Home Page Publications
Home Page Publications
The area of command languages has
been rich in activity since I built my first command language
application seven years ago. The following are what I believe
are the current major trends for command languages:
* There is consistent growth in the number of new command language interpreters introduced each year.
This includes new DCL, REXX, and
Mac-based command language interpreters.
In addition, these interpreters are appearing on an increasing number of operating systems and environments.
Command languages available on a
variety of platforms include Batch language (BAT), CLIST, DCL,
Icon, Python, REXX, Perl, and various UNIX shells.
This means that portability and
interoperability of command language programs across operating
systems are increasingly becoming a real concern for many programmers.
* Command language compilers
are appearing resulting in faster run times. Compilers are
available for Icon and REXX and are being vigorously requested
for other command languages.
* Command languages standards
are starting to appear such as UNIX shells (POSIX 1003.2)
and REXX (ANSI X3J18).
* Graphical User Interface-based
(GUI) environments such as MS Windows®, X Windows, and
Macintosh® OS are beginning to have command languages
available such as Bridge, Visual BASIC, and Frontier.
* 1992 will mark the appearance
of the first "one stop" reference on the major command
languages (Batch, CLIST, DCL, Perl, and REXX) written by the
author and published by Van Nostrand Reinhold. Included are sections
on portability of command language commands across operating systems,
extending each command language, and choosing a command language
(including comparison charts and benchmarks). The appearance of
this book is hoped to further interest in command languages.
NESUG also has a history of being
involved with command languages:
* This year, for the first time,
has sponsored a "SAS and Command Language" course given
November 1st at Pace University Graduate Center by the author.
Has had a Host Systems section
since 1989. Table 1 shows the breakdown by command language for
Host Systems (and talks that mention command languages but were
in other sections).
Table 1 Breakdown of NESUG Command
Language Talks 1989-1991
| General | German(1991) |
| Batch (OS/2) | Darden(1990 ) |
| (DOS) | Greenleaf & Kalyandrug(1989) |
| CLIST | German(1989,1990), Vye and Dickstein(1989) |
| DCL | Strekel(1989) Vizard(1990) |
| REXX | German(1990) Gleason(1989)
Lee(1990) Whitley(1991) |
| UNIX Script | Painter(1990) |
Since NESUG '89, I having been presenting
information on how to take advantage of this "grand alliance"
-- in developing the right type of application the first time.
(For the record, I've been accused of knowing seven command languages
but there are only six that I will acknowledge!)
Last year, I presented a model that
showed how command languages can be used anywhere in the
SAS development cycle. Knowing the type of application determined
the tasks to perform and corresponding commands to do them. This
paper, builds on last year's effort and does a selective content
analysis of past SUGI presentations to tap this important information
resource.
[A version of this paper was presented
for the Boston Area SAS Users Group earlier this year. The presentation
at NESUG 91 will be topics entirely different from this paper:1)
choosing a command language, 2) command language tips and traps,
and 3) command language style. Copies are available on request.]
Command Language Presentations
at SUGI (1981-1989)
The above years were picked as a
sample of convenience. (Ok, I confess they were the proceedings
I had lying around the office.] One could make the claim that
study of SAS and command languages mirrors the growth of the SAS
system. Table 2 shows some of these trends by years. Keep this
framework in mind as we discuss SUGI papers a little later on.
Table 2 Trends in command language papers at SUGI
| 1981-1985 | The height of "SAS on the mainframe frenzy." Talks were on front-ends or applications using CLIST and ISPF (TSO), or EXEC2 (CMS) |
| 1986-1989 | A greater diversity of SAS and command language talks:
- Using DCL as a "macro"
language.
- Using batch files for front-end to SAS - Using UNIX shells to pipe data to/from SAS programs |
| 1990s | Command language presentations appear to be on the decline due to the arrival of some SAS products that do the same thing.
Some talks on building applications across multiple operating systems. |
To begin tracking down the number
of SUGI SAS and Command Language articles is a difficult task.
For example, an article on a SAS TSO CLIST application may be
found under Command lists, CLISTs, ISPF, TSO CLISTS, TSO/E CLISTS,
MVS. The best way to track down articles is to use the master
index as a starting point, followed by the index of the specific
proceeding, and ending up looking at the table of contents for
a specific proceeding. Using only one of these is usually not
sufficient.
Table 3 presents a breakdown of
presentations given between 1981 and 1989 at SUGI. The first column
represents the articles mentioned in the SUGI master index. The
second column represents the articles mentioned in the index for
the specific proceeding.
Table 3 SUGI Command Language Presentations
1981-1989
| Topic | Master Index | Index Each Year |
| General | 2 | 1 |
| Batch | 0 | 5 |
| CLIST | 7 | 22 |
| DCL | 3 | 6 |
| REXX | 2 | 6 |
| UNIX Shell | 1 | 1 |
The remaining part of this paper
will discuss 1) various tips and recommendations these presenters
made, 2) some questions to ponder about SAS and command languages.
SAS and Command Language Applications
At last year's NESUG, I proposed
that all command language applications can be divided into two
topics:
*Housekeeping Applications
For some environments, SAS needs
a command language program to perform "housekeeping"
(such as allocating a dataset) before being invoked. Also placed
under this category were system utilities (such as archiving datasets/files)
and login scripts.
*End-User Applications
By using other environments, command
languages can be used to produce user-oriented, full-screen self-contained
applications.
Each command language application
developed by SUGI presenters fits into one of these two categories.
Space does not permit me to list references by category
Tips from Past SUGI papers
The following are some of the tips, traps, and
conclusions reached by SUGI command
language presenters: [Comments in brackets are mine.]
*Hofacker and Hoffman(1986) mention
applications can interoperate between PC-DOS and VM if using SAS/RTERM,
VM/Bond and REXX.
*Landon(1986) uses the technique
of having a menu where the user selection is the name of the batch
file invoked which sets up the proper path and data files. [With
the arrival of the CALL command in DOS 3.3, invoking nested batch
files and returning from them became much cleaner.]
* Young(1988) notes that file compression
can reduce SAS and other files by 10-20%, [Young was using ARC.
Later compression utilities would probably yield higher rates.
He uses the ECHO command to provide a value to CHKDSK (ECHO
N | CHKDSK /F). He also employs FIND to find a string in one
file and redirect it to another (FIND "SAS" TEXT.TXT
> TEXT2.TXT) [With DOS 5.0, FIND can now search for case-insensitive
arguments.]
*Davis and Sweetland(1987) mention
that the CHKDSK/F command should be placed in the batch file that
invokes SAS. They mention how batch files (with PATH statement)
can be used to customize the SAS environment.
* Economou (1989) uses a variety
of interesting techniques. These include: 1) IF EXIST dataset
...COPY (implied then), 2) BASIC program to set printer top
of form, and 3) sort directory before backup.
This is just a sample of the many
tips on CLISTS:
Tilanus et al (1982) report that
SAS, CLISTs, ISPF, and SAS/FSP are a good combination for prototyping.
They also use the ISPF system variables &ZUSER to "remember"
who the current user is. ISPF table and file tailoring service
were also used.
Dickinson(1982) built CLISTs that
users used as commands to invoke and process SAS programs. This
saved the user worrying about such things as the syntax of the
ALLOC command. However, ISPF panels were also available. This
gave users a choice on the interface to use. Porting to EXEC2
on CMS was also mentioned.
Timmins and Lucke(1982) used the
CLIST QSAM facility to build a control file used to determine
which SAS macro to invoke.
LaRue(1982) makes the following
observations: 1) Where a CLIST library is allocated in a SYSPROC
concatenation means everything, 2) Use a READ following by 26
variables to set a title. [For current CLIST users, the following
sequence would reduce "variable rot."]
READ
SET &TITLE=&STR('&SYSDVAL')
Hudson and McBride(1982) has a good
definition of end-user command language applications
"Allows the user to function in a conversational mode
with the system while maintaining strict control over
data entry"
Sauer and Allee(1986) offer these
tips when using TSO commands in SAS macros:1)Use TSO commands
in CLISTs [and now EXECs] instead of %TSO and X statements to
suppress operating system messages. 2)Use the macro variable SYSRC
and the CLIST EXIT statements to pass the result of a TSO
command back to the SAS system.
Suchoski and Magnuson(1986) used
the return code based on ACF protections to see if a user could
use a CLIST.
Folsom(1986) invoked SAS and then
passed the program as part of a DATA-ENDDATA structure.
Corcoran(1986) Coded a SUBMIT
* followed by the JCL statements to build and submit a JCL
program at run-time.
Blaylock and Salvo(1985) [Back in
the days where SAS macros on VMS were only a future enhancement.
used an INQUIRE to capture values then wrote them out to a file.
[Note the preferred method is to use READ. INQUIRE converts input
to uppercase and symbolic substitution can also unexpectedly take
place.]
Khoury and Laskarzewski (1985) used
the F$SEARCH lexical function to search for a dataset before deleting
it. A series of WRITE statements translated the data. [Would the
availability of F$FAO (formatted output) lexical function back
then have changed the look of this program?]
Rhoads(1988) discusses how to build
a command procedure to submit SAS batch jobs. This one of several
papers on this topic. [To me, it seems more efficient to create
your own SAS DCL verb such as Gibes(1989). If you look at SAS
system libraries, this is the way the SAS image is currently invoked.]
Oakley(1987) is interesting because
both an IBM and VMS solution are presented for the same problem.
She has a good example of how to conditionally include SAS statements
[Before the days of VMS SAS macros. ]
Martin(1989) has a COM file used
to monitor who is using SAS.
Cantrell and Wahl(1989) is interesting
to note in passing as a case of an application converted from
CLIST to DCL.
REXX remains the one language available
on nearly all the platforms that the SAS system is. (This includes
DOS, OS/2, CMS, TSO, MVS, UNIX, and possibly soon VMS.) Here
are some tips and guidelines in using REXX:
Lahr(1989) The user just presses
enter and REXX does the rest...
Siu and Johnson(1989) uses a REXX
program to update a master database and checks if any users of
the system are still logged on.
Day and Keyser (1989) used REXX
to create a SAS dataset (using the almost portable EXECIO.) This
could not be easily done using SAS.
Gleason (1989) dreams of a UNIX
REXX interpreter [available today from the Workstation Group]
and the desire to port REXX wherever SAS is. His 1987 article
gives a good overview of REXX across various environments and
discusses the desire for a REXX SAS interface.
LeBlanc(1988) Uses REXX to access
the SAS sample library. He does a LISTFILE of all files with a
type of HELPMSG. The titles from each file are then extracted
and written to a file.
Pan(1988) placed REXX values into
SAS variables. No code is shown doing this.
James et al. (1986) is one of the
first REXX papers presented at SUGI. It also mentions replacing
CLISTS that were nothing more than a series of questions with
SAS/AF.
Burlew(1987) used REXX calling XEDIT
to run a specified sample library member.
Denney(1987) presents a variety
of REXX tricks including invoking SAS/FSP and integrating SAS
HELP with CMS help.
Last but not least to the SAS/Command
Language world are UNIX shell languages. A variety of shells are
available. The command language for system administrators appears
to be Perl (Practical Extraction and Report Language.) Future
NESUG and SUGI talks should discuss how to use Perl with SAS.
It is also available on the Amiga, DOS, and OS/2. Here are a few
tips on UNIX shells and SAS:
Ziegler, Betancourt, and Kelly(1988)
is one of the first paper presented at SUGI that discusses SAS
on UNIX platforms. Examples of the interaction between SAS, system
commands, and external scripts are shown.
Betancourt(1989) has a good example
where he does a file listing, redirects it to a file which is
then processed by SAS.
Johnson(1989) also shows how you can pipe output
to or from a system command (or
a command language script.) This fits well with UNIX's philosophy
of combining tools serving as filters (via pipes and redirects)
to get a job done.) Examples are also provided passing parameters
between a script and a SAS program.
Thoughts to Ponder
Unfortunately, this can only be
a subset of the good information presented from past SUGI papers.
I hope it encourages future investigators to look at this area.
I leave you with the following questions to ponder:
* Are command languages really needed
for SAS applications anymore with the arrival of SAS/AF (and SCL),SAS/ASSIST,
and SAS/EIS?
* If yes, then is it time for SAS
to support a standard command language across all platforms and
perhaps include an interface to that command language? (The best
candidate so far is REXX.)
Will command languages successfully
make the transition as support tools to graphical user interfaces?
Will Microsoft's desire to push
Visual BASIC and Quick BASIC hurt the future of REXX?
Will the forthcoming REXX standard
make any difference?
Is the portability of command
languages still a concern any more?
Getting in touch with me/Trademarks
Hallett German
GTE Laboratories Inc
40 Sylvan Road
Waltham, Ma 02254
617-466-2290
hhg1@bunny.gte.com
SAS ® and all other SAS products
mentioned is a registered trademark of the SAS Institute
UNIX® is a registered trademark
of ATT
Windows, MS, and DOS are trademarks
of Microsoft
CMS, ISPF, OS/2, and TSO are trademarks
of IBM
VAX, VMS, DCL, and DEC are trademarks
of the Digital Equipment Corporation.
References
Betancourt, Randy "Using the SAS System for Controlling Distributed Processing under UNIX and PC DOS" SUGI 14
Proceedings
1989 pp. 1630-1636
Blaylock, Dewey and Salvo Larry
" An Environmental Data System using SAS Software on the
VAX 11/780" SUGI 10 Proceedings 1985 pp.
675-678
Burlew, Michele "Accessing
SAS Software Sample Program Libraries Using CMS REXX" SUGI
12 Proceedings 1987 pp. 966-967
Cantrell, Casey and Douglas Wahl
"Exploiting SAS System and DCL System Features to Replicate
an MVS CLIST Management Reporting System under VMS" SUGI
14 Proceedings 1989 pp. 782-785
Corcoran, Chris "SAS and the IBM Dialog Manager at Miles
Pharmaceuticals" SUGI 11
Proceedings 1986 pp. 587-592
Day, Bob "A Quick and User-Friendly
Comprehensive Front-end for the SAS/GRAPH Procedures " SUGI
14 Proceedings 1989 pp. 1467-1471
Darden, Sarah "The SAS System
Under OS/2" NESUG 90 Proceedings 1990 pp. 206-209
Davis Gordon and Scott Sweetland
"Efficiency and Performance Considerations for the SAS System
under PC DOS " SUGI 12 Proceedings 1987 pp. 779-787
Denney, Michael "The SAS System
Under CMS: Tricks of the Trade" SUGI 12 Proceedings 1987
pp. 974-977
Dickinson(1982) "A General Purpose TSO Interface
To SAS Using SPF" SUGI 7
Proceedings 1982 pp. 454-459
Economou, Tenya " A PC Menu-Driven
System: Integrating the SAS System for Personal Computers with
DBASE III PLUS and DOS Batch Files" SUGI 14 Proceedings
1989 pp. 992-998
Folsom, Todd "A SAS/FSP Software
and CLIST System for Easy Management of a Macro invertebrate Reference
Collection" SUGI 11 Proceedings 1986 pp. 341- 345
German, Hallett The Command Language
Advisor for PCs, Minicomputers, and Mainframes Van Nostrand
Reinhold (Forthcoming April 1992)
German, Hallett "The Grand Alliance: Using TSO/E CLIST and REXX Command Languages in the SAS®
Program Development Cycle"
NESUG 90 Proceedings 1990 pp. 216-219
German, Hallett TSO/E Clists: Basics, Applications,
and Advanced Techniques 1990
Van Nostrand Reinhold New York, NY
German, Hallett "Using TSO/E CLISTs and ISPF for
Generating SAS Programs" NESUG
89 Proceedings 1989 pp. 67-73
Gleason , Chapman "A CMS VM Batch Facility REXX EXEC for the SAS System" SUGI 14 Proceedings 1989
pp. 673-675
Gleason, Chapman "An Introduction
to SAA REXX for SAS Programmers" NESUG 89 Proceedings
1989 pp. 74-79
Gleason,Chapman "Software Mirrors in a Mainframe/Micro
Environment" SUGI 12 Proceedings
1987 pp. 840-857
Greenleaf, Andrew and Sivaram Kalyandrug
"Trauma Registry System Using the SAS System and Dbase III
Plus" NESUG 89 Proceedings 989 pp. 243-250
Hofacker, Charles and Paul Hoffman
"How will PC SAS And Organizations Come to Grips with Each
Other?" SUGI 11 Proceedings 1986 pp. 614-6
Hudson, Michael and Roberta McBride
"SAS,SAS-ETS, and CLIST: A User-Oriented Approach to Management
Reporting" SUGI 7 Proceedings 1982 pp. 505-507
James, Harriet et al. "Techniques
Used for Converting An On-Line Production System From a Prior
Release of SAS Software to Version 5.0.8 of SAS Software SUGI
11 Proceedings 1986 pp. 453-464
Johnson, Bernadette "The SAS
System Under UNIX and Derivative Operating Systems : New Features,
New Power" SUGI 14 Proceedings 1989 pp. 1637-1646
Khory, Philip "Converting CLINFO Dataset Sets Into SAS
Datasets" SUGI
10 Proceedings, 1985,
pp. 679-683
Landon, Robert "DBASE III to
SAS PC DOS - A File Management System" SUGI 11 Proceedings
1986 pp. 623-9
Lahr, Robert "SIR on SSHENET:
Developing a Common Student Information Database for 14 Universities
and Sharing It Under CMS" SUGI 14 Proceedings 1989
pp. 888-892
LaRue, Dennis W. "Techniques
for Creating Special TSO/CMS Commands that Generate and Execute
SAS Programs" SUGI 7 Proceedings 1982 pp. 495-500
LeBlanc, William, Robert Syren, Gonzalo Garriga
"On-line Access to the SAS
Sample Library, Usage Notes, and Tutorials in VM/CMS" SUGI
13 Proceedings 1988 pp. 1423-1427
Lee, Han-li "Using REXX to
Build a Front-end User Interface" NESUG 90 Proceedings
1990 pp. 220-224
Martin, Thomas "Tricks and Tips for Supporting the SAS
System Under VMS" SUGI
14 Proceedings 1989 pp. 507-512
Painter, Sally "The SAS System
Under UNIX Operating Systems and Derivatives" NESUG 90
Proceedings 1990 pp. 203-205
Pan, Rong Q. "Use of the SAS
System, CMS REXX, and ISPF to Generate Standardized Vital Signs
Reports for Clinical Studies" SUGI 13 Proceedings
1988 pp. 881-886
SAS Institute Master Index to
SUGI Proceedings 1989 edition
Sauer J. and Allee M. "Development
of a SAS Based Interactive System Used to Analyze Pharmaceutical
Production Data " SUGI 11 Proceedings 1986 pp. 273-278
Siu, Maria and LuAnn Johnson "Customized Database Management System for Clinical Research Data Using SAS/FSP, IBM DMS Panels, and IBM VM/SP REXX"
SUGI 14 Proceedings,
1989 pp. 1087-1092
Strekel, Giles "Waiting for
Version 6" NESUG 89 Proceedings 1989 pp. 87-90
Suchoski, Pamela and Dan Magnuson "SPARES - An Interactive Estimated System Using SAS Software, CLISTSs, ISPF, and COBOL. SUGI 11 Proceedings
1986 pp. 303-309
Tilanus, Erik, Backer and Esselman
"Pronto - A System for Maintenance of Duty Rosters of Shift
Rosters" SUGI 7 Proceedings 1982 pp. 405-409
Timmins, Pat and Larry Locke "
SAS Panels -- Using SAS with the new SPF's Dialog Manager"
SUGI 7 Proceedings pp. 466-471
Vye, Christian and Craig Dickstein
"SAS Macros and SAS/AF Menu System for reading Compustat
Files" NESUG 89 Proceedings 1989 pp. 217-221
Vizard, John "DCL Standards
for SAS Applications Running on VMS Host Computers" NESUG
90 Proceedings 1990 pp. 210-215
Whitley, Chris "Interfacing
REXX and SAS: Using REXX to Create Front-End Processors for SAS
Programs" NESUG 91 Proceedings 1991
Young, James "Strategies for
Shrinking PC Data Sets and Finding More Free Bytes" SUGI
13 Proceedings 1988 pp. 613-8
Zeigler,Carl et al."SAS Software for the UNIX Environment" SUGI 13 Proceedings 1988 pp. 46-50