new front end became more
important than the delay.
Ken Lidster's
spacious office was crammed with four
tables and three more lines were brought
in. The front end was redesigned.

NOP: It will be nice to have a no-op in a
string of sequential decimals in CASE and
USING blocks.
SCAN: Use it in WHILE and UNTIL lines to
look for either the beginning or the end
of the condition. Do you have any idea of
the number of lines of code this will
replace? Remember how you did it in DIBOL
v3b?

DBL will go to field test in June. Ken is
quite confident it will ship in July. I'd
sure hate to have to make a statement like
that!

IF-THEN-ELSE and other structured extensions will still be present in DBL. By
the way, there are many
undocumented
variations of IF-THEN-ELSE that caused
some to avoid v2.2. Check them out before
you say it can't be done!

With all the effort put into the TSX-Plus
version of DBL, what does DISC see for the
future of RT-11? Milton Campbell's "RT-11
Perspective" column in the April Hardcopy
brought up the question; he correctly made
the distinction between the "end" of RT-11
and the end of enhancements
to
it.
Lidster sees "no end in sight" because of
TSX-Plus. There is no other small system
with the features for the price of TSXPI us •
He sees a limi t of ten user,:running "average" programs because of th .
RT-11 file structure,
espeCially
the
directory
system,
which TSX-Plus has
improved with its caching algorithms.

Multi-dimensional arrays: This is what we
had hoped it would be.
It replaces the
old computational method of addressing
positions within a set of data. How many
times has your code for addressing parts
of a tax table looked like an engineering
algorithm? It is nice to see this feature
in a DIBOL-like language.
Fixed-point decimal: Now we
decimal point in our numbers.

can

put

Why is DBL being released so late? Those
of us in the front line of this business
know some of the problems that interfere
with getting a product out.
Ken Lidster
of DISC gave the flow of how it went since
I first knew about their version 4 plans
back in 1983:

With products like TSX-Plus and DBL, the
picture keeps getting better all the time!
DIBOL MESSAGE INTERFACE
DIBOL users have had to modify their
programs somewhat to handle messages under
TSX-Plus.
A user has contributed (1) a
routine that will handle DIBOL messages to
RTSORT via SEND/RECV, and (2) a more
general DIBOL message
interface
that
allows sending and receiving messages with
no chance of message misdirection. It is
for use with DIBOL earlier than v8.QQ.

Before DISC was to start work on the
TSX-Plus version of the new DBL, it wanted
to have its micro program in full swing.
The project took five programmers seven
man-years and the time block slippage was
3-4 months; not all that bad if you
consider the size of the project. Well,
the TSX-Plus
project
was
originally
targeted for mid-1984, then refined to
October 1984. Due to the micro slippage,
it became January 1985.

Contact me for directions on how to get a
copy. It is well written and documented
and would be an essential part of someone's DIBOL/TSX kit.

Porting to TSX-Plus from MS-DOS ran into a
major problem-with the PDP-11 addressing
limitation
of 64Kbytes, affecting the
memory resident overlay structure. A new
front end to the TSX-Plus system had to be

CONTRIBUTIONS
Send your routines (and
like to) to:
-14-

money,

you'd

TSX-Plus Users' Group Newsletter
May 1985
John J. SteTtlart
Minne Assn. of Comm. & Industry
480 Cedar St.
St. P~ul, MN 55101
(612) 292-4684

V5.1C adds several new features.
Chief
among them is the ability to set character
length and parity for DZ11, DZV11, DH11,
and DHV11 ports via commands; running
SYSMON on Hazeltine and ADM3A terminals in
addition to VT52, VT100, and VT200 terminals; two new CL handler special functions;
increased numbers of ACCESS and ASSIGN
commands; a change to the DIBOL DTSUB
routine to handle record numbers larger
than 65,535; and a change to TSAUTH which
inhibits display of account passwords.

CURRENT RELEASES
The latest release of TSX-Plus is Version
5.1C.
For PDP-l1/LSI-l1
users,
this
release principally fixes several bugs
left over in 5.1B, but it also makes
TSX-Plus runnable on the PRO-380. If you
are under support, you might want to pick
up this version.
To the best of our
knowledge, there is no release currently
under field test, although S & H is
planning another new release tn the near
future (hopefully, we will see the end of
5.1x before x gets to Z).

V5.1C also corrects over 35 bugs in V5.1B
ranging from stack overflows when using a
parallel printer (LP handler) to incorrect
SYSMON time displays when the system clock
ran at 50 Hertz. As usual, S & H Technical Support has been responsive in ac=
knowledging
and
correcting
problems
encountered with new releases.

-I:J-

TSX-Plus Users' Group Newsletter
May 1985
.TITLE

ONLINE - TELL EVERYBODY THE SYSTEM IS UP

COPYRIGHT (C) 1985 HORIZON DATA SYSTEMS
1901 WILDFLOWER TERRACE
RICHMOND VIRGINA
ONLINE SENDS A USER-DEFINED MESSAGE TO ALL TSX-PLUS TIME SHARING LINES
THAT ARE NOT PRESENTLY LOGGED ON. ONLINE SENDS THE MESSAGE BY
ASSIGNING A FREE (EXTRA) CL UNIT TO EACH LINE IN TURN, AND WRITING
THE MESSAGE TO THE CL UNIT IF THE LINE IS NOT IN USE. TO USE ONLINE,
THEREFORE, THE "CLXTRA" PARAMETER IN TSGEN MUST BE AT LEAST 1.
TO RUN ONLINE, USE THE FOLLOWING COMMAND SEQUENCE:
R ONLINE
<INSERT MESSAGE TEXT HERE>
THE MESSAGE IS PRECEDED BY <CR><LF> AND TWO BELLS, AND IS FOLLOWED
BY <CR><LF><LF> WHEN SENT.
YOU MUST DEFINE TWO PARAMETERS IN THIS PROGRAM BEFORE ASSEMBLING IT.
THE FIRST, "CLUNIT", IS SET TO THE UNIT NUMBER OF THE FREE (EXTRA)
CL USED TO SEND THE MESSAGE. THE SECOND, "V51C" IS SET TO 1 IF YOUR
TSX-PLUS IS VERSION 5.1C OR LATER, 0 OTHERWISE. WHILE RUNNING UNDER
5.1C, ONLINE USES .SPFUN 262 TO ENSURE THE MESSAGE HAS BEEN DELIVERED
TO EACH TERMINAL BEFORE REASSIGNING THE CL UNIT. OTHERWISE, A TIMED
WAIT OF ONE SECOND IS USED BETWEEN SENDS TO PREVENT CONFUSION CAUSED
BY RAPID CHANGES TO CL ASSIGNMENTS.
THIS PROGRAM MAY BE FREELY COPIED OR MODIFIED BUT NOT FOR COMMERCIAL
PURPOSES.
USER DEFINED PARAMETERS
CLUNIT=O
V51C=1
•MCALL
.IF
.MCALL
.IFF
.MCALL
•EN DC

JUSE CLO TO SEND MESSAGES
;RUNNING UNDER V5.1C
.EXIT,.GTLIN,.LOOKUP,.PURGE,.WRITW
EQ V51C
•TWAIT
.SPFUN

CONSTANTS
BELL=7
CHAN=O
CR=15
ERRBYT=52
LF=1i"
NOLINE=3
OUTCHK=262

jBELL CHARACTER
JUSE CHANNEL 0 FOR IIO
;RETURN CODE
;EMT ERROR BYTE
;LINE FEED CODE
;LINE DOES NOT EXIST
;OUTPUT CHECK .SPFUN CODE

-16-

TSX-Plus Users
May 1985

Group Newsletter

3TART: :
i$:

2$:

MOV
.GTLIN
TSTB
BNE
DEC
MOV
MOVB
BNE
SUB
ASR
CLR

;POINT TO USER MESSAGE TEXT
;GET MESSAGE TEXT
;LOOK FOR TERMINATING NUL
;KEEP LOOKING TILL FOUND
;POINT TO NUL
;POINT TO TRAILER STRING
;COpy TRAILER OVER
; UNTIL TERMINATOR FOUND
;GET MESSAGE LENGTH
;CONVERT TO WORD COUNT
;START WITH LINE 0

flMSG BUF j R3

R3 , II PROMPT
(R3)+
1$
R3
UTRAIL,RO
(RO)+,(R3)+
2$
flMSG ,R3
R3
R2

CHECK NEXT TIME SHARING LINE
;

3$:

INC
MOV
MOV

R2
R2,LINE
IISEIZE,RO

EIvrr

375

BCC
CMPB
BNE
.EXIT

4$
@HERRBYT,HNOLINE
3$

;BUMP THE LINE NUMBER
;PUT IN CONTROL BLOCK
;POINT TO CONTROL BLOCK
;SEIZE LINE AS CL
;BRANCH IF WE GOT IT
;SEE IF ALL LINES DONE
;TRY NEXT LINE IF NOT
;ELSE WE ARE DONE

OPEN A CHANNEL TO THE TERMINAL AND WRITE THE MESSAGE
4$:

.LOOKUP UAREA,#CHAN,UDEVBLK
BCS

5$:

;OPEN THE CHANNEL
;IGNORE IF FAILS
;OUTPUT THE MESSAGE

.WRITW #AREA,#CHAN,#MSG,R3,#0
.IF
EQ V51C
•TWAIT #AREA,#TIME
;WAIT A SECOND FOR MESSAGE
.IFF
.SPFUN fIAREA, fICHAN, f,IOUTCHK, !,IAREA, 4/1 , ·fIO, flO ; SEE IF OUTPUT IS DONE
TST
AREA
;CHECK CHARACTERS PENDING
BNE
5$
;CHECK AGAIN IF NOT DONE YET
.ENDC
•PURGE flCHAN
;CLOSE UP THE CHANNEL
CLR
LINE
;LlNE = MEANS DROP CL
MOV
flSEIZE ,RO
;POINT TO CONTROL BLOCK
EMT
;RELEASE THE LINE
375
BR
;TRY THE NEXT LINE
3$

CL LINE SEIZE/RELEASE CONTROL BLOCK
;

SEIZE:
LINE:

•BYTE
•WORD
.WORD

0,155
CLUNIT

;et UNIT NUMBER

;LINE NUMBER TO SEIZE

;EMT FUNCTION CODES

DEVICE I/O CELLS
AREA:

.BLKW

T_Tnnn
IEVBLK: .wunu

TIME:

.RAD50
.IF
.WORD

6.
ARCLO+CLUNIT
"TEMP TMP"
EQ V51C
0,60.

;EMT AREA
;DEVICE NAME FOR LOOKUP
;DUMMY FILE NAME
JONE SECOND TIMER BLOCK
-17-

TSX-Plus Users' Group Newsletter
May 1985
.ENDC
CHARACTER BUFFERS
;

MSG:
•BYTE
MSGBUF: .BLKB
• BLKB
•EVEN
TRAIL: .BYTE
PROMPT: .ASCII
• EVEN
•END

CR,LF,BELL,BELL
81.
4•

;MESSAGE HEADER
;FOR USER MESSAGE TEXT
;ALLOW FOR TRAILER

CR,LF,LF,O
"MSG>"(200)

;NULL TERMINATED TRAILER
;.GTLIN PROMPT

START

-18-

--,----

ANSWERS PRICE LIST
Answers .......................... 7 54
Answers (requiring thought). . . . . . .. 1 .25
Answers (correct) . . . . . . . . . . . . . . . .. 2.50
Dumb Looks Are Still Free

TSX-PLUS USER'S GR01JP

~
Six Phases of a Project
• Enthusiasm
• Disillusionment
• Panic
• Search for the guilty
• Punishment of the Innocent
• Praise and honors for the non-participants
TSX-PLUS USER'S GROUP

TSX-PLUS USERS' GROUP
clo ZIA CORPORATION
P.O BOX 351
MORRIS PLAINS NJ 07950
ON-SITE MEMBERSHIP APPLICATION
Please type or print neatly. Limit data to 36 characters
per line so it all fits on our mailing labels.
Name [NA]:
Title [TI]:
Department [DP]:
Company Name [CO]:
Address, Box, MiS [AD]:

City, State, Zip [CT]:
Country [CN]:

-~---.--~---------

.. -

---.-----~.~=-

-----

Computer Types [CP]:
Storage Devices [MS]:
rO'T'l.
Printer Types '-.&.,.
....
..I •

Terminal Types [TM]:
S & H Software Used (check all that apply):
[

] TSX

[

] TSX-Plus

[

] DPS

[

] RTSORT

[

] COBOL-Plus

Programming Languages Used (check all that apply):
i..

] MACRO

[

] FORTRAN

[

] PASCAL

[

] C

[

] DBL

[

] DIBOL

[

] APL

[

] Other

[

] BASIC

PLEASE briefly describe what you would like the Users' Group to do:

Submit only one application per person. You may copy this form as
required.
Include payment with your application and send to the
address at the top of this form. Memberships are $15.00 per year
in USA, Canada, and Mexico, $20.00 per year elsewhere. All prices
are in US $; foreign subscribers must remit by check or money order
payable at a US bank. Submitted applications become the property
of the TSX-Plus Users' Group.

Tsxpua
c/o Glenn Burnett
Zia Corporation
P.O. Box 361
Morris Plains, NJ 07960

FIR~ASS

u.s. POSTAGE
PAID
PERMIT No. 13
Morris Plains, NJ 07950