Digital PDFs

AA-MO63A-TB

June 1982

159 pages

Original

9.3MB

Document:	AA-M063A-TB SORT-MERGE Users Guide Feb82
Order Number:	AA-MO63A-TB
Revision:	0
Pages:	159
Original Filename:	AA-M063A-TB_SORT-MERGE_Users_Guide_Feb82.pdf

OCR Text

TOPS-10
SORT/MERGE
User's Guide
AA-M063A-TB

February 1982

This document describes user procedures for the
SORT/MERGE stand-alone utility program.
This document supersedes the SORT/MERGE User's Guide,
Order No. AA-0997D-TB, published May 1978.
OPERATING SYSTEM:
SOFTWARE:

TOPS-10 V7.01
SORT/MERGE V04C

Software and manuals should be ordered by title and order number. In the United States. send orders
to the nearest distribution center. Outside the United States. orders should be directed to the nearest
DIGITAL Field Sales Office or representative.

Northeast/Mid-Atlantic Region

Central Region

Western Region

Digital Equipment Corporation
PO Box CS2008
Nashua, New Hampshire 03061
Telephone:(603)884-6660

Digital Equipment Corporation
Digital Equipment Corporation
Accessories and Supplies Center Accessories and Supplies Center
1050 East Remington Road
632 Caribbean Drive
Schaumburg. Illinois 60195
Sunnyvale. California 94086
Telephone :(312)640-5612
Telephone:( 408) 734-4915

digital equipment corporation. marlboro. massachusetts

First Printing .. June 1977
Revised,
May 1978
Revised,
February 1982
Copyright ,c:, 1977, 1978, 1982, Digital Equipment Corporation. All Rights Reserved.

The information in this document is subject to change without notice and should
not be construed as a commitment by Digital Equipment Corporation. Digital
Equipment Corporation assumes no responsibility for any errors that may
appear in this document.
The software described in this document is furnished under a license and may
only be used or copied in accordance with the terms of such license.
No responsibility is assumed for the use or reliability of software on equipment
that is not supplied by DIGITAL or its affiliated companies.
The following are trademarks of Digital Equipment Corporation:
DEC
DECUS
DECSYSTEM-20
DECwriter
DIBOL
EduSystem

~DmDDmD

DECnet
DECsystem-10
PDT
RSTS
RSX
VMS
RT

lAS
MASSBUS
PDP
UNIBUS
VAX
VT

The postage-prepaid READER'S COMMENTS form on the last page of this
document requests the user's critical evaluation to assist us in preparing future
documentation.

Contents

Preface
Chapter 1

Getting Started with SORT/MERGE
1.1
1.2
1.3

1.4
1.5
1.6
1.7
1.8
1.9

Chapter 2

Creating a Multifield ASCII Test File . . . . . . . . . . .
Determining Record Length, Key Position, and Key Length
Sorting a Multifield File. . . . . . . . . .

1-1
1-2
1-3

1.3.1
1.3.2
1.3.3
1.3.4

1-3

Sorting the File on the Name Field.
Sorting the File on the Age Field. .
Sorting the File on the Grade Field
Sorting the File on Two Keys

Sorting a Line-Sequenced ASCII File . . .
Sorting Files Containing Tabs . . . . . . .
Sorting an ASCII File Containing Variable-Length Records
Sorting a Multiline ASCII File. . . . .
Sorting Uppercase and Lowercase Text.
Sorting Nontext Files . . . . . . . . .

1-5
1-5
1-6
1-7
1-8

1-10
1-11
1-12
1-13

How to Use SORT/MERGE
2.1
2.2
2.3

Introduction. . . . . .
Running SORT/MERGE
Command Formats. . .
2.3.1
2.3.2
2.3.3

2.4
2.5

The SORT/MERGE Command Scanner.
Merging Files . . . . . . . . . . . . .
Using Command Files. . . . . . . . .

Using SORT/MERGE from a COBOL Program.
Using SORT/MERGE from a FORTRAN Program

2-1
2-2

2-4
2-5
2-7
2-7
2-8
2-8

III

Chapter 3

SORT/MERGE Switches
3.1

3.2

3.3

3.4

3.5

Required Switches.

3-2

3.1.1
3.1.2

IRECORD-SIZE Switch
IKEY Switch.

3-2
3-2

3.1.2.1
3.1.2.2
3.1.2.3
3.1.2.4
3.1.2.5
3.1.2.6

3-3
3-4
3-5
3-6
3-6
3-8

/KEY Switch Format.
Key Starting Position
Key Length
Key Collating Order
Key Data Type.
Key Sign Status

Recording Mode Switches

3-10

3.2.1
3.2.2
3.2.3
3.2.4

3-10
3-10
3-11
3-11

IASCII Switch.
ISIXBIT Switch
IEBCDIC Switch.
IBINARY Switch.

File Switches

3-11

3.3.1
3.3.2
3.3.3
3.3.4
3.3.5
3.3.6
3.3.7
3.3.8
3.3.9
3.3.10

3-11
3-12
3-12
3-12
3-14
3-15
3-15
3-15
3-16
3-16

I AFTER-ADV ANCING Switch.

IALIGN Switch
IBEFORE-ADV ANCING Switch.
IBLOCKED Switch.
IFIXED Switch.
IFORTRAN Switch.
INOCRLF Switch
IRANDOM Switch .
ISEQUENTIAL Switch.
IV ARIABLE Switch

Control Switches.

3-16

3.4.1
3.4.2
3.4.3
3.4.4
3.4.5
3.4.6
3.4.7
3.4.8
3.4.9
3.4.10
3.4.11

3-16
3-17
3-17
3-19
3-20
3-21
3-21
3-21
3-22
3-22
3-23

IBUFFER-PAGES Switch
ICHECK Switch .

ICOLLATE Switch.
IERROR-RETURN Switch.
IFATAL-ERROR-CODE Switch

ILEAVES Switch.
IMAX-TEMP-FILES Switch.
IPHYSICAL Switch
ISTATISTICS Switch.
ISUPPRESS-ERROR Switch.
ITEMPORARY -AREA .Switch

Tape Switches.

3-23

3.5.1
3.5.2
3.5.3
3.5.4
3.5.5
3.5.6
3.5.7
3.5.8

3-23
3-23
3-24
3-24
3-25
3-25
3-26
3-26

I ANSI-ASCII Switch.

IDENSITY Switch.
IINDUSTRY -COMPATIBLE Switch
ILABEL Switch
IPARITY Switch.
IPOSITION Switch.
IREWIND Switch
IUNLOAD Switch

Chapter 4

File Formats
4.1

Recording Modes. . . . . . . . .

4-1

ASCII Recording Mode ..
SIXBIT Recording Mode.
EBCDIC Recording Mode
BINARY Recording Mode

4-2
4-2
4-2
4-3

4.1.1
4.1.2
4.1.3
4.1.4
4.2

File Formats. . . . . . . . .

4· ·3

Fixed-Length. ASCII.

4-5

4.2.1

COBOL Fixed-Length ASCII.
FORTRAN Fixed-Length (Random) ASCII

4 ·5
4-6

Variable-Length ASCII . . . . . . . . . . . . . .

4-8

4.2.1.1
4.2.1.2
4.2.2

4.2.2.1
4.2.2.2
4.3

4-12

COBOL Fixed-Length SIXBIT .

4-13

Variable-Length SIXBIT . . . . . . .

4-14

COBOL Variable-Length SIXBIT

4-15

EBCDIC File Formats. . . . . . . '. . .

4-16

4.4.1
4.5

4.5.1
4.5.2
4.5.3
4.5.4
4.6

COBOL Fixed-Length EBCDIC .
COBOL Variable-Length EBCDIC.
COBOL Blocked Fixed-Length EBCDIC
COBOL Blocked Variable-Length EBCDIC.

4-25

COBOL Binary File Formats.

4-25

COBOL ASCII Mixed-Mode Binary.
COBOL SIXBIT Mixed-Mode Binary.
COBOL EBCDIC Mixed-Mode Binary

4-25
4-27
4-28

FORTRAN Binary File Formats . . . . . . . .

4-30

4.6.1.1
4.6.1.2
4.6.1.3
4.6.2

4.6.2.1
4.6.2.2
4.6.2.3
4.6.2.4

FORTRAN Random Binary (with LSCWs)
FORTRAN Random Binary (without LSCWs).
FORTRAN Sequential Binary (with LSCWs) .
FORTRAN Sequential Binary (without LSCWs)

4-31
4-33
4-35
4-37

SORT/MERGE Error Messages
5.1
5.2

Chapter 6

4-17
4 -18
4-20
4-21

BINARY File Formats. . . . . . . .
4.6.1

Chapter 5

4--9
4-10

Fixed-Length SIXBIT . . . . . . . . .
4.3.1

4.4

COBOL Variable-Length ASCII . . . . .
FORTRAN Variable-Length (Sequential) ASCII

Message Format.
Error Messages .

5-1
5-1

SORT/MERGE Performance Considerations
6.1

Performance Overview ..

6-1

The Sort Phase .
The Merge Phase

6-2
6-3

6.1.1
6.1.2

6.2

Performance Considerations . .
6.2.1
6.2.2
6.2.3

Tree Size . . . . . . .
Number of I/O Buffers .
Number of Merge Passes.

6-9
6-9
6-10
6-11

Appendix A Summary of SORT/MERGE Commands and Switches
A.l
A.2
A.3
A.4

Commands...........
Switches (in Alphabetic Order).
Switches (by Function).
Switches (by Type) . . . . . . .

· A-I
· A-2
· A-I0
· A-II

Appendix B Collating Sequences and Conversion Tables
Glossary
Index
Figures
Key Argument Tree . . . . . . . . . . . .
COBOL Fixed-Length ASCII . . . . . . .
FORTRAN Fixed-Length (Random) ASCII.
COBOL Variable-Length ASCII . . . . . .
4-4 FORTRAN Variable-Length (Sequentiall ASCII.
4-5 COBOL Fixed-Length SIXBIT . .
4-6 COBOL Variable-Length SIXBIT .
4-7 COBOL Fixed-Length EBCDIC . .
4-8 COBOL Variable-Length EBCDIC.
4-9 COBOL Blocked Fixed-Length EBCDIC
4-10 COBOL Blocked Variable-Length EBCDIC
4-11 COBOL Standard Binary and ASCII Mixed-Mode Binary
4-12 COBOL Standard Binary and SIX BIT Mixed-Mode Binary.
4-13 COBOL Standard Binary and EBCDIC Mixed-Mode Binary
4-14 FORTRAN Standard and Mixed-Mode Random Binary with LSCWs
4-15 FORTRAN Standard and Mixed-Mode Random Binary. . . . . .
4-16 FORTRAN Standard/Mixed-Mode Sequential Binary with LSCWs
4-17 FORTRAN Standard and Mixed-Mode Sequential Binary
6-1 Binary Tree . . . . . . . . . . . . . . . . . . .
6-2 The Operation of the SORT/MERGE Binary Tree. . . . .

3-1
4-1
4-2
4-3

3-4
4-6

4-7
4-9
4-11
4-13
4-16

4-17
4-19
4-21
4-24
4-26

4-27
4-29
4-32
4-34
4-36

4-38
6-2
6-4

Tables
3-1
3-2
B-1
B-2
B-3

Field Descriptors for COBOL. . . . . . . . . . . . . . . . . . . . · 3-7
Field Descriptors for FORTRAN . . . . . . . . . . . . . . . . . . · 3-8
ASCII and SIXBIT Collating Sequence and Conversion to EBCDIC. · B-2
ASCII to SIXBIT Conversion. . . . . . . . . . . . .
· B-4
EBCDIC Collating Sequence and Conversion to ASCII . . . . . . . · B-6

Preface

This manual is written for those who want to sort data using the
SORT/MERGE utility program. The intended audience is composed of programmers and/or persons having user experience with a computer system.
As a user's guide, this manual is a combination of reference and explanatory material that should be sufficient for anyone with a reasonable
amount of experience to successfully sort any of the file formats supported
by SORT/MERGE. This experience should include some programming experience in a high-level language or simply user experience with a computer system. Specifically, you should be knowledgeable of the various
internal representations of data and with the file format of the data you
intend to sort.
However, if you are a beginning user who has never used a sorting utility,
Chapter 1 has been written to help you get started with the SORT/MERGE
utility. This chapter is devoted to sorting ASCII text files, and includes
some transitional material as a bridge into more complex types of sorts.
Where applicable, shaded areas in the examples show changes in the
SORT/MERGE command strings.
If there is any term used in this manual that you are unfamiliar with, refer
to the Glossary at the end of this manual for a definition of the term.
The software required to run SORT/MERGE is MACRO version 53A or
later and LINK version 4A or later. This manual reflects version 4C of the
SORT/MERGE software.

vii

References are made in this manual to the latest editions of the following
documents:
• TOPS-10ITOPS-20 COBOL-68 Language Reference Manual
• TOPS-10ITOPS-20 COBOL-74 LanguaRe Reference Manual
• TOPS-10 Monitor Calls Manual
• TOPS-10ITOPS-20 FORTRAN LanRuaRe Manual
• TOPS-10 Operating System Conlmands Manual

viii

Chapter 1
Getting Started with SORT/MERGE

This chapter is written for the beginning SORT/MERGE user who has a
simple ASCII or line-sequenced ASCII text file to sort. It demonstrates how
to create an ASCII file using SOS, how to calculate the position and length
of the key field, and how to use SORT/MERGE to sort the file.

1.1

Creating a Multifield ASCII Test File
To create and use a practice file, log in and create the sample file below.
(The examples use the SOS editor.) The file is a student record file consisting of last name, age, and final grade. It is 17 characters long and has the
following format:
name field age field
grade field -

position 1 to 10
positions 14 and 15
position 17

The following illustrates the format of each entry in the file:
100 00 00 00 01 :I. :I. 11 :1.:1.1
.( ?B 90 :1.2 34 ::,:;6/
,)\.)
12 34 '"'

NN NN NN NN NN
name field

At)
age field

(":J
grade field
MR·S1712·81

1-1

The following example illustrates creating the file with SOS. Each field of
the record is separated from the other fields by blanks (not tabs).
• SOS TEST. I NCREl}
Input: TEST.IN
00100
12345678901234567fRU)
00200
SMITH
19 efRU)
00300
JONES
32 Miilll
00400
ABERNATHY
19 B(RET)
00500
BROWN I NG
20 e(RET)
00600
HART
19 AfRU)
00700
WILSON
20 BfRU)
00800
HILL
20 AfRU)
00900
$

rEb
*D100(RE]
1 Lines (00100/1) deleted
* E S(RE'D
[TEST.INJ

NOTE
Line 00100 is temporarily used as a line of numbers to help
format the file.
Before the file is saved, the line of formatting numbers is deleted (line
00100). Also, the ES command, in SOS, is used to save the file without line
numbers. The saved version of the file looks like this:
• T Y PET EST. I Nru
SMITH
JONES
ABERNATHY
13ROWN I NG
HART
WILSON
HILL

19 e
32 A
19 B
20 e
19 A
20 B
20 A

1.2 Determining Record Length, Key Position, and Key Length
Now that the file is created, you must determine the following:
1.

The record length.

Whether the records are fixed-length or variable-length.

The starting position and length of the key field. (The key field is that
part of the record you want to sort on.)

Each line in the sample file is 17 characters long. Because each line is a
separate record, the record length is 1 7 characters long also. You do not
count line terminators, such as RETURN, as part of the record length.
Because each record is exactly 17 characters long, the records are
fixed-length.

1-2

Getting Started with SORT/MERGE

The starting position and length of the key field depends on which field of
the record you wish to sort (name, age, or grade). If you want to sort on the
name field, the key starts at position 1 and is 10 characters long:
OO~)~OOO~11111111
·L• ,.)
3· 4 I::·
( ·7 <:' <r' .) 1 ... ) ··r j\ I~· I: .. ~
,,_
..J (.)
\.>. (
•...• :I·'? '"' ,.) .'

NNNNNNNNNN

AA G

key field
MR·S·1713·81

If you desire to sort on the age field, the key starts at character position 14
and is 2 characters long:
00000000011111111
1234::..16?U90 1. :::.~34::.:jb/

NNNNNNNNNN

AA G
L-J
key field
MR·S·1714·81

If you wish to sort on the grade field, the key starts at character position 17
and is 1 character long:
OOO~OOOO)11111111

12345678901234567

NNNNNNNNNN
U
key field
MR·S-1715-81

1.3 Sorting a Multifield File
1.3.1

Sorting the File on the Name Field

Run SORT/MERGE by giving the command R SORT at monitor level:
.R SORTiBTI!

NOTE
Those areas of the command string that change in the following examples are shaded.
When SORT/MERGE responds with the prompt ~*', type the output file
specification:
*TEST;OUT

Getting Started with SORT /MERGE

1-3

Type an equals sign (=):
*TEST,OUT=

Type the input file specification:
*TEST,OUT=TEST,IN

Type the /FIXED switch to indicate that you are sorting fixed-length
records:
*TEST,OUT=TEST,IN/FIXED

Type the /RECORD switch and give the record length of 17:
*TEST,OUT=TEST,IN/FIXED/RECORD:17

Type the /KEY switch, then type the starting position of the key as 1, a
colon (:), and the length of the key as 10:
*TEST,OUT=TEST,IN/FIXED/RECORD:17/KEY:I:10

Type a colon and indicate if the file is to be sorted in ASCENDING or
DESCENDING order:
* TEST, OUT = TEST, IN 1 F I>(ED 1 RECORD: 171 KEY : 1 : 10: ASCEND I NG@

Type a carriage return
and wait for SORT/MERGE to perform the sort.
·When the sort is complete, SORT/MERGE responds with the following
display:
[SRT){PN Expanding to L1GP]
Gorted 7 records
12 KEY COtTIParisons, 1.71 per record
I~O record leaves in tTletTIOn'
o run s
0:00:00 CPU titTle, IG,L13 MS per record
O:OO:L12 elapsed titTle

'fhe sorted file, TEST. OUT appears as:
• TY PE TEST, OUTru
ABERNATHY
BROWN I NG
HART
HILL
JONES
SMITH
lAlILSON

1-4

19 B
20 C
19 A
20 A
32 A
19 C
20 B

Getting Started with SORT /MERGE

To sort the file in descending order, enter the following command string to
SORT/MERGE:
*TEST t OUT=TEST t IN/FI){ED/RECORD: 17 IKEY: 1: 10 :OESC!;NOING@]

The sorted file appears as:
+ TY PE

TEST + OUT@]

WILSON
SMITH
JONES
HILL
HART
BROWNING
ABERNATHY

20 B
19 C
32 A
20 A
19 A
20 C
19 B

1.3.2 Sorting the File on the Age Field
To sort the file on the age field, simple change the /KEY: arguments in the
command string to indicate that field, for example,
*TEST tOUT = TEST tIN I F I ){EDI RECORD: 17 I K EY ::{:~1:?: ASCEND I NGffil

The sorted file appears as:
tTY PE TEST t OUTffil
SMITH
ABERNATHY
HART
BROWNING
WILSON
HILL
JONES

19 C
19 B
18 A
20 C
20 B
20 A
32 A

1.3.3 Sorting the File on the Grade Field
To sort the file on the grade field, simply change the IKEY: arguments in
the command string to indicate that field, for example,
*TEST + OUT= TEST t I N/F U{ED/RECORD: 17 IKEY: 17Ft: ASCEND I NGffil

The sorted file appears as:
tTY PE TEST + OUT@]
JONES
HART
HILL
ABERNATHY
WILSON
SMITH
BROWNING

32 A
19 A
20 A
19 B
20 B
19 C
20 C

Getting Started with SORT /MERGE

1-5

1.3.4 Sorting the File on Two Keys
The previous examples demonstrated how to sort the file on anyone of the
fields in the record by changing the /KEY: switch arguments. It is often
necessary to sort a file on more than one field. For example, you might
want to sort your test file by age and also have each of the age groups
sorted by grade. To do this, type the following command string to
SORT/MERGE:
*TEST. OUT =TEST. IN / F 1>(ED / RECORD: 1 7/K EY: ILl: Z :ASCE:NDING
# / KEY: 1 7 : 1 : AS C END I N G(Rff)

-fREJl

NOTE
To continue a command string line, you must type a dash (-)
followed by a carriage-return/line-feed. SORT/MERGE then
prompts you with a pound-sign (# ). You can then continue
typing the command string as shown above.
The sorted file appears as: .
• TYPE TEST.OUT(RET)
HART
ABERNATHY
SMITH
HILL
WILSON
BROWNING
JONES

19
19
19
20
20
20
32

A
B
C
A
B
C
A

1.4 Sorting a Line-Sequenced ASCII File
In the first example, you saved TEST.IN without line numbers. The internal octal representation of the first record of TEST.IN would appears as:
Internal Representation-

-Bit35
123 115 r1:1.1
124 :1.:1.0 0
--·8· • M· • I • M'fu- ij..,1i r----

r-- - -

Data-

.-.-

040 040 -.-040 040 040 c -()- - n l l - M"
"

•

040 06:1. OJ:!. 0
-.040
-.- 040
.Ii "
-':L"- il9" r----

.-.-

'-.--

040 103 015 012 000
.(Ii -c,~

-':::-F- Niif.. r---MR-S-1716-81

1-6

Getting Started with SORT IMERGE

If the file is saved with line numbers, an additional six characters are
added to each record. These characters are the five ASCII digits of the line
number plus the horizontal tab. If TEST.lN has record-sequence numbers,
the octal representation of its first record appears as:
:1.
060 060 061 ()6() 06() r- ·0· ·0· 1-0:1. ..- Ii ().- A()"

r-- - -

0:1.1.
I-

12~3

1.1.5 1:1.1 124

()

ff1:- ~S· f-."'M. -1"- "y'ii r- - -

.. .

110 040 040 040 ()40 r- 0- -

li'1T·- • I i -.-II

•

r--H -

()40 ()40 061 r- 0- 040 040 --Ii -.- iIi"

C"--jj -11-

•

071 ()4() 103 015' 0:1.2

()

r.9 u- "H -'1(;" -(~f I . T r- - MR-S-1717-81

NOTE
Bit 35 is turned on in a line-sequence word. Thus, the
line-sequence sequence word shown above is not the same as
a data word containing the ASCII characters ~00100'.
The addition of the six characters increases the record length by six and
shifts the key starting position six places to the right. To sort a
line-sequenced version of TEST.lN on the name field, you must give the
following command:
*TEST • OUT =TEST. I NI F I )(EiItEC:URtf:'Z31KEY;1(fo: ASCEND I NG(@)

Thus, the IRECORD: switch argument and the IKEY: switch argument
were increased to include the length of the line sequence number.
The sorted file appears as:
.TYPE TEST.OUT(@)

00300

ooaoo
00500
00700
00200
00100

OOGOO

ABERNATHY
BROWNING
HART
HILL
JONES
SMITH
WILSON

19 B
20 C
19 A
20 A
32 A
19 C
20 B

Note that the line-sequence numbers are now out of order. To correct this,
you can edit the file (with SOS) and renumber it with the N command.
However, as soon as you open the file, SOS generates a warning message
for each line that is out of sequence. If you are working with a large file,

Getting Started with SORT IMERGE

1-7

you can have a tedious wait (while the warning messages are displayed)
before you can renumber the file. For this reason, it is recommended that
you sort ASCII files that are not line-sequenced and defer adding linesequence numbers until after the file is sorted.
,r

1.5 Sorting Files Containing Tabs
Tab characters are peculiar in that they are never printed or displayed on
an output device. Rather, they constitute an instruction to the output device (or the monitor if TTY NO TAB is set) to skip a certain number of
spaces before printing the next character. Tab characters can have the
following problems for you when using SORTiMERGE:
1.

While a single tab character can introduce as many as seven blank
spaces into a printed line, those seven "blanks" are not stored internally as a group of ASCII blank characters. (An ASCII blank is represented by octal 040, but a single ASCII tab character is represented by
octal all.) If you mistakenly interpret a single tab character to be the
number of spaces it produces on the output device, then you are calculating an incorrect record length and an erroneous starting position for
the key field. If the tab character falls within the key field, then you are
calculating an erroneous length for the key field.

Because tab characters are not printed, they are impossible to detect on
an ordinary printout or terminal display.

Tabs that are used randomly make it impossible to calculate a record
length or key field starting position that is accurate for each record in
the file. If tabs are used in the key fields of some records, but blanks are
used in the key fields of other records, then you are unable to sort the
file correctly.

For example, the following lines appear identicaL but the first line contains
blanks, while the second line contains tabs.

1-8

SMITH

Getting Started with SORT IMERGE

The octal representation of the first line appears as:
123 115 1:1.1 124 110

r-·S • -'i1.-

ilrii -·1''- -'f./.-

.-.-

04'0 040 040 ()4() 040
r-.- .f-.-.-.-._Ii

. -. .Ii .-.()40 04() 040 r()40
040
-Ii - . -

-.- .-

•

()40 062 064 -,-040 040
Ii
li'lH-41i
..
~:

-H-

040 :1.03

040 040 040
----- "C
r.-.II

ll
-

01~,i 012 000 000 O()O
r-CR -0::"" NUL -NlJL T~[n,:MR-S-1718-81

The octal representation of the second line appears as:
123 115 1:1.:1. :1.24

:1. :I. 0

r-..-c. "'M ilf· -"l'. IliT
II
-

\.')

H-

011 011 ()62 064 O:Ll
i=f( ").. --4 -j::f'(

-t:~T -

103

o1 ~j 012 000 000

.--·C· - -CR
-LJ.:' NUL..- -NUL..
'--

MR-S-1719-81

Although the two records (lines) shown above appear to be identical when
printed, their internal octal representation is very different. The record
length of the line using blanks is 25, but the record length of the line using
tabs is only 11. While the name field occurs in the sanle place in both
records, the age field and grade field do not.
The solution to the problem of tab characters is to avoid using them altogether or to use them only in a consistent manner. However, sometimes you
cannot know whether the file contains tab characters. A solution to this
problem is to use the PIP program to convert any tabs that may exist in
your file to the correct number of spaces that each tab represents. The
following example illustrates the commands you can give to PIP .
• R PI pm
*NOTAB.FIL=TEST.IN/wm

The IW switch causes PIP to convert the tabs in the file to the appropriate
number of spaces.

Getting Started with SORT IMERGE

1-9

1.6 Sorting an ASCII File Containing Variable-Length Records
The previous examples concentrated on files containing fixed-length records. SORT/MERGE can also sort variable-length records, providing you
specify a record length that is equal to the length of the largest record in
the file. For example, examine the following modified version of TEST.IN:
• TYPE TEST. I NID
1234567890123456
19 C SMITH
32 A JONES
19 B ABERNATHY
20 C BROWNING
19 A HART
20 B WILSON
20 A HILL

.Note how the record lengths vary from 11 characters to 16 characters. To
sort the above file on the name field, type the following command string to
SORT/MERGE:
nEST. OUT= TEST. I Nll,JAR I ABLE/RECORD: 1 6/KEY: 8: 9: ASCEND I NGID

The following notes apply to the sorting of variable-length records:
1.

If you specify a record length that is less than the maximum size record
in your file, then those records that are longer than the value you
specified are truncated on output.

Although the /VARIABLE switch is shown in the command string
above, you can omit it and still sort the file correctly. This is because
the /V ARIABLE switch is the default record-length type.

The /KEY switch is specified with the length of the longest field in the
record. Thus, for shorter fields, the key field extends beyond the end of
the field. This is permissable only if the data type switch is
/ALPHANUMERIC. (Data types are discussed in the last section of this
chapter.) However, you cannot specify a key field that extends beyond
the value given with the /RECORD switch.

The sorted file appears as:
• TYPE TEST. OUTID
1234567890123456
19 B ABERNATHY
20 C BROWNING
1.9 A HART
20 A HILL
32 A JONES
1.9 C SMITH
20 B WILSON

1-10

Getting Started with SORT /MERGE

1.7 Sorting a Multiline ASCII File
It is sometimes desirable to sort an ASCII text file in which each record
consists of two or more lines. For example, you may wish to sort a file of
records having the following format:
SMITH

19
C
JONES
32

A
Unfortunately, the control characters (carriage return/line feed) that are
used to format the text into lines are really ASCII record delimiters. If you
attempt to sort such a file, you will find that what you intended as fields of
a single record are interpreted as separate records by SORT/MERGE. The
results are not what you expected.
There are no elegant solutions to this problem. You could avoid using carriage returns until the actual end of your record, but you lose the ability to
format the record into separate lines. A better solution is to write a program that preprocesses your file before sorting and postprocesses your file
after sorting. In the preprocess phase, the program converts all carriage
return/line feeds, except the last one in the record, to some other character
pair that does not occur in the file. You sort the file, and then postprocess it
with the program to restore the carriage return/line feeds. The best solution
is to actually create the record with a program, rather than with a text
editor. This allows generating only one carriage return/line feed per record
and allows you to sort the records on any of the fields. However, as the
solutions to the problem grow more technically acceptable, you lose the
ease and informality of generating the file with a text editor.

1.8 Sorting Uppercase and Lowercase Text
The ASCII collating sequence (see Appendix B) is structured so that lowercase alphabetic characters have a greater value than their uppercase equivalents. As a result, if fields are composed of both uppercase and lowercase
characters and you sort your file according to the ASCII collating sequence,
then the file is sorted so that characters ~~A" through '~Z" come before characters "a" through ~~z", but all other characters have their normal position.
By using SORT/MERGE's alternate collating sequence facility, you can
specify a collating sequence in which uppercase and lowercase alphabetic

Getting Started with SORT /MERGE

1-11

characters are equivalent. This allows you to sort uppercase and lowercase
characters in normal alphabetic order. First, create a collating sequence
file of the form:
.505 UPPER.nn(BD)
Input: UPPER.TXT
00100
OOO-"@", "A"="a", "5"="b",
00200
" E " : " e", I. F " = " f" , "G" = " 9" ,
00300
"J":"j", "K"="f,", "L"="l",
OOLIOO
"0"="0'"
"P"="p", "0":"9'"
00500
"T"="t", "U":"u", "I.J"="!)",
00600
"y"="},,,, "Z"="z"(BIT)
00700
$

"C":"c",
"H" = " h" ,
"M"="!rl",
"R"=" r",
"1,..1"="1,1",

"D"=",j" ,(BIT)
"1" = " i " ., (BIT)
"N"="n" ,(BIT)
"5":"5" ,(BIT)
"){"="x" ,(BD)

~
[UPPER.TXTJ

When you give a command string, use the /COLLATE switch with the FILE
argument and the name of the file containing your alternate collating sequence. For example, the above collating sequence was stored in a file
named UPPER.TXT. To use the first command string shown in this chapter
for uppercase and lowercase text, specify the following:
*TE5T. OUT=TE5T. I N/F I ){ED/RECORD: 17 IKEY: 1 : 10: ASCEND I NG -(BIT)
#/COLLATE: FILE: UPPER. nn(BIT)

The result of the command string shown above is to sort file TEST.lN
according to the collating sequence stored in file UPPER.TXT. The results
are written to file TEST.OUT.

1.9 Sorting Nontext Files
This section is designed to help you extend your knowledge of more complex
types of sorts that SORT/MERGE can perform. If you are only interested in
sorting text files, then you do not need to read this section.
The following is the first command string shown in this chapter:
*TE5T.OUT:TE5T.IN/FIXED/RECORD:17/KEY:1:10:A5CENDING

For simplicity, you relied upon the default values for certain specifications
that SORT/MERGE requires. The complete, explicit command string
appears as:
* T EST. 0 UT = T EST. I NI AS C I I IF I:{ E DIRE COR D : 1 7 IKE 'I' : 1 : 1 0 : AS CEND I NG - (BIT)
#/ALPHANUMERIC(BD)

Two new items appear in the above command string:
1. /ASCII switch

2. /ALPHANUMERIC switch

1-12

Getting Started with SORT /MERGE

The /ALPHANUMERIC switch specifies the data type of the key field. The
data type specifies how the data is to be interpreted. For example, you read
in Section 1.5 that ASCII characters are represented internally by a 3-digit
octal number (or a 7-bit binary number). Because a single word of data on
TOPS-10 contains 36 bits, a word of data holds five ASCII characters, with
one bit unused (7*5 = 35). However, examined out of context, this word of
data appears to be simply a 36-bit binary number.
For example, consider the following 36-bit binary value:
011000101101010111001100001010000110
This value can be correctly interpreted as any of the following:
1.

26530857606

Fixed point decimal value

395671090000000000000.00

Floating point decimal value

3.9567109E + 20

Scientific notation

(159BC'

ASCII characters

Remember that any file you create is a series of these 36-bit words that are
subject to the various interpretations shown above. The interpretation of a
given word depends entirely on context. If you have created a text file, then
you are only interested in interpreting the data as characters. If you have
created a file of fixed-point decimal values, then you interpret the file
accordingly. You must specify the appropriate interpretation to
SORT/MERGE. If you specify an incorrect interpretation, then the results
of the sort, while technically correct, are not the results you expect.
((Data type" is the term used to refer to the interpretation of data; and the
data-type switch refers to the /KEY switch which is used to specify the
correct interpretation of data to SORT/MERGE. The switches for the simpler data types are:
1.

/ALPHANUMERIC

Alphabetic and numeric characters

/NUMERIC

Numeric characters

/COMPUTATIONAL

Fixed-point decimal numbers

/COMP1

Floating-point decimal numbers

/COMP3

Packed (4 bits) decimal

/PACKED

Same as /COMP3

See Section 3.2 for more information on data types.

Getting Started with SORT IMERGE

1-13

The second new switch, /ASCII, specifies the recording mode. The recording
modes that SORT/MERGE recognizes are:
1.

ASCII

7-bit bytes

SIXBIT

6-bit bytes

EBCDIC

9-bit bytes

BINARY

36-bi t bytes

All recording modes have a byte size associated with them. The byte size
specifies how many bits are used to represent a single unit of data in that
recording mode. For the first three recording modes, a byte is equivalent to
a character. (The first three recording modes have character sets associated
with them.) Thus, an ASCII character is represented by a group of 7 bits, a
SIXBIT character is represented by a group of 6 bits, and an EBCDIC
character is represented by a group of 9 bits. (See Appendix B for the
ASCII, SIXBIT, and EBCDIC character set.) A unit of data in binary recording mode requires all 36 bits of the word. Except for data produced by a
COBOL program, you can assume that any character data you generate is
ASCII data. COBOL programs generate SIXBIT data by default, unless you
have specified ASCII as the data type.
Refer to Section 4.1 for diagrams depicting the recording modes recognized
by SORT/MERGE. Also, see Chapter 3 for the discussion of each recording
mode switch.
There is also a new concept illustrated by the second command string. That
is the concept of file format, which is illustrated by the /ASCII /FIXED
switch combination. This particular file format is termed ~fixed-length
ASCII'. The file format specifies the type of control characters, control
words, or header words that are used to format the data into individual
records. Because these format words and control characters are not actually
part of the data, SORT/MERGE must strip them from the input, before the
actual sorting process can begin. (They are restored on output.) If you incorrectly specify the file format to SORT/MERGE, then the format words and
control characters are not properly stripped off and are sorted and output as
data. The results are quite unpredictable. (See Chapter 4 for a discussion of
file formats.)

1-14

Getting Started with SORT /MERGE

Chapter 2
How to Use SORT/MERGE

This chapter lists the information needed to sort a file and describes general command formats. Other topics, such as merging files and using command files, are also described. If you are a new user of SORT/MERGE, then
you should read Chapter 1 before you read this chapter.

2.1

Introduction
SORT/MERGE is a system program that arranges the records of one or
more files according to a user-specified sequence. You specify the key fields
on which the actual comparisons are to be made, and you also specify the
collating sequence to be used in ordering the file. The collating sequence
can be the normal ASCII, SIXBIT, or EBCDIC collating sequence; or it can
be a unique collating sequence which you have constructed for a particular
application. Additionally, you can specify that the file be sorted in ascending or descending order. SORT/MERGE performs two processes when sorting a file.
1.

The sort phase, where the input files are read and the records are sifted
into runs. These runs are output to one or more temporary areas. You
can specify up to 26 temporary areas.

The merge phase, where the runs are merged into fewer but longer
runs.

This process of sorting and merging continues until a single run remains;
this run becomes the output file.
Depending on file size, file order, and available memory, it is possible for
the runs to remain in memory. In such a case, temporary areas are not
used.

2-1

The time required for a given sort diminishes progressively as more disk
units and more memory are available. Although tape units can be used for
input and output files, disk units must be used for temporary areas. If the
input files are on disk, you must have enough temporary space to hold the
input files plus two additional copies of each input file.
The minimum memory required is an area large enough to contain the
following:
1.

Double buffers for
a. Each input device
b. The output device
c. One temporary disk file

At least 16 records

One record from each temporary file

2.2 Running SORT/MERGE
You must know the following information about the input files in order to
sort or merge them:
1.

Record size

Key size and location

Key collating order
a. Ascending
b. Descending

Key data type
a. Al ph an umeric
b. COMPI
c. COMP3
d. COMPUTATIONAL
e. FORMAT (FORTRAN ASCII floating point)
f.

Numeric

Key sign status
a. Algebraic sign used/not used for compares

2-2

How to Use SORT/MERGE

Recording mode
a. ASCII
b. BINARY
c. EBCDIC
d. SIXBIT

File attributes
a. Blocked
b. Fixed (random)
c. Variable (sequential)
d. FORTRAN (ASCII or binary)

In addition, there are control attributes and tape attributes that you can
specify. SORT/MERGE must have all this information in order to correctly
sort a file. However, you can rely on default switches and default key attributes, so that you need not specify all of the information required by
SORT/MERGE.
Although sorting a file can be a simple operation, several factors are often
introduced that make it difficult for an uninformed user to sort a file:
1.

The existence of control words in the file.
The use of control words is interwoven with the concept of file format,
and you may not understand one without understanding the other.
Chapter 4 contains diagrams of all major file formats supported by
SORT/MERGE, code segments from COBOL and FORTRAN programs
that generate these formats, and the SORT/MERGE switch or switch
combination used to specify these formats. You should study this
chapter until you know what switches to use in specifying your file's
format to SORT/MERGE.

The existence of line-sequence numbers in the file.
Chapter 1 describes the problems caused by line-sequence numbers.

3. The existence of tab characters in the file.
Chapter 1 describes the problems caused by tab characters in the file.
4.

U sing a key data type other than alphanumeric or numeric.
Calculating key position and key length for alphanumeric or numeric
keys is a relatively straightforward task. (See Chapter 1, if you do not
understand how to do this.) However, when using other key data types
such as COMPUTATIONAL or FORMAT, calculating key position and
length becomes more complex. If you are using a key data type other
than ALPHANUMERIC or NUMERIC, be sure to read the description

How to Use SORT/MERGE

2-3

of the /KEY switch in Chapter 3. You can also find the file format
diagram in Chapter 4 that describes your file, and study the key specifications in the diagram.
5. Sorting character data stored in a binary file.
If you specify your file as a standard binary file (lBINARY switch), then
you cannot use an alphanumeric or numeric field as a key field. If you
specify your file as a mixed-mode binary file (lBINARY/ASCII,
/BINARY/SIXBIT, or /BINARY/EBCDIC switches), then you can use an
alphanumeric or numeric field as a key field. However, the specification
of record length and key starting position changes from words for a
binary file to characters for a mixed-mode binary file. See Chapter 4 for
a more detailed discussion of standard and mixed-mode binary files.

2.3 Command Formats
There are five different functions that can be specified in a command string
to SORT/MERGE. The following describes each of these functions and illustrates the general format of the command string that performs the function.
1.

SORT -

sort the specified files.

switches OUTFIL switches = switches INFILI switches, switches
INFIL2 switches
2.

MERGE -

merge the specified files.

switches OUTFIL switches = switches INFILI switches, switches
INFIL2 switches /MERG E
3.

RUN - leave SORT/MERGE command level and run the specified
program.
/RUN (specified anywhere in the command string - program is started
when the sort or merge is complete.)

EXIT -

exit SORT/MERGE and return to monitor level.

/EXIT (specified anywhere in the command string - control returns to
monitor level when the sort or merge is complete.)
5.

HELP -

print the text of the help file.

/HELP (lHELP:SWITCHES prints available SORT/MERGE and SCAN
switches.)
SORT/MERGE uses SCAN, the TOPS-IO command scanner to parse the
command string. The effect of a switch in a command string depends on the
type of switch and its place in the command string. See Chapter 3 for more
details. Certain switches are not usable in FORTRAN-called sorts. These
are described in Section 2.3.4.

2-4

How to Use SORT/MERGE

File specifications are given in the standard form:
filnam.ext
Where ~filnam' is a 1- to 6-character filename, and ~ext' is a 1- to
3-character file extension. At least one input file specification and one
output file specification must.be given. Multiple file specifications are separated by commas. Note that multiple output file specifications are allowed
only if the output device is magnetic tape.
You can continue a command string onto a second line by typing a dash.
SORT/MERGE prompts the next line with ~#', and you can continue typing
the command string.

2.3.1

Merging Files

If you have two or more files that have been previously sorted on the same
keyes) and have the same file format, the files can be merged into one file
with the /MERGE switch. When merging files, you must specify the same
information as for a sort, plus the /MERGE switch, for example:
OUTFIL = INFIL1,INFIL2/RECORD:70/KEY:40:15:ASCENDING/MERGE

If you attempt to merge files that do not have the same file formats and are
not already in sorted order on the same key field(s), then you receive unpredictable results. To safeguard against this, you can specify the /CHECK
switch. This switch causes SORT/MERGE to check the files and generate
an error message if any out-of-sequence records are found. See Chapter 3
for more information on the /CHECK switch.

2.3.2 Using Command Files
A command file is an ASCII file with SORT/MERGE commands stored in it.
By specifying ~@filespec', you can cause SORT/MERGE to read a command
file and execute the commands stored in it. The following illustrates a
command file created with SOS. Note that SORT/MERGE accepts the file
with or without line-sequence numbers .
• 808 FILE.CCLOOl
Input: FILE.CCL
00100
80RT1.FIL=IN1.FIL/RECORD:80/KEY:1:10OOl
00200
80RT2.FIL=IN2.FIL/RECORD:80/KEY:l:10OOl
00300
MERGE.FIL=80RT1.FILt80RT2.FIL/RECORD:80/KEY:1:10/MERGEool
00400
I E}-( I TOOl
00500
$

•

[FILE.CCLJ

The following command causes SORT/MERGE to execute the command file
shown above:
*@FILE.CCLOOl

How to Use SORT/MERGE

2-5

If you do not specify a file extension, SORT/MERGE searches for a file
specification of the form: filnam.CCL.

2.4 Using SORT/MERGE from a COBOL Program
COBOL users can run SORT/MERGE from a COBOL program by using the
COBOL SORT and MERGE statements. Because this statement is part of
the COBOL syntax, it is not described here. See the COBOL manual
(COBOL-68 and/or COBOL-74 Language Reference Manual) for information on the SORT or MERGE statement.

2.5 Using SORT/MERGE from a FORTRAN Program
FORTRAN users can call SORT/MERGE from a FORTRAN program by
using a CALL statement, for example:
CALL SORT('OUTPUT.FIL = INPUT.FILIRECORD:50IKEY: 1:5')

Note that the actual SORT/MERGE command string can be enclosed in
single quotation marks, or passed directly as an array containing the text
of the command. The following example illustrates a FORTRAN subroutine
that accepts a SORT/MERGE command string typed from a time-sharing
terminal, writes the string into an array, and passes the entire array to
SORT/MERGE.

NOTE
When you pass an array to SORT/MERGE in a FORTRAN
program, the array must end with a null (for example,
CMDO(21) equivalence to CMD(20)).

10
1000
2000

DIMENSION CMDO(21)
DIMENSION CMD(20)
EQUIVALENCE(CMD,CMDO)
TYPE 1000
FORMAT (' *'$)
ACCEPT 2000,CMD
FORMAT (20A5)
CALL SORT(CMD)
GO TO 10
END

If your installation has built a FORTRAN VM SORT program (FSORT),
then the CALL SORT statement in the above FORTRAN program executt~3
FSORT. FSORT contains the same features as the stand-alone
SORT/MERGE program.

2-6

How to Use SORT/MERGE

However, if your installation has not built a FORTRAN VM SORT, then
the FORTRAN-called sort is somewhat smaller but does not have access to
all the features of stand-alone sorts. The following are the restrictions:
1.

All switches must be specified to uniqueness. This means that
IRECORD must be specified at least to IREC, since this version of sort
does not have the concept of a switch unique in any abbreviation. For
example, if you give the following switch from SORT/MERGE command
level:

IH:S

you receive a display of all the switches that can be specified to
SORT/MERGE. Any switch marked with an asterisk (*) is recognized
even if only one letter of the switch is typed. For instance, IRECORD
can be specified as IR. This version of sort does not have this facility and
generates an error message if you do not type enough characters of the
switch to make it unique.
2.

Switches must always follow file specifications, for example,
Iswitchllswitch2 TEST.OUT = Iswitch3/switch4 TEST.IN ...
is illegal, while:
TEST.OUT/switchllswitch2 = TEST.IN/switch3/switch4 ...
is legal.
There is no concept of modified switches.

Switches cannot be read from the SWITCH.INI file.

The following switches are not recognized:
./BLOCKED

• ICOMP3 (or IPACKED)

• ICOMPUTATIONAL
./EBCDIC

./SIXBIT
• all SCAN switches (except IDENSITY)
5.

There is no continuation line feature. If a command string is excessively long, it should be passed directly to a FORTRAN sort.

How to Use SORT/MERGE

2-7

There are two SORT/MERGE switches that are of special interest to
FORTRAN users:
1.

/ERROR:

/FATAL:

The first switch, /ERROR:, allows you to specify an error-condition GOTO
address. If SORT/MERGE encounters an error, control transfers to the
specified address in the FORTRAN program that called SORT/MERGE.
The second switch, /FATAL:, allows you to save a SORT/MERGE error code
in a specified address. See Chapter 3 for more details on these two switches.

2-8

How to Use SORT/MERGE

Chapter 3
SORT/MERGE Switches

This chapter gives detailed explanations of all SORT/MERGE switches and
certain SCAN switches that are useful to you. (For a quick-reference list of
SORT/MERGE switches, see Appendix A, SORT/MERGE Functions and
Switches.)
There are six categories of switches in SORT/MERGE:
1.

Required switches

Key data type switches

Recording-mode switches

File switches

Control switches

Tape switches

Each of these switch groups vanes according to the range of its effect,
which can be:
1.

Global - Affects the entire process. Global switches can be placed anywhere in the command string without changing their effect.

Local - Affects only the preceding file specification. Local switches can
also be placed in front of file specifications, but they still affect only
that file specification.

3-1

Position dependent - Depends on its position in the command string. If
placed before a file specification, a position dependent switch affects
that file and all following files until an equal sign or an end-of-line
character is encountered. Thus, the switch is position dependent over
the file specifications that follow it. If placed after a file specification, it
affects only that file specification. for example.
*/Switch-A OUT.FILISwitch-B = ISwitch-C FIRST.FIL/Switch-D -~EJI
# ISwi tch-E SECOND. FILISwi tch-F/Swi tch-G THIRD. FILISwi tch-HfRET)

Switches A and B apply to the output file (OUT.FILL Switch C applies
to all input files. Switch D applies only to the first input file
(FIRST.FILL Switch E applies to the second and third input files
(SECOND.FIL and THIRD.FILL Switch F applies only to the second
input file. Switches G and H apply to the third input file.
4.

Modified position dependent - Modified position dependent switches
work as follows (the IBLOCKED switch is used for example purposes):
a. If IBLOCKED is not specified anywhere in the command string,
then no files are blocked.
b. If IBLOCKED is specified anywhere in the command string, then
those files for which no explicit IBLOCKED switch is specified have
a blocking factor determined by default. The default blocking factor
is taken from the last position dependent IBLOCKED switch in the
command string. If there is no position dependent IBLOCKED
switch, then the first IBLOCKED switch in the command string is
used. Thus, in the following command,
OUT.FIL = IN1.FIL.IN2.FIL.IN3.FIL

all files are unblocked. In this command.
OUT.FIL = IN1.FIL.IN2.FIL.IN3.FILIBLOCKED:3

all files have a blocking factor of 3. In this command,
OUT.FIL = IN1.FIL.IN2.FIL'BLOCKED:2!BLOCKED:3 IN3.FIL.IN4.FIL

OUT.FIL, INl.FIL, IN3.FIL. and IN4.FIL have blocking factors of 3.
IN2.FIL has a blocking factor of 2.

NOTE
Key data type switches are in a special category and can only
be specified after the key switch they are intended to modify.

3-2

SORT/MERGE Switches

3.1

Required Switches
Every SORT/MERGE command string must contain the following two
switches: /RECORD and /KEY. Both are global and required. These
switches are described below.

3.1.1 IRECORD Switch
The /RECORD switch specifies the length of the record and has the
following format:
/RECORD:n
For all file formats except standard binary (described with the /BINARY
switch), n indicates the number of characters in the record. For a standard
binary file, n indicates the number of words in the record.
If the file contains any control words (such as SIXBIT and EBCDIC header
words or FORTRAN Logical Segment Control Words) or control characters
(such as ASCII carriage return/line feed), you should not include thesE:
control words in the record length. If you specify a value with the
/RECORD switch that is less than the actual size of the records you are
sorting, the records are truncated for variable-length records. For
fixed-length records, you receive an error message or undefined results. If
you are sorting variable-length records, you should specify the length of
the largest record in the file.

3.1.2 IKEY Switch
The /KEY switch describes the characteristics of the key field, the field on
which comparisons are made when the records in a file are sorted. Three
characteristics of the key field are:
1. key starting position
2.

key length

key collating order

NOTE
Chapter 4 contains diagrams of key fields and key specifications for all major file formats recognized by SORT/MERGE.
You should read that chapter after reading this section.

SORT/MERGE Switches

3-3

/KEY Switch Format -

3.1.2.1

The /KEY switch has the format:

/KEY:n:m I:xl
where:
n

is the position of the first character or word of the key. (If the key
begins with the first character or word of the record, then n = 1.)

is the length of the key.

is the key collating order, either ASCENDING or DESCENDING.

The key position and key length values must be specified. If you do not
specify a collating order, then it defaults to ASCENDING.
3.1.2.2 Key Starting Position - The key starting position indicates where in
the record the key field begins. This position is usually specified in characters. However, for COMPU and COMPI data types, there are file formats
that require the key starting position to be given in words, rather than
characters. The following illustrates when to use characters and when to
use words:

Data Type

Use

ALPHA
NUMERIC
FORMAT
COMP3
COMPU

Characters
Characters
Characters
Characters
Characters (mixed-mode binary)
words (standard binary)
Characters (mixed-mode binary)
words (standard binary)

COMPI

If the file is a mixed-mode binary file (a binary file containing characters)
and is specified to SORT/MERGE as a mixed-mode binary file (for example,
/BINARY/ASCII), then the key starting position is always calculated in
characters. This is true even for COMPU or COMPI keys. However, if a
mixed-mode binary file is specified to SORT/MERGE as being a standard
binary file (for example, /BINARY), then the key starting position for
COMPU or COMPI keys is specified in words; and the fields containing
characters cannot be used as key fields. See Chapter 4 for more details on
mixed-mode binary files.
If you are using two adjacent key fields such that some key field is preceded
by a COMP3 (or PACKED) key field, then use the following formula to
calculate the first character position in the second key field.
(n + I + 1)

+s
2

3-4

SORT/MERGE Switches

where:
n

is the number of decimal digits in the COMP3 field.

is the starting position of the COMP3 field.

When using this formula, round a fractional result to the next lower
integer.
The key length refers to the length of the key field.
Key length is specified either in characters or decimal digits, depending on
the data type:

3.1.2.3 Key Length -

Data Type

Use

ALPHA
NUMERIC
FORMAT
COMPU
COMPI
COMP3

Characters
Characters
Characters
Decimal digits
Decimal digits
Decimal digits

Key Collating Order You can specify either ASCENDING or
DESCENDING collating order for any file that you wish to sort. If the key
data type is ALPHA, then ASCENDING causes SORT/MERGE to sort the
file in direct agreement with the character set used in the file (ASCII,
SIXBIT, or EBCDIC). DESCENDING causes SORT/MERGE to sort the file
in reverse agreement with the character set used. If the key data type is
FORMAT, NUMERIC, COMPUTATIONAL, COMPl, or COMP3, then
ASCENDING causes SORT/MERGE to sort the file in ascending numeric
order. DESCENDING causes SORT/MERGE to sort the file in descending
numeric order.
3.1.2.4

3.1.2.5 Key Data Type - Data type refers to the type of data that makes up
the key field. All key data type switches modify the key switch that they
follow. The data type switch can be one of the following:

/ALPHANUMERIC (/ALPHA)
indicates that the key field is composed of alphabetic and numenc
characters.
/NUMERIC
indicates that the key field is composed of numeric characters. If the key
field contains an algebraic sign (+' or t_', or over-punched trailing sign),
and you wish to exclude the sign in the key comparisons, then specify the
/UNSIGNED switch also. Otherwise, the sign status defaults to /SIGNED.
This data type can be used only for data without a decimal point or with an
implied decimal point. The data cannot include the character t.'.

SORT/MERGE Switches

3-5

IFORMAT

indicates that the key field is in one of four FORTRAN ASCII numeric
formats. The IFORMAT switch has the following format:
IFORMAT:nPaw.d

where ~n' is a scaling factor, 'a' is a FORTRAN format type, 'w' is the width
of the key field, and ~d' is the number of decimal places in the number. The
~n' can be a positive or negative number. If the 'nP' is not specified, no
scaling is done. If the ~w.d' is not specified, FORTRAN free format is assumed. The ~a' can be one of the following arguments:
Argument

Format

D
E
F
G

Double-precision scientific format
Scientific format
Floating-point format
General format

NOTE
A key having a FORMAT data type always implies that the
recording mode is ASCII.
ICOMPUTATIONAL (/COMPU)
indicates that the key field is a fixed-point binary number. If the number is
10 decimal digits or less, then the key occupies one word of storage. If the
number is greater than 10 decimal digits, then the number occupies two
words of storage. This data type is signed by default; so if you wish to
exclude the sign from the key comparisons, you should also specify the
IUNSIGNED switch.
ICOMPI

indicates that the key field is a floating-point binary number. If the number is 10 decimal digits or less, then the key occupies one word of storage. If
the number is greater than 10 decimal digits, then the number occupies two
words of storage. This data type is signed by default; so if you wish to
exclude the sign from the key comparisons, you should also specify the
IUNSIGNED switch.
ICOMP3

(or IPACKED)

indicates that the key field is an IEBCDIC packed-decimal number. This
data type is signed by default; so if you wish to exclude the sign from the
key comparisons, you should also specify the IUNSIGNED switch.
The following two tables illustrate the various field descriptors for COBOL
and FORTRAN prograllls, and give the appropriate SORT/MERGE data
type for each program field descriptor:

3-6

SORT/MERGE Switches

Table 3-1:

Field Descriptors for COBOL

COBOL PHRASE

SORT/MERGE
KEY DATA TYPE

PIC A( )
PIC X( )
PIC 9( )
PIC S9( )
PIC 9( )COMPUTATIONAL
PIC 9( ) COMP-l
PIC 9( ) COMP-3

IALPHA
INUMERIC/UNSIGNED
INUMERIC/SIGNED
ICOMP
ICOMPI
ICOMP3

IALPHA

NOTE
Data types INUMERIC, ICOMP, ICOMP1, and ICOMP3 can
be further specified as ISIGNED or IUNSIGNED. COBOL
COMP items can be single precision (1-10 decimal digits,
occupying one word of storage) or double precision (11-18
decimal digits, occupying two words of storage).

Table 3-2:

Field Descriptors for FORTRAN

FORTRAN FORMAT
DESCRIPTOR

SORT/MERGE
KEY DATA TYPE

A
D
E
F
G
H
I
L
0
R

IALPHA

IFORMAT:Fw.d
IFORMAT:Gw.d
IALPHA
INUMERIC
IALPHA
IALPHA
IALPHA

FORTRAN
DATA TYPE

SORT/MERGE
KEY DATA TYPE

INTEGER
REAL

ICOMPU
ICOMPl

/FORMAT:Dw.d
/FORMAT:Ew.d

NOTE
Data types INUMERIC, ICOMPU, ICOMP1, and IFORMAT
can be further specified as ISIGNED or IUNSIGNED. If the
specified length of a COMPU or COMP1 key is 10 decimal
digits or less, SORT/MERGE interprets the key as one word
long. If the specified length is 11 decimal digits or more,
SORT/MERGE interprets the key as two words long.

SORT/MERGE Switches

3-7

The key sign status indicates whether or not the
algebraic sign of the key field is to be used in making key comparisons. Key
fields can incorporate the algebraic sign in the following ways:
3.1.2.6 Key Sign Status -

Key Data Type

Sign Representation

NUMERIC
COMPU
COMPI
COMP3

Characters ~ +' or ~-', or over-punched character
Sign bit (bit 0)
Sign bit (bit 0)
4-bit trailing sign bit

Specifying /SIGNED causes SORT/MERGE to use the algebraic sign in
making key comparisons. Specifying /UNSIGNED causes SORT/MERGE to
ignore the algebraic sign in making key comparisons.

WARNING
If you specify the /SIGNED switch, then the method of sign
representation used in your key field can affect the calculation of key starting position and length. Note the effect of
various ·sign representations on the key field:
NUMERIC data type
1.

If the key field is written with a COBOL field descriptor of the form:
PICTURE S99, then the sign is represented by an overpunch on the last
numeric character in the field. For example, -521 would be represented
by 52J. This type of sign representation does not use an extra character
position to represent the sign, and the key field is not affected.

If the key field contains l-123' (where the entire field consists of characters), and you specify the /SIGNED switch, then the sign occupies a
character position and is part of the key. As the sign character precedes
the key, the key starting position begins with the sign character, not
with the first numeric character in the field. Therefore, the key length
must include the sign character.

FORMAT data type
If the key field was written in any FORTRAN scientific notation format,
then the sign is represented by the character ~ +' or l_'. For example, in the
following data item: -0.51E + 03, the sign (as well as the period and the
letter ~E', the exponent's sign, and the exponent) occupy character positions
within the key and are considered part of the key. This is true for either
/SIGNED or /UNSIGNED sign status.
COMPUTATIONAL and COMPI data types
The sign is represented by a single bit for these data types, and the key
field is not affected by sign status.

3-8

SORT/MERGE Switches

COMP3 data type

(or PACKED)

In EBCDIC packed-decimal notation, the sign is represented by a 4-bit
pattern in the last digit position of the field. The length of the sign representation is never added to the key length. The key field is not affected by
the sign status. However, the sign representation can affect a key field
position following the COMP3 key field.
To determine the length of a COMP3 (or PACKED) data type key field in
9-bit bytes, add the length (L) of the field in digits plus two divided by two,
truncating any remainder. Thus, a nine digit key field equals a key field
length of five bytes.
L + 2 / 2 = key length field
9 + 2 / 2 = 5 bytes

3.2 Recording Mode Switches
Recording-mode switches designate the coding scheme used in the data
files that you are sorting. There are four recording-mode switches in
SORT/MERGE, and all are global:
1.

/ASCn

/EBCDIC

/SIXBIT

/BINARY

The first three switches (fASCn, IEBCDIC. and ISIXBIT) have collating
sequences associated with them. and only one of these three switches can be
specified in a command string.

3.2.1

/ ASCII Switch

The /ASCn switch indicates that the file is recorded in ASCn mode. This
switch has the following format:

/Ascn
Characteristics:
1.

Defaults to IV ARIABLE, though iFIXED can be specified instead. If
IASCn/BINARY is specified. then the default is IFIXED. and
/VARIABLE can only be given with lABell/BINARY/FORTRAN.

The ICOMP3 (or /PACKED) switch cannot be given with the iKEY
switch when the recording-mode is ASCII.

SORT/MERGE Switches

3-9

3.2.2 ISIXBIT Switch
The ISIXBIT switch indicates that the fi Ie h n'corded in SIXBIT mode. This
switch has the following format:
ISIXBIT
Characteristics:

Defaults to IV ARIABLE, although FIXED can be specified. If
ISIXBIT/BINARY is specified, FIXED is the default. and VARIABLE
can only be specified with /SIXBITBINARY· FORTRAN.

The ICOMP3 (or IPACKED) and FORMAT data type switches cannot
be specified with the ISIXBIT switch.

3.2.3 IEBCDIC Switch
The IEBCDIC switch indicates that thp fi Ie is recorded in EBCDIC mode.
This switch has the following format:
IEBCDIC

Characteristics:
1.

Defaults to IFIXED, though VARIABLE can be specified. If
IEBCDIC/BINARY is specified. then VARIABLE can only be specified
with IEBCDIC/BINARY/FORTRAN.

The IFORMAT data type switch cannot be specified with the EBCDIC
switch.

3.2.4 IBINARY Switch
The IBINARY switch indicates that the file is recorded in binary mode.
This switch has the following format:
IBINARY

Characteristics:

I
3-10

Defaults to IFIXED. The ,VARIABLE switch is allowed only for
IBINARY/FORTRAN.

If the ICOMP3 (or IPACKED) switch is specified with the !KEY switch,
then the default is IBINARYiEBCDIC. and ASCII or /SIXBIT cannot
be specified.

If the ICOMPU or ICOMPI switch is specified with the IKEY switch,
then the default is IBINARY/SIXBIT. although IASCII or IEBCDIC can
be specified to override this and the key size is specified in characters,
not words.

If the IFORMAT switch is specified with the IKEY switch, then the
recording mode is IBINARYiASCII. and neither ISIXBIT nor IEBCDIC
can be specified.

SORT/MERGE Switches

3.3 File Switches
File switches are used to specify various file attributes such as word alignment, blocked files, and file type (fixed or variable). All file switches are
global, except the IBLOCKED switch, which is a modified position dependent switch.

3.3.1

I AFTER Switch

The IAFTER switch indicates where the output record is written in relation
to the carriage-return/line-feed characters. This switch has the following
format:
IAFTER

Characteristics:
1.

The IAFTER switch specifies to place the output record after a
carriage-return/line-feed.

This switch is designed to be compatible with COBOL-74 programs.

The IAFTER switch is a local switch.

The I AFTER switch is not required for input files.

3.3.2 IALIGN Switch
The IALIGN switch indicates that each output record begins on a word
boundary. This switch has the following format:
IALIGN
Characteristics:
1.

Your file must be recorded in ASCII mode.

This is a global switch, but affects only the output file.

This switch increases the rate of data transfer, but produces a somewhat larger output file.

Line-sequenced ASCII files must be word aligned and are output in
word-aligned format whether the IALIGN switch is specified or not.

3.3.3 IBEFORE Switch
The IBEFORE switch indicates where the output record is written in relation to the carriage-return/line-feed characters. This switch has the
following format:
IBEFORE

SORT/MERGE Switches

3-11

Characteristics:
1.

The /BEFORE switch places the output record before the carriagereturn/line-feed.

This switch is the default to SORT/MERGE.

The /BEFORE switch is a local switch.

The /BEFORE switch is not required for input files.

3.3.4 IBLOCKED Switch
The /BLOCKED switch indicates the blocking factor to be used for blocked
files. This is a modified position dependent switch and has the following
format:
/BLOCKED:n
where:
n

is a decimal number indicating the number of logical records per logical block.

Characteristics:
On tape, each physical record is separated from the next record by an
inter-record-gap (lRG), which is a space containing no data. As each IRG is
approximately 95 words long, a large amount of tape storage is wasted if
there is an excessive number of IRGs. This occurs if the length of a physical
tape record is not appreciably larger than the size of an IRG. The default
size of a physical tape record is set by the monitor at 200 (octal) words. You
can override the default value with the SET BLOCKSIZE monitor command (see the TOPS-10 Operating System Commands Manual). An alternative to relying on the system parameter for physical tape record size is to
specify the /BLOCKED switch to SORT/MERGE. The argument to this
switch specifies how many logical records are contained in a physical tape
record. The optimal range of physical tape record size is about 500 to 2000
words. By specifying a blocking factor that creates a physical record size
that falls within that range, you can substantially improve utilization of
tape storage space by reducing the number of IRGs. This is illustrated in
the following diagram:
Physical Records:
1
(128 words)
,------,

,----,

I i i

Rl I R2 I'RG L~~~48~~~J-;~--"RG I R71~ I'RG I R9 eEI~_E~EJ

R ~ logical record (62 words)

Physical Records:
(256 words)

~~L~~~'RG t~I_~~]RiI-~~ IIRGL~9E-~~~~~1~ii8
R = logical record (62 words)

3-12

SORT/MERGE Switches

MR-S-1698-81

Note that the second file uses half as many IRGs as the first file.

NOTE
EBCDIC magnetic tapes must be blocked. If no blocking
factor is specified for an EBCDIC magnetic tape, a blocking
factor of 1 is assumed.
Blocking files also causes somewhat faster access to disk. However, as the
size of the physical disk record is computed on the basis of disk blocks (128
words), the blocking factor must be chosen carefully; or you can cause an
excessive waste of disk storage space. Use the following formula to calculate physical record size:
(n + c1) * m + c2
where:
n
m
c1
c2

record length in words
blocking factor
record control words in a single record
block control words in a single block

Now compare the calculated value with the closest multiples of 128 (that is,
128, 256, 384, 512, and so on). If the calculated value is slightly less than or
exactly equal to a multiple of 128, then the blocking factor makes efficient
use of disk storage. If the calculated value does not meet the previous
conditions, then you should try increasing or decreasing the blocking factor
until the conditions are met. The following example demonstrates the effect
that selection of blocking factors has on disk storage utilization:
Record Length: 30
Blocking Factor: 5

Record Length: 30
Blocking Factor: 4

Wasted Space
1 Physical
Record
(1 block)

~---...-.. Disk Block Boundary

(128words)

2 Physical
Records
(2 blocks)

~1-"-~:.L.L.~..L..4f'--Disk Block Boundary

(128 words)

Wasted Space
Wasted Space
~~L..LJ..~"",- Disk Block Boundary

(128 words)
Physical Record Size: 256 words
Wasted Space: 106 words per block

ILL.<~~~~.... Disk Block Boundary

(128 words)
Physical Record Size: 128 words
Wasted Space: 8 words per block
MR-S-1699-81

3.3.5 IFIXEO Switch
The /FIXED switch indicates that the file contains fixed-length records.
This switch has the following format:
/FIXED

SORT/MERGE Switches

3-13

Characteristics:
This switch mayor may not be in effect by default. depending on the particular recording-mode switch being used. See the description of the appropriate recording-mode switch to determine when IFIXED is in effect by
default.

3.3.6 IFORTRAN Switch
The IFORTRAN switch indicates that the file is a FORTRAN file written in
either ASCII or binary recording mode. This switch has the following
format:
IFORTRAN

Characteristics:
1.

If you specify IFORTRANIASCIL then IALIGN is in effect by default.

2 .. If you specify IFORTRAN/BINARY, then SORT/MERGE checks for
Logical Segment Control Words (LSCWs). Refer to Chapter 4 for a description of LSCWs.

3.3.7 INOCRLF Switch
The INOCRLF switch (No Carriage Return/Line Feed switch) indicates that
both the input and the output files are IFIXED IASCII records containing
no carriage control characters. This switch has the following format:
INOCRLF

Characteristics:
1.

The INOCRLF switch is a global switch.

Both the input and output files must be fixed length.

This switch should be specified when you sort text files that contain no
carriage-return/line-feed. Such files may come from a foreign source or
you may be sorting the files to go to a foreign source.

3.3.8 IRANDOM Switch
The IRANDOM switch indicates that a FORTRAN random file is to be
processed by SORT/MERGE. This switch has the following format:
IRANDOM

Characteristics:
This switch has exactly the same effect as the IFIXED switch. The
/RANDOM switch is provided for the convenience of FORTRAN users.

3-14

SORT/MERGE Switches

3.3.9 /SEQUENTIAL Switch
The /SEQUENTIAL switch indicates that FORTRAN sequential files are to
be processed by SORT/MERGE. This switch has the following format:
/SEQUENTIAL
Characteristics:
This switch has exactly the same effect as the /VARIABLE switch. The
/SEQUENTIAL switch is provided for the convenience of FORTRAN users.

3.3.10 /VARIABLE Switch
The /V ARIABLE switch indicates that the input file
variable-length records. This switch has the following format:

contains

/VARIABLE
Characteristics:
1.

The /VARIABLE switch mayor may not be in effect by default, depending on the particular recording mode you are using. See the description
of the appropriate recording-mode switch to determine when
/V ARIABLE is in effect by default.

For general processing, variable length records process slower than
fixed length records.

3.4 Control Switches
The control switches are used to control parameters such as memory size,
tree size, and temporary area allocation, and to perform other functions
such as checking files to be merged and specifying alternate collating
sequences. All control switches are global, except for the /PHYSICAL
switch, which is a position dependent switch.

3.4.1

/CHECK Switch

The /CHECK switch indicates to check the order of input files during a
merge. This switch has the following format:
/CHECK
Characteristics:
1.

The /CHECK switch can be used only if the /MERGE switch has been
specified.

This switch increases the number of comparisons and causes some increase in run time. However, it protects against erroneously merging
unsorted files.

SORT/MERGE Switches

3-15

If out-of-sequence records are found, then the following error message
is generated:
?SRTMRS MERGE RECORD n NOT IN SEQUENCE FOR filename

3.4.2 /COLLATE Switch
The ICOLLATE switch indicates the collating sequence to be used when
sorting a file. This switch has the following format:
ICOLLATE:argument
Characteristics:
One of the following arguments must be specified with the ICOLLATE
switch:
1. ASCII

EBCDIC

~3.

FILE :filename

LITERAL:/collating sequencel

ADDRESS:address

The first two arguments (ASCII and EBCDIC) allow you to sort data in one
recording mode according to the collating sequence associated with another
recording mode.
For example,
if you
have specified
IASCII/COLLATE:EBCDIC for a file, then the data is interpreted as ASCII
data, but is sorted according to the EBCDIC collating sequence. Similarly,
you can specify IEBCDIC/COLLATE:ASCII and sort EBCDIC data according to the ASCII collating sequence.
The remaining arguments (FILE:, LITERAL:, and ADDRESS:) allow you to
specify your own collating sequence, either as a file or as a literal. The
ADDRESS argument is for FORTRAN users in a format of ~n construction,
where ~~n" is the number of the calling argument in the calling argument
list. (See the ~~addr" description of the IERROR: switch in Section 3.4.4,) It
is essential the FORTRAN user supplies a new complete sequence that
replaces the existing ASCII, SIXBIT, or EBCDIC sequence for purposes of
sorting and collating. The individual items in your collating sequence can
be specified as follows:
"A","B","C","D","E" = "F","G"-"N",117 = "P",121-"S","T" = 120,125 = 126

There are three functions illustrated in the above example:
1.

3-16

Equivalence - allows you to make particular characters sort as if they
were equal to other characters.

SORT/MERGE Switches

~~E" = ~~F" indicates that ((F" is assigned the same internal value as

~~E". Thus, ~~F" has been moved from its normal position in the collat-

ing sequence (whichever sequence you are using) and now has the
same position as (~E". "E"s and "F"s are equivalent and are sorted
accordingly.
b. 117 = ~~P" indicates that '~P" is equal to the character with the octal
value 117 (that is, ~(O") in the ASCII sequence. This does not affect
the character whose normal internal represen ta tion
is 117.
c.

~~T" = 120 indicates that the character whose octal value is 120 (that
is, (~P") is now equivalent to T in the new collating sequence.

d. 125 = 126 indicates that the character whose octal value is 126 (that
is, (~V") is equivalent to the character whose octal value is 125 (that
is, (~U"L
2.

Abbreviation - allows you to specify a range of characters without
having to type each one individually. For example, "G"-"N" is an abbreviated form of ~(G",~(H",((I",'(J","K","L","M","N".

Completion - SORT/MERGE must have a complete alternate collating
sequence before it can correctly sort or merge the data. (The term 'complete alternate collating sequence' means an alternate collating sequence in which every character from the recording mode in effect is
specified,) If you do not specify an alternate collating sequence that is
complete, then SORT/MERGE completes it by adding the missing characters (in their normal order) to the end of the partial collating sequence you specified. For example, if you specify only the following
items for your collating sequence:

(which is equal to ~~D"-('F") and the recording mode is ASCII, then the
actual collating sequence is made up of the characters you specified,
followed by the normal ASCII collating sequence (minus the characters
you specified). The complete alternate collating sequence is as follows:

To further illustrate the various ways an alternate collating sequence can
be defined, note how the following four sequences are equivalent:

101-104,105 = 107,132

SORT/MERGE Switches

3-17

NOTE
All data in the collating sequence file or in the literal is
ASCII data.
If you wish to store your alternate collating sequence in a file, you can
create the file using a text editor and save it with or without line sequence
numbers.
If you specify your alternate collating sequence as a literal, the sequence
must be delimited by a character that does not appear in the literal itself,
for example:
ICOLLATE:/LITERAL:rA","B","C"-"N"1
or
ICOLLATE:/LITERAL:M"A","B","C"-"N"M
If you specify a complete collating sequence as a literaL then you may not
have any character to use as a delimiter. unless you do one of the following:
1.

Specify a range of characters, such as "A"-"Z". This frees characters B
through Y for use as delimiters.

Specify an octal value for some character. For instance, if you specify
057 instead of "I", then you can use the slash character (/) as a
delimiter.

If the recording mode is SIXBIT, then you can successfully use any lowercase character as a delimiter.

3.4.3 /CORE Switch
The ICORE switch indicates the size of SORT/MERGE's low segment. This
switch has the following format:
ICORE:nK

specifies low segment in K (1024 words)

ICORE:nP

specifies low segment in pages (512 words)

Characteristics:
SORT/MERGE's default memory-allocation algorithm generally tries to

use about half of the available physical memory on the system, up to your
job's physical memory limit. If necessary, SORT/MERGE exceeds your job's
physical limit (causing the sort to become virtual), but it never exceeds
128K. The ICORE switch allows you to override the defaulting algorithm.
See Chapter 6, SORT/MERGE Performance Considerations, for more information on the use of the ICORE switch.

3-18

SORT/MERGE Switches

3.4.4 fERROR Switch
The /ERROR switch indicates the absolute address that execution control
should transfer to in the event that SORT/MERGE encounters a fatal error.
This switch has the following format:
/ERROR:addr
Characteristics:
The ~addr' is an octal integer specifying the absolute address that control
should transfer to. This switch was designed for FORTRAN users, who can
specify it in one of two ways:
CALL SORT('OUTFIL = INFILIswitches/ERROR:3454673')

or
CALL SORT(,OUTFIL = INFILIswitches/ERROR:'2',$99)

In the first example, an absolute octal address is used.
In the second example, the expression 'A2' is used to indicate that the argument to the /ERROR switch is the second (A2) calling argument in the
calling argument list, the statement label 99 (specified as $99). This allows
you to indirectly specify a symbolic address. At compile time, the symbolic
address is converted to a relocatable address. Then, at load time, the relocatable address is converted to the absolute address that SORT/MERGE
requIres.
The /ERROR switch is a global switch.

3.4.5 fEXIT Switch
The /EXIT switch indicates that you want to stop execution and exit to
TOPS-IO command level. This switch has the following format:
/EXIT

3.4.6 fFATAL Switch
The /FATAL switch indicates the absolute address in which to store the
error code if SORT/MERGE encounters a fatal error. This switch has the
following format:
/FATAL:addr
Characteristics:
The ~addr' is an octal integer specifying the absolute address where the
error code should be stored. The error code is a ASCII value of the form
SRTxxx, where xxx is a 3-letter code for the error type. These codes and
their associated error messages and explanations are listed in Chapter 5,
Error Messages.

SORT/MERGE Switches

3-19

This switch is designed for FORTRAN users who can specify it in one of two
ways:
CALL SORT('OUTFIL= INFILIswitches/FATAL:347231')

or
CALL SORT('OUTFIL= INFILIswitches/FATAL:'2',ERRCOD)

In the first example, an absolute address is given.
In the second example, the expression (A2' indicates that the argument to
the /F ATAL switch is the second calling argument in the call string:
ERRCOD. This allows you to indirectly specify a symbolic address. At compile time, the symbolic address is converted to a relocatable address. Then,
at load time, the relocatable address is converted to the absolute address
that SORT/MERGE requires.

3.4.7 ILEAVES Switch
The /LEAVES switch indicates the size of the binary tree that
SORT/MERGE is to use during the sort phase of a sort. This switch has the
following format:
/LEAVES:n
Characteristics:
1.

The argument (n' is a decimal number that indicates how many leaves
(or terminal nodes) the binary tree is to have.

This switch does not affect the merge phase of a sort and is illegal with
the MERGE command.

Ordinarily, you would rely on SORT/MERGE's default tree algorithm to
define the size of the tree. However, if you are concerned with improving
the performance of a particular sort, then you can specify the tree size with
the /LEA VES switch. See Chapter 6, SORT/MERGE Performance
Considerations, for more information on using the /LEAVES switch.

3.4.8 IMAXTEMP Switch
The /MAXTEMP switch indicates the maxinlum number of temporary files
that can be used during a sort or merge. This switch has the following
format:
/MAXTEMP:n
Characteristics:

3-20

This switch is a global switch.

The maximum number (n) of temporary files that you can specify is 26.
This is also the default number, if you do not specify the /MAXTEMP
switch.

SORT/MERGE Switches

3.4.9 IMESSAGE ,Switch
The IMESSAGE switch indicates how much of an error warning or informational message you want printed. This switch has the following format:
IMESSAGE:argument
Characteristics:
The possible arguments are:
1.

I NOIPREFIX - determines whether or not the message code (for example, SRTxxx) is printed.

INO lFIRST - determ.ines whether or not the text portion of the message is printed.

The arguments can be combined by specifying them in parenthesis. For
example, IMESSAGE:(NOPREFIX,FIRST) prints the text of the message
while
suppressing
the
message
code.
If you
specify
IMESSAGE:(NOPREFIX,NOFIRST), however, you still get the text portion
of the message.

3.4.10 IOPTION Switch
The IOPTION switch indicates the lines in SWITCH.INI that you want to
reference. This switch has the following format:
IOPTION:name
Characteristics:
1.

The ~name' argument must appear in the SWITCH.INI file In the
form of:
SORT:name/switchllswitch2/switch3 ...

By specifying this switch, you can force SORT/MERGE to read that
particular line(s) in SWITCH.INI that 'name' refers to and apply any
switches in the line to your sort.

3.4.11

IPHYSICAL Switch

The IPHYSICAL switch indicates to suppress logical name assignments in
file specifications. The result is that only physical devices are searched for
the data file(s). This switch has the following Jormat:
IPHYSICAL

3.4.12 IPRIORITY Switch
The IPRIORITY switch indicates the disk priority for your SORT/MERGE
job. This switch has the following format:
IPRIORITY:n
SORT/MERGE Switches

3-21

Characteristics:
1. The argument ~~n" can be a numeric value from -3 to 3. A value of + 3 is
the highest priority that can be specified; -3 is the lowest priority.
Higher priority jobs may process faster depending upon the queue
priority.

The default setting for the /PRIORITY switch is O. However, the default
can be changed with the DISK. monitor call. (See the TOPS-10 Monitor
Calls Manual,)

3.4.13 fRUN Switch
The IRUN switch indicates that you wish to run a specified program from
SORT/MERGE command level. This switch has the following format:
/RUN:filespec I/switch(es) I
Characteristics:
The ~filespec' is the name of the progranl. and '/switch(es)' can be any of the
following:
1. IRUNCORE:n - Run the program with a memory allocation of nK
(K = 1024 words) or nP (p = 1 page or 512 words).

/RUNOFFSET:n - Run the program. starting at the address + n. If
this switch is omitted, the default value is O. If the argument 'n' is
omitted, the argument defaults to 1. This switch is generally only useful to compilers. Once this switch is used. the argument in effect applies
to all subsequent runs, unless /RUNOFFSET is specified again with a
new argument.

ITMPFIL:nam:~string' -

Store the string in TMPCOR file ~nam'. The

~nam' can be no more than three characters long. This switch is useful

for passing a command string to a program.

3.4.14 fSUPPRESS Switch
The ISUPPRESS switch indicates to suppress various kinds of messages,
depending on which argument you specify. This switch has the following
format:
/SUPPRESS:argument
Characteristics:
The possible arguments are as follows:

3-22

NONE -

Suppress no messages.

INFORMATION ning with T).

SORT/MERGE Switches

Suppress all information messages (those begin-

WARNING - Suppress all information messages and all warning messages (those beginning with (lk').

FATAL - Suppress all information message, all warning messages,
and all fatal error messages (those beginning with (7').

ALL - Suppress all messages, except messages beginning with ($'.
(This has the same effect as F AT AL. )

NOTE
It is not possible to suppress those messages beginning with
($', since these are operator intervention messages (for exam-

ple, to mount the next reel of a multireel tape file).

3.4.15 ITEMP Switch
The /TEMP switch indicates that you wish to specify the names of the
devices for temporary file storage. This switch has the following format:
/TEMP
Characteristics:
1.

For stand-alone sorts, the number of devices is fixed at 26.

For COBOL and FORTRAN-called sorts, the devices you specify depends on the number of 110 channels available to SORT/MERGE.

One temporary file is created on each device that you specify. For example, if you specify the following temporary areas:
DSKA:/TEMP,DSKC:/TEMP,DSKB:/TEMP
The runs are appended to the above areas as follows:
DSKA: - /RUN lIRUN 4/RUN 7/ ... RUN n
DSKC: - /RUN 2/RUN 5/RUN 8/ ... RUN n + 1
DSKB: - /RUN 3/RUN 6/RUN 9/ ... RUN n + 2

If you do not specify this switch, then SORT/MERGE writes its temporary files (up to 26, as needed) on DSK:.

3.5 Tape Switches
Tape switches are used to define attributes of the tape read/write process
and to control the tape drive itself.

3.5.1

IDENSITY Switch

The /DENSITY switch indicates the tape density to be used for reading or
writing a magnetic tape. This switch has the following format:
/DENSITY:n

SORT/MERGE Switches

3-23

Characteristics:
1.

The argument (n' is a decimal number indicating the density and can
have one of the following values:
• 200
• 556
• 800
• 1600

• 6250
2.

The IDENSITY switch is a position dependent switch.

The default is installation dependent and can be set with the SET
DENSITY command. (See the TOPS-J 0 Operating System Commands
Manual.)

3.5.2 IINDUSTRY Switch
The IINDUSTRY switch indicates that a magnetic tape is to be read or
written in industry-compatible mode. This switch has the following format:
IINDUSTRY
Characteristics:
The IINDUSTRY switch is meaningful only for EBCDIC magnetic tapes.

3.5.3 ILABEL Switch
The ILABEL switch indicates the tape label status to SORT/MERGE. This
is a modified position dependent switch. This switch has the following
format:
ILABEL:argument
Characteristics:
The possible arguments are:
1. ANSI - If a tape label processor is present, ANSI-STANDARD labels
are read or written. Otherwise, this is similar to
ILABEL:NONSTANDARD. Thus, labels are skipped over on input and
omitted on output.

3-24

DEC - Currently equivalent to ILABEL:ST ANDARD, but preferred
for specification of COBOL-standard labels.

IBM - If a tape label processor is present, IBM-STANDARD labels are
read
or
written.
Otherwise,
this
is
similar
to
ILABEL:NONSTANDARD. Thus, labels are skipped over on input and
omitted on output.

SORT/MERGE Switches

NONSTANDARD - The file has nonstandard labels. SORT/MERGE
bypasses them on input and omits them on output. These labels are
assumed to be the size of one physical record.

OMITTED - The file has no labels, and SORT/MERGE does not perform any label checks.

STANDARD - The file has COBOL-standard labels. SORT/MERGE
reads and writes them on magnetic tape.

If you do not specify the /LABEL switch, then SORT/MERGE expects:
1.

Standard labels for a file on magnetic tape

No labels for a file on any other device

NOTE
Labels are ignored on disk because disk is a directory device.

3.5.4 /PARITY Switch
The /PARITY switch indicates the parity to be used when reading or writing a magnetic tape. This switch has the following format:
/P ARITY :argument
Characteristics:
1.

This switch is a position dependent switch.

The possible arguments are:
• EVEN
• ODD

The default is ODD parity.

3.5.5 /POSITION Switch
The /POSITION switch indicates that you wish to position the magnetic
tape before the file is read or written, but after rewinding (if required J. This
switch has the following format:
/POSITI ON :argument
Characteristics:
1.

The argument can be a signed positive or negative number. Positive
arguments skip that number of files. Negative arguments backspace
that number of files.

This switch is ignored for devices other than tape drives.

The default is to position at the first file on the tape.

SORT/MERGE Switches

3-25

3.5.6 IREWIND Switch
The IREWIND switch indicates to rewind the tape before the file is read or
written. This switch has the following format:
IREWIND

Characteristics:
1.

This switch is ignored for devices other than tape drives.

The IREWIND switch is a local switch.

3.5.7 ISTANDARD Switch
The ISTANDARD switch indicates that the data is to be read and written
in STANDARD-ASCII mode. This switch has the following format:
ISTANDARD
Characteristics:
1.

This switch is a modified position dependent switch.

The ISTANDARD switch is meaningful only for ASCII recording mode.

3.5.8 IUNLOAD Switch
The IUNLOAD switch indicates to rewind and unload the tape after the file
is read or written. This switch has the following format:
IUNLOAD

Characteristics:

3-26

This switch is ignored for any devices other than tape drives.

For multireel files, this switch affects only the disposition of the last
tape. All intermediate tapes are UNLOADed regardless of the setting
of this switch.

The IUNLOAD switch is a local switch.

Once the tape is unloaded, the tape cannot be read from or written to
unless the operator reloads the tape drive.

SORT/MERGE Switches

Chapter 4
File Formats

SORT/MERGE is capable of reading and writing data in four recording
modes and in a variety of file formats. The following sections describe the
four recording modes and all major file formats that are recognized by
SORT/MERGE. Each file format is described in detail, and the COBOL code
segments and/or FORTRAN programs that generate each file format are
given. In addition, each code segment or program has a detailed illustration
of the records it produces, and a description of the SORT/MERGE command
strings needed to sort the record on any of its fields.

4.1

Recording Modes
The recording mode specifies the byte size of the data and, except for binary
mode, also specifies the character set used. The four recording modes and
their respective byte sizes are:

RECORDING MODE

BYTE SIZE

ASCII
SIXBIT
EBCDIC
Binary

7 bits
6 bits
8 bits
36 bits (1 word)

The following sections describe the recording modes in more detail.

4-1

4.1.1

ASCII Recording Mode

An ASCII word consists of five characters left-justified in the word. Each
character is represented by a 7-bit byte:
ASCII RECORDING MODE

BIT NU M B E R _ 01 1 2 3 41 5 1 6 7 B 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
BINAR Y
0 o 0 01 0 1. o •
REPRE SENTATION- .1

•

DATA

• 0 0 •

• 0 • 0 0 0 0 • • X
C

BYTES: 5
= on bit
o = off bit
X = unused bit

•

MR-S-030-79

NOTE
A variant form of ASCII, line-sequence ASCII, sets bit 35 of
the line-sequence word to 1.
ASCII recording mode is specified to SORT/MERGE with the /ASCII
switch.

4.1.2 SIXBIT Recording Mode
SIXBIT is a compressed form of ASCII in which lowercase letters and a few
special characters are not used. A SIXBIT word consists of six characters
per word, with each character represented by a 6-bit byte:
SIXBIT RECORDING MODE

BIT N U M B E R - O 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
BINAR Y
REPR ESENTATION - - •

o 0 o •

o •

000

DATA

• 0 0

• 0

·• . ·.

BYTES· 6
• = on bit
o = off bit

MR-S-031-79

SIXBIT recording mode IS specified to SORT/MERGE with the /SIXBIT
switch.

4.1.3 EBCDIC Recording Mode
An EBCDIC word consists of four characters per word. Each byte is nine
bits long, but the first bit in each byte is unused. Each character is represented by eight bits:
EBCDIC RECORDING MODE

BIT NU M B E R - O 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
BINAR Y
X •
REPRE SENTATION - DATA

•

0 o 0 o 0 •

X •

...
1

• X

.•

•

· ..• ·
0

BYTES: 4
• = on bit
o = off bit
X = unused bit

4-2

File Formats

MR-S-1700-81

A variant form, used only for magnetic tape, is industry-compatible
EBCDIC. In this form of EBCDIC, there are four characters per word,
left-justified within the word. Each character is represented by an 8-bit
byte. The last four bits in the word are unused:
INDUSTRY-COMPATIBLE EBCDIC RECORDING MODE

BIT NU M B E R - O 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
BINAR Y
REPRE SENTATION
DATA

--. · .

•

o •

.. • .

•

..•

•

... • .
0

x x X x

BYTES: 4
• ~ on bit
o ~ off bit

MR-S-1701-81

Standard EBCDIC recording mode is specified to SORT/MERGE with the
/EBCDIC switch. Industry-compatible EBCDIC recording mode is specified
with the /EBCDIC/INDUSTRY switch combination.

4.1.4 BINARY Recording Mode
Unlike the recording modes previously mentioned, binary mode does not
specify a character set for the data. In binary mode, the entire 36-bit word
is interpreted as a single byte of binary data:
BINARY RECORDING MODE

BIT N U M B E R - O 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
BINAR Y
REPR ESENTATION
DATA

--.

000

. . . • . . ....
0

2,739,136

BYTES: 1
• ~ on bit
o ~ off bit

MR-S-034- 79

Binary recording mode is specified to SORT/MERGE with the /BINARY
switch.

4.2 File Formats
The file format specifies the structure of the record used to store the data.
The following sections describe all major file formats recognized by
SORT/MERGE. Each section includes a diagram of the file format and a
COBOL code segment or FORTRAN program that generates the file
format.
Note the following conventions that are used in the diagrams:
1.

Alphanumeric or numeric character data in a word is shown with each
individual character enclosed in a box. The box represents one byte.
Thus, a word of ASCII data is shown as follows:

MR-S-1364-81

File Formats

4-3

Binary data in a word (fixed- and floating-point numbers) is shown by
a number in the word:

MR-S-1365-81

EBCDIC packed-decimal values are shown as two decimal digits per
EBCDIC byte. The right half of the rightmost byte contains the sign.
Note that neither the digits nor the sign are EBCDIC characters.

COBOL signed nunleric data, such as produced by PIC S9(n), is shown
with the overpunched character, if the sign is negative. For example,
-12345 is shown as 1234N, with the N representing both the negative
sign and the value 5. DIGITAL's COBOL does not use overpunched
characters for positive sign representation, so diagrams depicting positive, signed numeric data do not show a sign. (Note that SORT/MERGE
accepts positive overpunched characters.)

Italicized characters in a diagram do not depict data; they label or
clarify parts of the diagram:

I RDW301 °
MR-S-1366-81

Heavy vertical lines are used to delimit individual fields within a
record:

lA B C 1 2
3

4 A 3

MR-S-1367-81

Padding, the use of blanks or nulls to force the next record to begin on
some boundary (for example, a word or disk-block boundary), is shown
by white space in the word: .

MR-S-1368-8 1

Note that you cannot consider padding as part of a record field, nor can you
use padding as part of a key field. However, the length of any padding must
be taken into account when calculating record length and key starting
position.

4-4

File Formats

4.2.1

Fixed-Length ASCII

A fixed-length ASCII file consists of records containing five characters per
36-bit word, with each group of five characters left-justified within the
word. Fixed-length ASCII records must end with a carnagereturn/line-feed. The following diagram illustrates the format of
fixed-length ASCII records:
WORD

RECORD

G!D

I
4

CD
~
~

CARRIAGE RETURN
LINE FEED
MR-S-035-79

The format is specified with the following SORT/MERGE switch
combination:
/ASCII/FIXED
4.2.1.1

COBOL Fixed-Length ASCII -

CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS ASCII.
DATA DIVISION.
FILE SECTION.
FD filename
01 record-1
02 field-1
02 field-2
02 field-3
02 field-4
02 field-5

VALUE OF ID "DATA FIL".
DISPLAY-7.
PIC X(6) VALUE "AB12EF".
PIC A(3) VALUE "GHI".
PIC 9(4) VALUE 3249.
PIC S9(6) VALUE -481253.
PIC S9(6)V9999 VALUE + 31458.5012.

Figure 4-1 illustrates the record produced by the code segment shown
above:

File Formats

4-5

Figure 4-1:

COBOL Fixed-Length ASCII

WORD

/KEY:1:6/ALPHA

/KEY:7:3/ALPHA

/KEY: 1O:4/NUMER IC/UNSI GNE D

/KEY:14:6/NUMERIC/SIGNED

/KEY:20:10 /NUMERIC/SIGNED

G::)
MR-S-1728-81

SORT/MERGE Comlnand String:

* 5 0 R TED • F I L = DAT. F I L / A5 C I I / F D: E D/ RE COR D : 29 / KEY: 1 : Gf ALP HA rRITl

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-1. Also, to retain the
FIXED format of the file shown in the example, no
ADVANCING clauses on WRITE statements can be used in
the COBOL code.
4.2.1.2 FORTRAN Fixed-Length (Random) ASCII PROGRAM:
DIMENSION A(2),C(2)
DOUBLE PRECISION Y
INTEGER J
REAL X

100

An) = 'ABCDE'; A(2) = 'F12'
B ='LMNO'
cn) = '12345'; C(2) = '6789'
D = '-1234'; 1=1
J
= 1234567890
X = 123456.4321
Y = -1435789432.456
OPEN (UNIT = 1,DEVICE = 'DSK',FILE = 'TEST.DAT',MODE = 'ASCII',
1ACCESS = 'RANDOM' ,RECORDSIZE = 68)
WRITE n#I,lOO)A,B,C,D,J,X,Y
FORMAT( 1A5,lA3,lA4,lA5,lA4,lA5,IlO,E17.8,D15.4)
CLOSE (UNIT= 1)
END

Figure 4-2 illustrates the record produced by the program shown above:

4-6

File Formats

Figure 4-2:

FORTRAN Fixed-Length (Random) ASCII

WORD

I
3

/KEY:1:8/ALPHA

/KEY:9:4/ALPHA

/KEY:13:9/NUMERIC/UNSIGNED

L..J

9
10

L..J

/KEY:22:5/NUMERIC/SIGNED
/KEY:27:10 /NUMERIC/UNSIGNED

/KEY:37:17 /FORMAT:E17.8

/KEY:54:15/FORMAT:D15.4

L..J ~ blank

MR-S-1729-81

SORT/MERGE Command String:
*SORTEO.FIL=TEST.DAT/FORTRAN/ASCII/RANDOM/RECORD:GBIKEY:1:B/ALPHAm

FORTRAN users can describe this file format with either of the following
SORT/MERGE switch combinations:
1.

/FORTRAN/ASCII/RANDOM

2. /FORTRAN/ASCII/FIXED

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-2.

File Formats

4-7

4.2.2 Variable-Length ASCII
Variable-length ASCII consists of records containing five characters per
36-bit word, with each group of five characters left-justified within the
word. Variable-length ASCII records must end with some combination of
the following:
1. carriage return
2.

line feed

vertical tab

form feed

The following diagram illustrates the format of variable-length ASCII
records:
A

C2Q

GO C2Q
D

3
4

GD GO
GD = CARRIAGE RETURN
GO = VERTICAL TAB
C2Q

= FORM FEED

CD = LINE FEED

MR·S·037·79

SORT/MERGE strips away all end-of-line characters and replaces them
with carriage-return/line-feed on output.
This format is specified with anyone of the following SORT/MERGE switch
combinations:
1. /ASCII/VARIABLE
2.

4-8

/ASCIIUVARIABLE in effect by default)

File Formats

4.2.2.1

COBOL Variable-Length ASCII -

VALUE OF ID "DATA FIL".
DISPLAY-7.
PIC X(7) VALUE "AB13521".
PIC S9(7)V99 VALUE -:3269.02.
PIC A(3) VALUE "ILM".
PIC 9(4) VALUE 1359.

01 record-2
02 field-1
02 field-2
02 field-3
02 field-4

DISPLAY-7.
PIC X(7) VALUE "EFGHI95".
PIC S9(7)V99 VALUE 42553.40.
PIC A(3) VALUE "LMN".
PIC 9(7) VALUE 3712536.

PROCEDURE DIVISION.
WRITE record-1 BEFORE ADVANCING.
WRITE record-2 BEFORE ADVANCING.

Figure 4-3 illustrates the record produced by the code segment shown
above:

Figure 4-3:

COBOL Variable-Length ASCII

WORD

1
~-

IKEY: 1 :7/ALPHA

1----------

IKEY:8:7/NUMERIC/SIGNED

IKEY:15:3/ALPHA

G:Q

I
I

NOTE
It IS unWise to speCify a key that extends
outSide the shorter records In a variable
length file, unless the data type IS ALPHA
NUMERIC. For other data types, the result
IS undefined and will result III a warning mes
sage. For thrs reason, a key specrfrcatlon IS
not given for the last field In the record
MR-S-1730-81

File Formats

4-9

SORT/MERGE Command String:

*S 0 RTEO. F I L =0 AT A • F I L / AS C I I / t.J AR I ABL E / RECOR 0 : LI 7 / K~V~rri'1'1?rC'pJ~Aru

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-3.
4.2.2.2 FORTRAN Variable-Length (Sequential) ASCII PROGRAM:
DIMENSION A(2),C(2)
DOUBLE PRECISION Y
INTEGER J
REAL X

100

200

A( 1) = 'ABCDE'; A(2) = 'F12'
B ='-1234'
J
= 1234567890
X = 123456.4321
Y = -1435789432.456
C( 1) = '12345'
OPEN (UNIT = 1,DEVICE = 'DSK' ,FILE = 'TEST1.DAT' ,MODE = 'ASCII',
1ACCESS = 'SEQOUT')
WRITE (1,100)A,B,J,X,Y,C(l)
FORMAT(1A5,lA3,1A5,1l0,E9.2,D12.4,1A5)
A(l) = 'LMNOP'; A(2) = '22C'
B ='+3500'
J
= 4567912343
X =4569.723
Y = + 45982341234.234
C( 1) = '54402'; C(2) = '6789'
WRITE (l,200)A,B,J,X,Y,C
FORMAT(1A5,1A3,1A5,1l0,E9.2,D12.4,lA5,1A4)
CLOSE (UNIT = 1)
END

Figure 4-4 illustrates the record produced by the program shown above:

4-10

File Formats

Figure 4-4:

FORTRAN Variable-Length (Sequential) ASCII

WORD

I
4

/KEY: 1 :8 / ALPHA

/KEY:9:5/NUMERIC/SIGNED

/KEY:14:10 /NUMERIC/SIGNED

L.....I

6
0
8

G:D

G:Q
M

/KEY:24:9/FORTRAN:E9.2

/KEY:33:12/FORMAT:D12.4

NOTE

It IS unwise to specify a key that extends
outside the shorter records In a variable·
length file, unless the data type is ALPHA·

NUMERIC. For other data types, the result

sage. For this reason, a key specification IS

L.....I

IS undefmed and will result In a warnmg mes
not given for the last field in the record.

17
18

t·

G:D

L.....I = blank

MR-S-1731-81

SORT/MERGE Command String:
*SORTED.FIL=DATA.FIL/FORTRAN/ASCII/SEOUENTIAL -®m
RECORD: Ll8 ZWgW;l{f8':J~l,[pH:'Am

# I

FORTRAN users can specify this format with anyone of the following
SORT/MERG switch combinations:

IFORTRAN/ASCII/SEQUENTIAL

IFORTRAN/ASCII/VARIABLE

3. IFORTRAN/ASCIIUVARIABLE in effect by default)

File Formats

4-11

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-4.

4.3 Fixed-Length SIXBIT
In a SIXBIT file, characters are stored six per 36-bit word, and a SIXBIT
record must start and end on a word boundary. The left half of the first
word in the record contains one of the following:
1.

The record sequence number of COBOL magnetic tape records

Data specific to COBOL ISAM records (lSAM files cannot be directly
sorted by SORT/MERGE.)

Binary zeros

The right half of the first word contains the number of characters in the
record. To ensure that the record ends on a word boundary, the last word in
the record is padded with blanks, if necessary. When determining the size
of the record for memory considerations, you must take into account the
first word of the record (containing file-access information and a character
count) and the possible existence of padding characters (blanks) to enable
the record to end on a word boundary.
The following diagram illustrates the format of fixed-length SIXBIT records. Note that the character count is the same for each record:
WORD

L...I

FAD

F
L...I

L...I

F
L...I

FAD = FILE ACCESS DATA
CC

= CHARACTER COUNT

L...I

= BLANK (USED AS PADDING CHARACTER)
MR-S-039-79

This format
combination:

/SIXBIT/FIXED

4-12

File Formats

specified

the

following

SORT/MERGE

switch

4.3.1

COBOL Fixed-Length SIXBIT

CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS SIXBIT.
DATA DIVISION.
FILE SECTION.
VALUE OF ID "DATA FIL".
DISPLAY-6.
PIC X(4) VALUE "A13B".
PIC A(5) VALUE "CDEFG".
PIC 9(10) COMP VALUE 9654839218.
PIC X(2) VALUE "HI".
PIC 9(11) COMP VALUE 34567982314.
PIC 9(4) VALUE 1289.
PIC 9(5) COMP-1 VALUE 123.45.
PIC 9(11) COMP VALUE 12398756983.

FD filename
01 record-1
02 field-1
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6
02 field-7
02 field-8

Figure 4-5 illustrates the record produced by the code segment shown
above:

Figure 4-5:

COBOL Fixed-Length SIXBIT

WORD
CC

FAD

I c I

/KEY:5:5/ALPHA

9654839218

3
H

/KEY:l:4/ALPHA

/KEY:13:10 /COMP
/KEY:19:2/ALPHA

34567982314

6
1

/KEY:37:4/NUMERIC/UNSIGNED
/KEY:43:5/COMPl

123.45

/KEY:25:11 /COMP

9
I--

12398756983

/KEY:49:11 /COMP
MR-S-1732-81

SORT/MERGE Command String:

* S 0 RTED. F I L =0 AT A • F I L / S I )( 5 I T / F I )( E 0 / RECOR 0 : 6 <) :l$l;g l~:t~(~~1! F!:t!:~ru
NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-5.

File Formats

4-13

4.4 Variable-Length Sixbit
This format is the same as fixed-length SIXBIT, except that the character
count can vary from record to record. The following diagram illustrates the
format of variable-length SIXBIT records:
WORD

RECORD

FAD
2

L...I

3
4

FAD

FAD = FILE ACCESS DATA
CC

= CHARACTER COUNT

L...I

= BLANK (USED AS PADDING CHARACTER)

MR-S-041-79

This format is specified by either of the following SORT/MERGE switch
combinations:
1. /SIXBIT/VARIABLE

4-14

/SIXBITUV ARIABLE is in effect by default)

File Formats

4.4.1

COBOL Variable-Length SIXBIT

CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS SIX BIT.
DATA DIVISION.
FILE SECTION.
FD filename
01 record-l
02 field-l
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6
02 field-7
02 field-8

VALUE OF ID "DATA FIL".
DISPLAY-6.
PIC 9(7) COMP-l VALUE 123.4567.
PIC X(3) VALUE "A3C".
PIC A(3) VALUE "DEF".
PIC 9(3) VALUE -55.
PIC 9(10) COMP VALUE 1234567809.
PIC 9( 11) COMP VALUE 98765432108.
PIC X(2) VALUE "A2·'.
PIC 9(5) COMP VALUE 32571.

01 record-2
02 field-l
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6
02 field-7
02 field-8

DISPLAY-6.
PIC 9(7) COMP-1 VALUE 1395.678.
PIC X(3) VALUE "B5L".
PIC A(3) VALUE "LMN".
PIC 9(3) VALUE 79.
PIC 9(10) COMP VALUE 8176596821.
PIC 9(11) COMP VALUE 18976532150.
PIC X(2) VALUE "M5".
PIC 9( 11) COMP VALUE 12357986183.

PROCEDURE DIVISION.
WRITE record-I.
WRITE record-2.

Figure 4-6 illustrates the record produced by the code segment shown
above:

File Formats

4-15

Figure 4-6:

COBOL Variable-Length SIXBIT

WORD
CC

FAD

/KEY: 1: 7 /COMPl

123.4567

/KEY:7:3/ALPHA

/KEY:l0:3/ALPHA

1234567809

/KEY:13:3/NUMERIC/SIGNED

/KEY:19:10 /COMP
98765432108

'---

6
2

/KEY:25:11 /COMP
/KEY:37:2/ALPHA

32571

FAD

54
NOTE

1395.678
2

8176596821

It is unwise to specify a key that extends
outside the shorter records in a variablelength file, unless the data type is ALPHANUMERIC. For other data types, the result
is undefined and will result in a warning message. For this reason, a key specification is
not given for the last field in the record.

5
'---

18976532150

12357986183

6
M

MR-S-1733-81

SORT/MERGE Command String:
*SORTEO.FIL=DATA.FIL/SIXBIT/VARIABLE/RECORD:

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-6.

4-16

File Formats

4.5 EBCDIC File Formats
On disk and in memory, the characters in an EBCDIC file are represented
by 8 bits right-justified in 9-bit bytes. On tape, the characters in an
EBCDIC file are represented by 8-bit bytes, and 4 bytes occur per 36-bit
word. EBCDIC records are written only by COBOL programs.

4.5.1

COBOL Fixed-Length EBCDIC

Within a given file, fixed-length EBCDIC records all have the same record
length, and the record need not begin or end on a word boundary. The
following diagram illustrates the format of fixed-length EBCDIC records:
WORD

FAD

123.4567

IKEY: 1,7, COMPl

IKEY: 10,3, ALPHA

1234567809

IKEY: 13,3, NUMERIC, SIGNED

I F

IKEY: 19, 10, COMP
'-

98765432108

32571

FAD

1395.678

8176596821

18976532150

12357986183

IKEY: 25, 11, COMP
IKEY: 37,2, ALPHA

IKEY: 7,3, ALPHA

NOTE
It is unwise to specify a key that extends
outside the shorter records in a variablelength file, unless the data type is ALPHANUMERIC. For other data types, the result
is undefined and will result in a warning message. For this reason, a key specification is
not given for the last field in the record.

MR-S-043· 79

The file format is specified by the following SORT/MERGE switch
combination:
/EBCDIC/FIXED

File Formats

4-17

CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS F.
DATA DIVISION.
FILE SECTION.
VALUE OF ID "DATA FIL".
DISPLAY-9.
PIC 9(3) VALUE 123.
PIC X(5) VALUE "ABCDE".
PIC A(2) VALUE "LM".
PIC 9(9) COMP-3 VALUE 137958795.
PIC S9(6) COMP-3 VALUE -351235.

FD filename
01 record-l
02 field-l
02 field-2
02 field-3
02 field-4
02 field-5

Figure 4-7 illustrates the record produced by the code segment shown
above:

Figure 4-7: COBOL Fixed-Length EBCDIC

7 :9

1 :3

7 :9

5: +

/KEY:1:3/NUMERIC/UNSIGNED

/KEY:4:5/ALPHA
/KEY:9:2/ALPHA
/KEY: 11:9 /COMP3/UNSIGNED

I
I

5 :1

2 :3

5 :-

/KEY:16:6/COMP3/SIGNED
MR-S-1734-81

SORT/MERGE Command String:
*SoRTEo • F I L = DATA. F I L / EBCD I C / F I ){ED / RECORD: 19/KE'{ 1 £3;'<ALPHAru

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-7.

4.5.2 COBOL Variable-Length EBCDIC
In this file format, the record lengths can vary from record to record. Each
record contains a 4-byte Record Descriptor Word (RDW) at the head of the
record. The left half word of the RDW specifies a value equal to the number
of bytes in the record plus 4 (to allow for the length of the RDW itself). The
rightmost two bytes of the RDW must be zero. If they are nonzero, they
indicate spanned records, which are unsupported. The following diagram
illustrates the format of variable-length EBCDIC records:

4-18

File Formats

RECORD

WORD

ROW

ROW = RECORD DESCRIPTOR WORD
MR-S-045-79

This format
combination:

specified

the

following

SORT/MERGE

switch

/EBCDIC/V ARIABLE
CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS V.
DATA DIVISION.
FILE SECTION.
FD filename
01 record-1
02 field-1
02 field-2
02 field-3
02 field-4
02 field-5

VALUE OF ID "DATA FIL".
DISPLAY-9.
PIC S9(7) COMP-3 VALUE -1398569.
PIC S9(8) COMP-3 VALUE 57635937.
PIC 9(3) VALUE 596.
PIC A(2) VALUE "AB".
PIC X(5) VALUE "A13DE".

01 record-2
02 field-l
02 field-2
02 field-3
02 field-4
02 field-5

DISPLAY-9.
PIC S9(7) COMP-3 VALUE 5369787.
PIC S9(8) COMP-3 VALUE -53896156.
PIC 9(3) VALUE 593.
PIC A(2) VALUE "MN".
PIC X(8) VALUE "ILH5MLXY".

PROCEDURE DIVISION.
WRITE record-I.
WRITE record-2.

Figure 4-8 illustrates the record produced by the code segment shown
above:

File Formats

4-19

CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS F.
DATA DIVISION.
FILE SECTION.
FD filename
VALUE OF ID "DATA FIL"
BLOCK CONTAINS 1 RECORDS.
01 record-l
DISPLAY-9.
02 field-l
PIC 9(3 ) VALUE "194".
PIC X(5) VALUE "BDEFG".
02 field-2
PIC A(2) VALUE "MN".
02 field-3
PIC 9(5) COMP-3 VALUE 13796.
02 field-4
02 field-5
PIC S9(4) COMP-3 VALUE 1985.
DISPLAY-9.
PIC X(3) VALUE "762".
PIC X(5) VALUE "LANBH".
PIC A(2) VALUE "AB".
PIC 9(5) COMP-3 VALUE 76543.
PIC S9(4) COMP-3 VALUE -9764.

01 record-2
02 field-l
02 field-2
02 field-3
02 field-4
02 field-5

PROCEDURE DIVISION.
WRITE record-I.
WRITE record-2.

Figure 4-9 illustrates the record produced by the code segment shown
above:

Figure 4-9:

COBOL Blocked Fixed-Length EBCDIC

6 +

1 3

7:9
,
I

I
918

/KEY:1 :3/NUMERIC/UNSIGNED

/KEY:9:2/ALPHA

5 I

/KEY: 11 :5/COMP3/UNSIGNED

-------~~~

I
4

I 7:6

5:4

3'+
I

7:6

4'I

/KEY:4:5/ALPHA

/KEY:14:4/COMP3/UNSIGNED

-~~

MR-S-1736-81

File Formats

4-21

SORT/MERGE Command String:
*SORTEO.FIL=OATA.FIL/EBCOIC/FI)-(EO/BLOCKEO: 1/RECORO: 18 -rBrrJ
#IKEYl 1:3/NUMER I CIS I GNEOrBrrJ

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-9.

4.5.4 COBOL Blocked Variable-Length EBCDIC
In this file format, the record length can vary from record to record. Each
record contains a 1-word Record Descriptor Word (RDW) at the head of the
record. This word contains (in the left half word) a count of all bytes in the
record and in the RDW itself. The right half of the RDW must be zero. The
records are read and written in groups called blocks. The actual number of
records in a block depends on the blocking factor specified when the file was
created. Each block of a record contains a 1-word Block Descriptor Word
(BDW) which contains a count (in the left half word) of the bytes in the
block. The bytes of data, the bytes of the RDW for each record in the block,
and the four bytes of the BDW itself are included in the block count. The
following illustrates the format of blocked variable-length EBCDIC
records:
RECORD

WORD

BOW

ROW

BLOCK

L,...---

201

BOW

202

ROW

203

ROW

204
205
206
207

0
C

BOW = BLOCK DESCRIPTOR WORD
ROW = RECORD DESCRIPTOR WORD
MR-S-048-79

4-22

File Formats

This format IS specified with the following SORT/MERGE switch
combinations:
1.

/EBCDIC/V ARIABLE/BLOCKED:n

2. /EBCDIC/BLOCKED:n(!V ARIABLE is in effect by default)

NOTE
SORT/MERGE accepts this format from disk only.
SORT/MERGE does not correctly sort blocked
variable-length EBCDIC files on tape.
CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS V.
DATA DIVISION.
FILE SECTION.
FD filename
VALUE OF ID "DATA FIL"
BLOCK CONTAINS 1 RECORDS.
01 record-1
DISPLAY-9.
02 field-1
PIC S9(7) COMP-3 VALUE + 9356127.
02 field-2
PIC 9(7) COMP-3 VALUE 3987156.
02 field-3
PIC X(3) VALUE "198".
PIC A(2) VALUE "MN".
02 field-4
PIC S9(9) COMP-3 VALUE -569138279.
02 field-5
02 field-6
PIC X(6) VALUE "ABCDEF".
01 record-2
02 field-1
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6

DISPLAY-9.
PIC S9(7) COMP-3 VALUE -3295865.
PIC 9(7) COMP-3 VALUE 9378518.
PIC X(3) VALUE "196".
PIC A(2) VALUE "AL".
PIC 9(9) COMP-3 VALUE 569138279.
PIC X(9) VALUE "ABCDEFGHI".

PROCEDURE DIVISION.
WRITE record-1.
WRITE record-2.

Figure 4-10 illustrates the record produced by the code segment shown
above:

File Fornlats

4-23

Figure 4-10: COBOL Blocked Variable-Length EBCDIC

BOW

ROW

0
0
I

9:3

5:6

1 12

7 ;+

I
I

3:9
I

8: 7
1

1 5

6: +
1

/KEY:l:7/COMP3/SIGNED
/KEY:5:7/COMP3/SIGNED
/KEY:9:3/NUMERIC/UNSIGNED

,
4

5:6

,
I

5
6

2: 7

9:-

BOW

ROW

3:8

/KEY:14:9/COMP3/UNSIGNED

*/KEY:19:9/ALPHA

- - - ----- ------0
0

length file, unless the data type is ALPHA·

1
5,1
!
8,+

NUMERIC. For other data types, the result

9:5

8:6
~

,
7:, 8

5: 1
1

5:6

91 1

3:8
,

9:3

*NOTE
It IS unwise to specify a key that extends
outside the shorter records in a variable·

3:2
,

/KEY:12:2/ALPHA

9: 1

is undefined and will result in a warning
message.

I
I

5
6

2: 7

9:+

~--------

MR-S-1737-81

SORT/MERGE Command String:
*SoRTEo. F I L =DATA. F I L I EBCo I C It,JAR I ABLE I BLOCK EO: 1 I RECORD: 27 -(ill)
#/KEY:l:7/COMP3/SIGNED(ill)

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the command string with one or more of the
key specifications shown in Figure 4-10.

4-24

File Formats

4.6 Binary File Formats
Binary records consist of contiguous 36-bit words. Each record starts and
ends on a word boundary. Binary is the only recording mode which does not
have a character set associated with it, and standard binary records can
only be interpreted as COMPUTATIONAL and COMP1 binary numbers.
However, it is possible to associate a character set with binary records by
writing mixed-mode records. These records are specified with one of the
following SORT/MERGE switch combinations:
1.

IBINARY/ASCII

IBINARY/SIXBIT

IBINARY/EBCDIC

These formats are discussed in the following sections. Note that for each
file format diagram, two sets of key specifications and command strings are
given. The first set illustrates sorting the file in standard binary, and the
second set illustrates sorting the file in mixed-mode binary.

4.6.1

COBOL Binary File Formats

COBOL programs are capable of writing all three mixed-mode binary formats. Each format is discussed below.
4.6.1.1

COBOL ASCII Mixed-Mode Binary -

CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS BINARY.
DATA DIVISION.
FILE SECTION.
FD filename
01 BINARY-REC
02 field--1
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6
02 field-7

VALUE OF ID "DATA FIL".
DISPLAY-7.
PIC S9(10) COMP VALUE 12345678910.
PIC S9(10) COMP-1 VALUE 1246.597892.
PIC X(7) VALUE "ABCDE12".
PIC 9(11) COMP VALUE 12345678954.
PIC 9(3) VALUE "532".
PIC 9(14) COMP VALUE 12345678954.
PIC A(2) VALUE "LM".

Figure 4-11 illustrates the record produced by the code segment shown
above:

File Formats

4-25

Figure 4-11:

COBOL Standard Binary and ASCII Mixed-Mode
Binary

STANDARD BINARY

ASCII MIXED-MODE BINARY

IBINARY

IBINARY/ASCII

/KEY:1:1/SIGNED
/KEY:2:1/SIGNED
CAN'T BE SORTED AS CHARACTERS

/KEY: 1: 10 /COMP/SIGNED

1234568910

/KEY:6:10 /COMP1/SIGNED

1246.597892
3

/KEY:5: 2/UNSIGNED
6

CAN'T BE SORTED AS CHARACTERS

/KEY:8: 2 /UNSIGNED

I E

/KEY:21: 11 /COMP/UNSIGNED
12345678954

l
/KEY:31:3/NUMERIC/UNSIGNED
/KEY:36:14/COMP/UNSIGNED

12345678954967

CAN'T BE SORTED AS CHARACTERS 10

/KEY:11:7/ALPHA

l
/KEY:46:2/ALPHA
MR-S-1738-81

SORT/MERGE Command Strings:
Standard Binary:
*SORTEO. F I L =DATA. F I LIB I NARY IRECORD: 1 O/Kt::V: Ulf.$IGNEDOOl

Mixed-Mode Binary:
ItSORTED.FIL=DATA.FIL/BINARY/ASCII/RECORD:47 -00
1*:; K EY 1:1 01 CDMPU I S I GNED([E]

4-26

File Formats

4.6.1.2 COBOL SIXBIT Mixed-Mode Binary CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS BINARY.
DATA DIVISION.
FILE SECTION.
FD filename
01 BINARY-REC
02 field-1
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6
02 field-7

VALUE OF ID "DATA FIL".
DISPLAY-6.
PIC S9(10) COMPVALUE 12345678910.
PIC S9(10) COMP-1 VALUE 1234.592175.
PIC X(7) VALUE "ABCDE12".
PIC 9(11) COMP VALUE 12345678954.
PIC X(3) VALUE "532".
PIC 9(14) COMP VALUE 12345678954.
PIC A(2) VALUE "LM".

Figure 4-12 illu~trates the record produced by the code segment shown
above:

Figure 4-12:

COBOL Standard Binary and SIXBIT Mixed-Mode
Binary

STANDARD BINARY

SIXBIT MIXED-MODE BINARY
/BINARY/SIXBIT

/BINARY
WORD
/KEY:1 :l/SIGNED
/KEY:2:1/SIGNED
CAN'T BE SORTED AS CHARACTERS

/KEY:1 :10 /COMP/UNSIGNED

12345678910

/KEY:7:10 /COMP1/SIGNED

1234_592175
3

/KEY:5: 2/UNSIGNED
6
CAN'T BE SORTED AS CHARACTERS

/KEY:8: 2/UNSIGNED

L
5

B~.cIDIE

/KEY:13:7/ALPHA

/KEY:25:11 /COMP/UNSIGNED
12345678954

I 2 I

l
/KEY:37:3/NUMERIC/UNSIGNED
/KEY:43:14/COMP/UNSIGNED

12345678954967
9

CAN'T BE SORTED AS CHARACTERS 10

I 1

l
/KEY:55:2/ALPHA
MR-S-1739-81

SORT/MERGE Command Strings:
Standard Binary:
*SORTED.FIL=DATA.FIL/BINARY/RECORD: 10 -ru
~ ~ -~-E>!{J;:$ !;if:lt:lNS:;J~CiI'i.V~,~ru

File Formats

4-27

Mixed-Mode Binary:

* 5 0 R TED. F I L =D A T A • F I L / BIN A R Y / 5 I )( BIT / RE COR D : 56 -!RET)

#iKEV:25:11/COMPUjUNSJGNED~m .

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the standard binary command string with
one or more of the key specifications given on the left-hand
side of Figure 4-12. Likewise, replace the shaded portion of
the mixed-mode binary command string with one or more of
the key specifications given on the right-hand side of
Figure 4-12.
4.6.1.3 COBOL EBCDIC Mixed-Mode Binary CODE SEGMENT:
ENVIRONMENT DIVISION.
INPUT-OUTPUT SECTION.
FILE-CONTROL.
SELECT filename ASSIGN TO DSK
RECORDING MODE IS BINARY.
DATA DIVISION.
FILE SECTION.
FD filename
01 BINARY-REC
02 field-1
02 field-2
02 field-3
02 field-4
02 field-5
02 field-6
02 field-7
02 field-8
02 field-9

VALUE OF ID "DATA FlL".
DISPLAY-9.
PIC S9(10) COMP VALUE 12345678910.
COMP-1 VALUE 1246.597861.
PIC X(7) VALUE "ABCDE12".
PIC 9(11) COMP VALUE 12345678954.
PIC 9(3) VALUE "532".
PIC 9(14) COMP VALUE 12345678954967.
PIC A(2) VALUE "LM".
PIC S9(5) COMP-3 VALUE -72539.
PIC 9(8) COMP-3 VALUE 36193586.

Figure 4-13 illustrates the record produced by the code segment shown
above:

4-28

File Formats

Figure 4-13: COBOL Standard Binary and EBCDIC Mixed-Mode
Binary
STANDARD BINARY

EBCDIC MIXED-MODE BINARY

IBINARY

IBINARY (EBCDIC
WORD

/KEY:1: 1/SIGNED
/KEY:2:1/SIGNED
CAN'T BE SORTED AS CHARACTERS

IKEY:5: 2/UNSIGNED

CAN'T BE SORTED AS CHARACTERS

l
IKEY:25:3/NUMERIC/UNSIGNED

IKEY:29:14/COMP/UNSIGNED

8
12345678954967

9
10

CAN'T BE SORTED AS CHARACTERS

IKEY:9:7/ALPHA

IKEY: 17: 11 /COMP/UNSIGNED
12345678954

CAN'T BE SORTED AS CHARACTERS

IKEY:5:10 ICOMP1/SIGNED

1246.597861

/KEY:8: 2/UNSIGNED

IKEY:1 :10 /COMP/SIGNED

12345678910

l2 5:3

IKEY:37:2/ALPHA

9 :-

: 3

6 : 1

5! 8

9 : 3

:+ I

IKEY:39:5/COMP3/UNSIGNED
IKEY:42:8/COMP3/SIGNED
MR-S-1740-81

SORT/MERGE Command Strings:
Standard Binary:
*SORTED,FIL=DATA,FIL/BINARY/RECORD:12 -m

# IK.EY: ~ 5

::'41 !lfN8~;GNr;:·O.m

Mixed-Mode Binary:
*SORTED,FIL=DATA,FIL/BINARY/EBCDIC/RECORD:4G -m

# '1~K~E Y:":17=!:1}Cqtyip(J:lUI\(STGN E

File Formats

4-29

4.6.2 FORTRAN Binary File Formats
FORTRAN programs can generate two types of binary files: those with
Logical Segment Control Words (where MODE = (BINARY') and those
without Logical Segment Control Words (where MODE = (IMAGE'). Also,
each of these file types can be written, randomly (where
ACCESS = (RANDOM') or sequentially (where ACCESS = (SEQOUT').
Logical Segment Control Words (LSCWs) are used to delimit each record of
a file written with MODE = (BINARY'. If the file is random, only two
LSCW s occur per record: a code 1 LSCW at the beginning of the record and
a code 3 LSCW at the end of the record. If the file is sequential, any record
that crosses a block boundary has an additional LSCW at the beginning of
the block containing the remainder of the record. This additional LSCW is
a code 2 LSCW. The following diagram illustrates a sequential file containing all three types of LSCWs:
BLOCK 1
C

D
E

REC 1

D
E

BLOCK 2
C

D
E

REC 2

D
E

0
REC 2

REC 3

D
E

LSCW's

REC = RECORD
MR-S-1707-8i

The LSCW is a 36-bit word consisting of an octal code value (1, 2, or 3) in
the first nine bits of the word and a count value in the right half word. For
random files and sequential files that do not cross block boundaries, the
count values are as follows:
LSCW

COUNT

CODE 1
CODE 3

N umber of words in record + 1
Number of words in record + 2

For sequential records that cross a block boundary, the count values are as
follows:
LSCW

COUNT

CODE 1
CODE 2

N umber of words in the block
Number of words in the record that cross the
boundary + 1
N umber of words in the record + 3

CODE 3

block

A further complication is that the four file types previously discussed can
each be written in standard binary or mixed-mode binary (using the ASCII
character set). Thus, there are actually eight major binary file formats
written by FORTRAN programs:

4-30

Random Standard Binary with LSCWs

Random Standard Binary without LSCWs

File Formats

Random Mixed-Mode Binary with LSCWs

Random Mixed-Mode Binary without LSCWs

Sequential Standard Binary with LSCWs

Sequential Standard Binary without LSCWs

Sequential Mixed-Mode Binary with LSCWs

Sequential Mixed-Mode Binary without LSCWs

The values for record length, key position, and length change, depending on
whether or not the file is written in standard binary or mixed-mode binary.
Also, the existence of control words and the use of sequential versus random 110 requires changes in the SORT/MERGE command string used to
sort the file. Thus, it is very important that you understand the characteristics of a given FORTRAN-generated binary file before you attempt to
sort it.
The following sections describe all of the previously mentioned file formats.
Note that each file format diagram includes two sets of key specifications
and command strings. The first set illustrates sorting the file in standard
binary mode, and the second set illustrates sorting the file in mixed-mode
binary. Also, note that, for the sake of program simplicity, the sample
FORTRAN programs use literals to produce mixed-mode binary. The more
common practice is to use ENCODE, DECODE, and FORMAT statements
to produce mixed-mode binary.
4.6.2.1

FORTRAN Random Binary (with LSCWs) -

PROGRAM:
DIMENSION A( 128)
DOUBLE PRECISION Y; REAL X; INTEGER J
A(1)='ABCDE'; A(2)='FG'
A(3)='LM135'
A(4) = '-1234'
A(5) = '.1256'; A(6) = '7E + 03'
J
= 1234567890
X = 98765.43212
Y = 500400300.200101
I
=1
OPEN (UNIT= 1,DEVICE ='DSK',FILE ='DATA.FIL',MODE = 'BINARY',
1ACCESS = 'RANDOM',RECORDSIZE = 132)
WRITE (1#IlA,J,X,Y
CLOSE (UNIT = 1)
END

Figure 4-14 illustrates the record produced by the program shown above:

File Formats

4--31

Figure 4-14: FORTRAN Standard and Mixed-Mode Random
Binary with LSCWs
STANDARD FORTRAN RANDOM BINARY

MIXED-MODE FORTRAN RANDOM BINARY

/FORTRAN/RANDOM/BI NARY

/FO RTRAN/RANDOM/BI NARY/ASCII

WORD

CAN'T BE SORTED
AS CHARACTERS

L...I

/KEY:1 1:5/ALPHA

/KEY:1 6:5/NUMERIC/SIGNED

/KEY:2 1 :10 /FORMAT:E10.5

I
7

133 1

•

CODE 1 LSCW
/KEY:1 : 7/ALPHA

127

128

~BL OCK

BOUNDARY

..-'

/KEY:129:1

129

1234567890

IKEY: 641:10/COMP

/KEY:130:1

130

98765.43212

/KEY:6 46:10/COMP1

/KEY:131:2

131

500400300,200101

134

/KEY:6 51: 15 /COMP1

132

CODE 3 LSCW

L-I=BLANK
MR-S-"I741-81

SORT/MERGE Command Strings:
Standard Binary:
*SORTEO.FIL=DATA.FIL/FDRTRAN/RANDDM/BINARY/RECDRD:132 -m
#/KEY: 129: 1m

Mixed-Mode Binary:
*SORTED.FIL=DATA.FIL/FORTRAN/RANDOM/BINARY/ASCII/RECORD:GGO -m
#/KEY:G41:10/COMPUTATIONALm

4-32

File Formats

NOTE
To sort the file on the various fields of the record, replace the
shaded portion of the standard binary command string with
one or more of the key specifications given on the left-hand
side of Figure 4-14. Likewise, replace the shaded portion of
the mixed-mode binary command string with one or more of
the key specifications given on the right-hand side of Figure
4-14.
FORTRAN users can specify the file formats illustrated in Figure 4-14 as
follows:
Standard FORTRAN Random Binary:
1.

IFORTRAN/RANDOM/BINARY

IFORTRAN/FIXED/BINARY

Mixed-Mode FORTRAN Random Binary:
1.

IFORTRAN/RANDOM/BINARY/ASCII

IFORTRAN/FIXED/BINARY/ASCII

4.6.2.2 FORTRAN Random Binary (without LSCWs) PROGRAM:
DIMENSION A( 128)
DOUBLE PRECISION Y; REAL X: INTEGER J
An) = 'ABCDE'; A( 2) = 'FG'
A(3)='LM135'
A(4) = '-1234'
A(5)='.1256'; A(6)='7E+03'
J
= 1234567890
X = 98765.43212
Y = 500400300.200101
I
=1
OPEN (UNIT= 1,DEVICE ='DSK',FILE = 'DATA.FIL',MODE
1ACCESS = 'RANDOM',RECORDSIZE = 132)
WRITE (1#I)A,J,X,Y
CLOSE (UNIT= 1)
END

-c-

'IMAGE',

Figure 4-15 illustrates the record produced by the program shown above:

File Formats

4-33

Figure 4-15:

FORTRAN Standard and Mixed-Mode Random
Binary

STANDARD FORTRAN SEQUENTIAL BINARY

MIXED·MODE FORTRAN SEQUENTIAL BINARY

/FOR TRAN/SEQU ENTIA L!BINAR Y

/FORTRAN/SEQUENTIALIBINAflY IASCII

WORD

CAN'T BE SORTED
AS CHARACTERS

L....J

/KEY:11 : 5/ALPHA

/KEY:16: 5/NUMERIC/SIGNED

/KEY:21 : 10/FORMAT:E10.5

I ...

128

CODE 1 LSCW

/KEY:1:7 /ALPHA

J
0

0
__ .. __ 0_-

~BL OCK BOUNDARY

2
128
/KEY: 129: 1

...

129

1234567890

/KEY:641 :10/COMP

98765.43212

/KEY:646 :10/COMP1

_ . _ - - _..••_ - - . _ - - - /KEY: 130: 1

130

/KEY:131:2

131

CODE 2 LSCW

'--

500400300.200101

/KEY:651 :15/COMP1

132
3
L....J

135

...

CODE 3 LSCW

= BLANK
MR·S-1742-81

SORT/MERGE Command Strings:
Standard Binary:
*SORTED.FIL=DATA.FIL/RANDOM/BINARY/RECORD:132 -®ill
#JKEY: 129: 1m

Mixed-Mode Binary:
*SORTED.FIL=DATA.FIL/RANDOM/BINARY/ASCII/RECORD:GGO -m
#JKEY:G41: 10/COMPUTATIONAL(RET)

4-34

File Formats

IRANDOM/BINARY

2. IFIXED/BINARY
3. IBINARY(!FIXED in effect by default)
Mixed-Mode Random Binary:
1.

IRANDOM/BINARY/ASCII

2. IFIXED/BINARY/ASCII
4.6.2.3 FORTRAN Sequential Binary (with lSCWs) PROGRAM:
DIMENSION A(l28)
DOUBLE PRECISION Y; REAL X; INTEGER J
A(l)='ABCDE'; A(2)='FG'
A(3) = 'LMI35'
A(4) = '-1234' .
A(5)='.1256'; A(6)='7E+03'
J
= 1234567890
X = 98765.43212
Y = 500400300.200101
OPEN (UNIT = I,DEVICE = 'DSK' ,FILE = 'DATA.FIL' ,MODE = 'BINARY',
lACCESS = 'SEQOUT')
WRITE (l)A,J,X,Y
CLOSE (UNIT = 1)
END

Figure 4-16 illustrates the record produced by the program shown above:

File Formats

4-35

Figure 4-16:

FORTRAN Standard/Mixed-Mode Sequential Binary
with LSCWs

STANDARD RANDOM BINARY

MIXED-MODE RANDOM BINARY

IRANDOM/BINARY

IRANDOM/BINARY/ASCII

WORD
A

CAN'T BE SORTED
AS CHARACTERS

L-.I

/KEY:1:7/ALPHA
L-.I

/KEY:11 :5/ALPHA

/KEY:16:5/NUMERIC/UNSIGNED

/KEY:21 :10/FORMAT: E10,5

r----~~_~___~___~_3__~1

128

________ BLOCK BOUNDARY

/KEY:129:1

1234567890

/KEY:641 : 10/COMP

98765.43212

/KEY:646:10/COMP1

500400300.200101

/KEY:651: 15/COMP1

129

-.
/KEY:130:1

130

.....

/KEY:131 :2

131

-132

-.-~-"-

[1.......,1=

B LAN K

MR-S-1744-81

SORT/MERGE Command Strings:
Standard Binary:
*SORTED.FIL=DATA.FIL/FORTRAN/SEOUENTIAL/BINARY/RECORD:132-ru
#/KEY: 128: 1ru

Mixed-Mode Binary:
*SORTED.FIL=DATA.FIL/FORTRAN/SEOUENTIAL/BINARY/RECORD:GGO -ru
#/KEY:G41:10/COMPUTATIONAL(REJ)

4-36

File Formats

FORTRAN users can specify the file formats illustrated in Figure 4-16 as
follows:
Standard FORTRAN Sequential Binary:
1. IFORTRAN/SEQUENTIAL/BINARY
2. IFORTRAN/V ARIABLE/BINARY
Mixed-Mode FORTRAN Sequential Binary:
1. IFORTRAN/SEQUENTIAL/BINARY/ASCII

IFORTRAN/VARIABLE/BINARY/ASCII

There are no control
words or end-of-record control characters in this format. Therefore, if the
records are of variable length, it is impossible for SORT/MERGE to extract
a record length for each record. Thus, you cannot correctly sort this file
format unless you observe the following restrictions:
4.6.2.4 FORTRAN Sequential Binary (without LSCWs) -

You must ensure that all records are actually of fixed length.

You must specify IRANDOM (or IFIXED), even though the
MODE = ~SEQOUT' statement is used in the FORTRAN program.

PROGRAM:
DIMENSION A(128)
DOUBLE PRECISION Y; REAL X; INTEGER J
A(1)='ABCDE'; A(2)='FG'
A(3) = 'LMI35'
A(4) = '-1234'
A(5)='.1256'; A(6)='7E+03'
J
= 1234567890
X = 98765.43212
Y = 500400300.200101
OPEN (UNIT = I,DEVICE = 'DSK',FILE = 'DATA.FIL',MODE = 'IMAGE',
lACCESS = 'SEQOUT')
WRITE (1 )A,J ,X, Y
CLOSE (UNIT = I,DEVICE = 'DSK',FILE = 'DATA.FIL')
END

Figure 4-17 illustrates the record produced by the program shown above:

File Formats

4-37

FORTRAN users can specify the file formats illustrated in Figure 4-17 as
follows:
Standard Sequential Binary:
1. IRANDOM/BINARY
2.

IFIXED/BINARY

IBINARY(!FIXED in effect by default)

Mixed-Mode Sequential Binary:
1.

IRANDOM/BINARY/ASCII

IFIXED/BINARY/ASCII

IBINARY/ASCIIUFIXED in effect by default)

File Formats

4-39

Chapter 5
SORT/MERGE Error Messages

5.1

Message Format
SORT/MERGE messages have the following format:
1.

?SRTxxx

text

%SRTxxx

text

[SRTxxx

text

$SRTxxx

text

where:
?
%
[

$
SRT
xxx
text

= fatal error - program must be restarted
= warning message - program continues
= informational message - program continues
= operator intervention message - program continues when operator has replied
= SORT/MERGE mnemonic
= 3-letter code for the message
= explanation of the message

The following section contains the messages that SORT/MERGE generates.
The messages are in alphabetic order by the error code.

5-1

5.2 Error Messages
?SRT2Nl

TOPS-20 version of SORT/MERGE will not run on
TOPS-IO.
This version of SORT/MERGE was compiled for a
TOPS-20 system. It does not run on a TOPS-IO system.

%SRTALN

ANSI label not written.
No system tape label processor is present and, therefore,
ANSI labels have been skipped over, but not checked.

%SRTANL

ANSI label not checked.
No system tape label processor is present and, therefore,
no ANSI labels have been checked.

?SRTARL

ASCII record length incorrect.
One of the following occurred:

%SRTATD

You supplied an incorrect record length by including
end-of-line characters or by not including sequence
numbers in the record length count.

The record is not ASCII.

Attempt to use temporary device failed-ignoring it.
SORT/MERGE was unable to write a temporary file on
the device you specified by means of the /TEMP switch.
SORT/MERGE ignores this device.

?SRTATF

At least two input files required for MERGE.
You specified the /MERGE switch, but only specified one
input file. A merge is performed on two or more input
files.

?SRTBNV

BINARY mode does not support variable length records.
Variable-length records are not allowed in binary files,
except for FORTRAN binary.

?SRTCFE

Collating sequence input file error.
SORT/MERGE encountered an error while trying to read
your collating sequence file. Try rerunning the sort.

?SRTCFS

Collating sequence file specification in error.
You gave an incorrect file specification for the /COLLATE
file. Verify that your specification is correct.

5-2

SORT/MERGE Error Messages

%SRTCLC

Cannot lower core after SORT.
In a COBOL sort, someone expanded memory during a
sort or merge. SORT/MERGE cannot, in this case, free the
memory it was using, but otherwise continues normally.

?SRTCLS

Collating sequence literal specification in error.
Your /COLLATE:/LITERAL: switch argument is incorrect. Check for errors and respecify the command string.

?SRTCND

Collating sequence not defined.
In a FORTRAN-called sort, you specified an argument
with the /COLLATE switch that SORT does not recognize.

?SRTCSD

Cannot set density to ~nnn'.

•

The specified device cannot be set to density ~nnn'.
?SRTCSM

Cannot set hardware data mode on ~~device".
The specified ~~device" cannot be set to industrial compatible mode.

?SRTCTL

Command string too long.
The command string to
(5 * 128 **-2) characters.

?SRTCWB

FSORT

longer

than

CaMP key must be on word boundary.
You specified a key starting position that did not begin on
a word boundary. All CaMP and COMP1 keys must begin
on a word boundary.

?SRTDDV

Double device illegal.
In a FORTRAN-called sort, you specified more than one
device name for the same file specification.

?SRTDFK

Data mode switches must follow a /KEY switch.
In a FORTRAN-called sort, the data type switch must
follow the key switch.

?SRTDND

Device ~dev' is not a disk. All scratch devices must be
disks.
You specified a nondisk temporary device. This is illegal.

?SRTDNE

Device ~dev' does not exist.
You specified a nonexistent device.

SORT/MERGE Error Messages

5-3

•

?SRTELN

EBCDIC tape labels not supported.
No system tape label processor is present, and therefore
SORT/MERGE cannot process EBCDIC labels.

?SRTFCE

Fortran command error.
This is an internal FSORT error. Please contact your software specialist or send an SPR.

?SRTFCN

Attempt to free an 110 channel not retained or released.
This is an internal SORT/MERGE error and is not expected to occur. Please contact your software specialist or
send an SPR.

%SRTFCR

CORE MANAGEMENT ERROR AT RELSPC.
This is an internal SORT/MERGE error and is not expected to occur. Please contact your software specialist or
send an SPR.

?SRTFCW

FORTRAN binary control word incorrect.
You specified /FORTRAN/BINARY and SORT/MERGE
checked for Logical Segment Control Words (LSCWs).
However, SORT/MERGE could not find any or all valid
LSCWs. Check that you have specified a file that actually
contains LSCWs and that you have properly specified
/RANDOM or /SEQUENTIAL.

?SRTFEA

Formal argument count exceeds actual argument count.
The argument referenced by ~n exceeds the argument supplied by the user.

?SRTFMI

Output switch illegal in input file.
You specified an output-only switch such as /ESTIMATE
(SCAN switch), on the input side of the command string.

?SRTFMO

File switches illegal in output file.
You specified an input-only switch such as /TEMP, on the
output side of the command string.

?SRTFMR

Attempt to free more memory than was originally
retained.
This is an internal SORT/MERGE error and is not expected to occur. Contact your software specialist or send
an SPR.

5-4

SORT/MERGE Error Messages

?SRTFNT

Filename may not be specified with /TEMP device.
You have specified a file name with a temporary device,
such as DSK:FOO.BAR/TEMP. The /TEMP switch can
only be specified with a device name.

?SRTFSA

/FORMAT switch argument error.
Verify that the argument you specified with the
/FORMAT switch is a valid format descriptor accepted by
SORT/MERGE.

?SRTFSM

/FORMAT switch must be preceded by /KEY switch.
You can only specify the /FORMAT switch after the /KEY
switch it is intended to modify.

?SRTFUF

FILOP. function failed for file.
This error message is followed by the ?SRTLRE
LOOKUP/ENTER error message (see Table 5-1).

%SRTIBL

IBM label not checked.

No system tape label processor is present and, therefore,
no IBM labels were checked.
?SRTICS

Illegal user supplied collating sequence.
SORT/MERGE discovered illegal characters in your collating sequence. Check the sequence for errors.

?SRTIDS

Illegal/DENSITY: value specified.
You specified an illegal value with the /DENSITY switch.

?SRTIEC

Input error from indirect command file.
SORT/MERGE received an error while trying to read the
indirect command file passed to it by a FORTRAN
program.

?SRTIIF

Illegal indirect filespec.
The indirect command file specification
Check it for errors.

?SRTILC

incorrect.

Illegal character x in numeric field.
A nonnumeric character 'x' has been found in a field described as numeric.

%SRTILN

IBM label not written.
No system tape label processor is present and therefore,
IBM labels were skipped over on input and are not written on output.

SORT/MERGE Error Messages

5-5

?SRTINA

KI-IO version of SORT will not run on KA-IO.
This version of SORT/MERGE was compiled for a KI-IO.
You cannot use it on a KA-IO.

?SRTINS

Input file not specified.
You neglected to specify an input file.

?SRTIRE

Input read error, status 'xxxxxx'.
SORT/MERGE got error code 'xxxxxx' from the monitor
GETSTS UUO while trying to read a file. See the
TOPS-10 Monitor Calls Manual for the meaning of this
code.

?SRTJAL

Junk in ASCII line.
In a fixed-length ASCII file, the characters after the record were not CR, LF, FF, or VT. You may have specified
/FIXED for a variable-length file, or you may have specified the wrong record size.

?SRTKAI

Key argument incorrect.
You specified an argument other than ASCENDING or
DESCENDING with the /KEY switch.

?SRTKCB

Key comparison code too big.
The 2000 words allocated for key extraction and/or key
comparison code has been exceeded. Rebuild SORT with a
larger FTZXSZ internal parameter.

?SRTKEB

Key extraction code too big.
Same as error message ?SRTKCB.

?SRTKLR

Key length required.
You must specify both key length and key starting position with the /KEY switch.

?SRTKNR

Key not fully contained in record.
You have given a key starting position and length to a
variable length file such that a record does not fully contain the key.

?SRTKOR

Key outside of record.
You specified a key that extends outside the record (that
is, extends past the value set with the /RECORD switch).
If the data type is ALPHA and your file is
variable-length, the key you specified can extend past the
smaller records, but in no case can a key extend past the
/RECORD value.

5-6

SORT/MERGE Error Messages

?SRTLNC

LABEL not correct for (filespec'.
The label on the tape does not correspond to the file specification given. Verify that you mounted the correct tape.

?SRTLNI

KL-10 version of SORT/MERGE will not run on KI-10 or
KA-10.
This version of SORT/MERGE was compiled for a KL-10.
You cannot run it on a KI-10 or a KA-10.

?SRTLRE

ENTER error (n' (filespec'
LOOKUP
RENAME
DELETE
An ENTER, LOOKUP, RENAME, or DELETE error occurred for file (filespec'. The error code (n' can be found in
Table 5-1, Error Codes, located at the end of this chapter.

$SRTLRI

Load reel (n' of output file (filespec' type CONTINUE
when ready.
SORT/MERGE is ready to read the records from the next
reel of a multireel input file. After the tape is properly
mounted, type CONTINUE, and SORT/MERGE resumes
reading the file.

$SRTLRO

Load reel (n' of output file (filespec' type CONTINUE
when ready.
SORT/MERGE requires an additional magnetic tape to
continue writing the output file. When the new tape is
properly mounted, type CONTINUE, and SORT/MERGE
resumes writing the file.

?SRTMCS

Multiple collating sequences not allowed.
You can only specify one collating sequence.

?SRTMGF

Monitor GETTAB failed nnnnnn.
This is an internal software error and is not expected to
occur. Contact your software specialist or send an SPR.

?SRTMOI

Multiple output specs are illegal.
You can specify more than one output file specification
only if you are doing output to a magnetic tape.

?SRTMOM

Multiple output specs only on magtapes.
You can specify more than one output file specification
only if you are doing output to a magnetic tape.

SORT/MERGE Error Messages

5-7

%SRTMRS

MERGE record (n' not in sequence for (filespec'.
SORT/MERGE has detected an out-of-sequence record in
an input file during execution of the /MERGE switch
when /CHECK was specified.

?SRTMSC

Mode switch conflict.
You may have specified two contradictory recording mode
switches, such as /ASCII/EBCDIC. You can only pair recording mode switches for mixed-mode binary
/BINARY/ASCII,
/BINARY/SIXBIT,
or
/BINARY/EBCDIC. Or, you may have specified a data
type that conflicts with the recording mode switch, such
as using /COMP3 (or /PACKED) with /ASCII.

?SRTMSD

Multireel tape files with other than STANDARD or DEC
labels not supported.
SORT/MERGE has filled up your output tape and needs
another, but only tapes with STANDARD or DEC labels
can be multi reel. Use a larger tape, or a tape with one of
these label types.

?SRTMTE

Max Temp Files must be in the range 3 to 26.
The /MAXTEMP switch must have a value between 3
and 26.

?SRTMUF

Monitor UUO failed xxxxxx.
This is an internal SORT/MERGE error and is not expected to occur. Contact your software specialist or send
an SPR.

%SRTNCS

Not enough core specified.
The amount of memory that you specified with the /CORE
switch was insufficient. SORT/MERGE uses the mInImum amount of memory for the sort or merge.

?SRTNDV

Null device illegal.
The null device is illegal in a /COLLATE: switch.

?SRTNEC

Not enough core.
SORT/MERGE did not have enough memory to perform
the sort. Try specifying more memory with the /CORE
switch. Also, you may need to have your core limits increased by the system administrator.

SORT/MERGE Error Messages

?SRTNEH

Not enough 110 channels for SORT/MERGE.
SORT/MERGE cannot get enough 110 channels from the
OTS. Check your FORTRAN or COBOL program and
close some files before calling SORT/MERGE.

?SRTNFS

No filename specified for labeled tape ~filespec'.
If you specify the /LABEL switch for a magnetic tape,
then you must include a file specification which agrees
with the file sI?ecification on the tape label.

%SRTNLN

Non-standard label not written.
When you specify /LABEL:NONSTANDARD,
SORT/MERGE skips over nonstandard labels on input
and does not write them on output.

%SRTNLS

Not enough leaves specified.
You specified a value with the /LEAVES switch that was
too small. SORT/MERGE uses the minimum tree size of
16 leaves.

?SRTNRL

Name required with labeled magtape.
If you specify the /LABEL switch for a magnetic tape,
then you must include a file specification which agrees
with the file specification on the tape label.

%SRTNSL

N on-standard label not checked.
SORT/MERGE does not support nonstandard tape labels.

?SRTNSW

No temporary device is writeable.
None of the temporary devices you specified is available.

?SRTOFF

OPEN failed for (filespec'.
SORT/MERGE could not open the specified file. Verify
that the file you specified actually exists.

?SRTOKR

At least one key is required.
You neglected to specify at least one key in your command string.

?SRTONS

Output file not specified.
You neglected to specify an output file.

?SRTOOF

Only one /FORMAT switch per /KEY switch.
You specified more than one /FORMAT switch with the
same /KEY switch. This is illegal.

SORT/MERGE Error Messages

5-9

?SRTOPF

OPEN or LOOKUP failure for indirect command file.
Verify that the command file you specified actually exists.

?SRTOWE

Output write error, status xxxxxx.
SORT/MERGE got error code 'xxxxxx' from the monitor
GETSTS UUO while trying to read a file. See the
TOPS-10 Monitor Calls Manual for the meaning of this
code.

?SRTPRI

Priority must be in range -3 to + 3.
The argument of the /PRIORITY: switch must be in the
range -3 to + 3.

?SRTRIE

Record incomplete at E-O-F.
An end-of-file was encountered in the middle of a record.
The input file is probably damaged or in the wrong
format.

?SRTRLO

RELEASE called out of sequence. SORT not active.
A COBOL program executed a RELEASE verb when a
sort or merge was not in progress.

%SRTRNI

Record number inconsistent, ~n' read, ~m' written.
This is an internal SORT/MERGE error and is not expected to occur. Contact your software specialist or send
an SPR.

?SRTROS

Reel no. ~n' out of sequence for ~filespec'.
The operator mounted the wrong reel of a multireel file.

?SRTRSR

Record size required.
You neglected to specify the record size.

%SRTRTI

Record truncation on input.
One or more variable-length records were longer than the
record size specified with the /RECORD switch. The extra
data is lost.

?SRTRTO

RETURN called out of sequence. SORT not active.
A COBOL program executed a RETURN verb when a sort
or merge was not in progress.

?SRTSAT

Standard ASCII requires TU70 drive.
To read/write standard ASCII mode on tape,
SORT/MERGE requires a TU70 tape drive. Other tape
drives do not have the hardware capability for standard
ASCII mode.

5--10

SORT/MERGE Error Messages

?SRTSFD

SFD depth greater than 5.

The SFD nesting is greater than that allowed.
?SRTSFF

Switches must follow filespecs.
In a FORTRAN-called sort, switches must follow the file
specification. They cannot precede them.

?SRTSRM

SORT/MERGE will not run on this machine.

•

This version of SORT/MERGE was compiled for a machine other than the one you are using.
?SRTSRS

Spanned records not supported.
The Record Descriptor Words (RDWs) in your
variable-length EBCDIC file are nonzero. This indicates
spanned variable-length EBCDIC records, a file format
that SORT/MERGE does not support.

?SRTSSE

Swi tch scanning error.
This is an internal SORT/MERGE error and is not expected to occur. Contact your software specialist or send
an SPR.

?SRTSVR

Switch value required.
You neglected to specify an argument for a switch that
requires one.

?SRTSWP

Temporary structure ~dev' is write-locked.
You specified a write-locked device with the /TEMP
switch, and SORT/MERGE could not use it.

?SRTTMD

Too many digits in key.
You specified a CaMP or COMPl key with more digits
than would fit in two words (for double precision) of
storage.

%SRTTMT

Too many temporary structures specified.
You have specified more temporary devices than
SORT/MERGE is able to use. The extra areas are ignored.

?SRTUDL

Unknown delimiter.
An illegal character was found in the command string.

?SRTUKS

Unknown switch ~/switch'.
You specified a switch that SORT/MERGE does not
recognize.

SORT/MERGE Error Messages

5-11

•

?SRTUSV

Unknown switch value.
You specified a switch value
SORT/MERGE does not recognize.

argument

that

Expanding to nK I

rSRTXPN

For stand-alone sorts, this indicates how much memory is
being used for the sort. For COBOL or FORTRAN-called
sorts, this is an internal SORT/MERGE debugging message which only occurs if the feature test switch
FTDEBUG is nonzero.

5.3 Error Codes
The error codes in Table 5-1 are returned in the AC on RUN and GETSEG
monitor calls, in the right half of location E + 1 on 4-word argument
blocks of LOOKUP, ENTER, and RENAME monitor calls, and in the right
half of location E + 3 on extended LOOKUP, ENTER, and RENAME monitor calls.

&-12

Table &-1:

Error Codes

Symbol

Code

Explanation

ERFNFlk

File not found, illegal filename (0,*) file names do not match
(UPDATE), or RENAME after a LOOKUP failed. On a FILOP.,
this error is given if the specified device cannot perform I/O in the
direction indicated.

UFD does not exist on specified file structures. (Incorrect
project-programmer number.)

ERPRT7c

Protection failure or directory full on DTA.

ERFBM'If

File being modified (ENTER. RENAME l.

ERAEFlk

Already existing filename (RENAME or FILOP.), different
filename (ENTER after LOOKUP) or supersede (on a nonsuperseding ENTER). Two LOOKUPs or two ENTERs were
performed.

ERISU7c

Illegal sequence of monitor calls (RENAME with neither
LOOKUP nor ENTER. or LOOKUP after ENTERl.

ERTRNlk

One of the following: '
1.

Transmission. device. or data error (RUN, GETSEG only).

Hardware-detected device or data error detected while reading the UFD RIB or UFD data block.

Software-detected data inconsistency error detected while
reading the UFD RIB or file RIB.

ERNSFo/c

Not a saved file (RUN, GETSEG only).

ERNEClk

Not enough core (RUN, GETSEG only).

SORT/MERGE Error Messages

Table 5-1 (Cont.):

Error Codes

Symbol

Code

Explanation

ERDNA%

Device not available (RUN, GETSEG only),

ERNSD%

No such device (RUN, GETSEG only),

ERILU%

Illegal monitor call (GETSEG only). No 2-register relocation
capability.

ERNRM%

No room on this file structure or quota exceeded (overdrawn quota
not considered).

ERWLK%

Write-lock error. Cannot write on file structure.

ERNET%

Not enough table space in free core of monitor.

ERPOA%

Partial allocation only.

ERBNF%

Block not free on allocated position.

ERCSD%

Cannot supersede an existing directory (ENTER).

ERDNE%

Cannot delete a nonempty directory (RENAME).

ERSNF%

Subdirectory not found (some SFD in the specified path was not
found).

ERSLE%

Search list empty (LOOKUP or ENTER was performed on generic
device DSK and the search list is empty).

ERLVL%

Cannot create an SFD nested deeper than the maximum allowed
level of nesting.

ERNCE%

No file structure in the job's search list has both the no-create bit
and the write-lock bit equal to zero, and has the UFD or SFD
specified by the default or explicit path (ENTER on generic device
DSK: only).

ERSNS%

GETSEG from a locked low segment to a high segment which is
not a dormant, active, or idle segment. (Segment not on the swapping space.)

ERFCU%

The file cannot be updated.

ERLOH%

The low segment overlaps the high segment (GETSEG).

ERNLI%

The user is not logged in (RUN).

ERENQ%

The file still has outstanding locks set.

ERBED%

The file has a bad .EXE file directory (GETSEG, RUN).

ERBEE%

The file has a bad extension for an .EXE file (GETSEG, RUN).

ERDTB%

The file's .EXE directory is too big (GETSEG, RUN).

ERENC%

Network capacity has been exceeded, not enough space for connect
message (LOOKUP/ENTER).

ERTNA%

Task not available (LOOKUP/ENTER/RENAME).

ERUNN%

Unknown network node; node went down during connect
(ENTER).

SORT/MERGE Error Messages

5-13

Chapter 6
SORT/MERGE Performance Considerations

This chapter gives a general description of the SORT/MERGE binary tree
and describes how the operation of SORT/MERGE is affected by tree size,
buffer size, and number of merge passes. This is not a complete description
of SORT/MERGE internals. This chapter simply gives you enough information to modify SORT/MERGE's performance by use of the /LEAVES and
/CORE switches.

6.1

Performance Overview
SORT/MERGE consists of two major sets of routines, I/O routines and sorting routines. The I/O routines strip off header words, control words, and
control characters on input and restore them on output. This is necessary as
only pure data can be passed to the sorting routines. There is nothing that
you can do to improve the performance of the I/O routines. However, if the
file format is not correctly specified, the data is not correctly extracted from
the file; header words, control words, or control characters are erroneously
passed to the sorting routines. Thus, the results are unpredictable.
The sorting modules perform two functions: sorting and merging data.
These two functions are described more fully in the following sections.
Since you cannot actively modify the operation of the merge phase, most of
the discussion is concerned with the sort phase. You can, however, optimize
the number of merge passes that occur. That technique is described in this
chapter.

6-1

SORT/MERGE uses a variable-size binary tree to sort and merge input
files. Figure 6-1 illustrates a simple binary tree:

Figure 6-1:

Binary Tree

8
MR-S-1720-81

The tree size is measured by the number of its leaves and corresponds to the
number of records the tree holds. For example, the tree shown above has
eight leaves, and thus eight records.

NOTE
Although SORT/MERGE's minimum tree size is 16 leaves,
this chapter uses fewer leaves in the examples for
readabili ty.
If you do not specify the size of the tree (with the /LEAVES switch), then
SORT/MERGE picks the largest tree greater than 16 leaves that fits in
your available memory space.

6.1.1

The Sort Phase

The following sections should be read in conjunction with Figure 6-2. This
figure illustrates how a binary tree is used to sort and merge data. To keep
the figure as simple as possible, the following conditions and conventions
have been used:
1.

The sorting tree is limited to four leaves (versus a minimum of 16
leaves for SORT/MERGE).

Letters are used to represent records (not characters) of arbitrary size.
The key field is assumed to be as long as the record.

For the purposes of this chapter, ~~smallest" refers to an ordering that
fully takes into account multiple keys with possibly differing
ASCENDING or DESCENDING key arguments.

4. The leaves (terminal nodes) of the tree are numbered, and these numbers represent the ~~address" of the leaves.
5.

6-2

The upper, or nonterminal, nodes contain pointers to the LOSING key
of any two keys that are compared. An additional ~node', shown outside
the tree, points to the WINNING key in the tree. For example, if the

SORT/MERGE Performance Considerations

records are being sorted in ascending order, then the object is to find
the smallest key in the tree at that time. The nonterminal nodes point
to the larger of any two keys compared, while the external (node' points
to the smallest key in the tree.
6. The sort shown in the figure is limited to a maximum of two temporary
files (rather than the maximum of 26 temporary files for
SORT/MERGE).
The sorting process works as follows: SORT/MERGE fills the tree to its
capacity with records from the input file. (Multiple input files are treated
as one file.) All records in the tree are compared. Then the smallest record
is ejected from the tree and written to a temporary file. The ejected record's
position in the tree is then filled with the next input record, and the process
is repeated. In the special case where the number of records initially in the
tree is less than or equal to the capacity of the tree, ejected records are
written directly to the output file.
SORT/MERGE continues reading records until one is read that is less than
the record just output. This record and all subsequent ones less than the
record currently being ejected from the tree are marked to be sent to a new
run. When one of the records marked for the new run reaches the top of the
tree, the collection of records currently in the temporary file constitutes the
longest sequence of ordered records SORT/MERGE could generate, given
the size of the tree. SORT/MERGE now creates a new run, beginning with
the currently ejected record. The new run is appended to the end of the next
temporary file, after any existing runs. SORT/MERGE begins each run,
except the first in a file, with a run marker containing the run number of
the new run. SORT/MERGE continues sorting records and writing runs
until all records have been read. At this point, the sort phase is over.

6.1.2 The Merge Phase
For the rnerge phase, SORT/MERGE reinitializes the tree to contain as
many leaves as there are temporary files, and then fills the tree with the
first record from each temporary file. The smallest record is then written to
a new temporary file, and another record is read into the tree from the
same file that the currently ejected record was read from. New merge
passes are initiated as long as more than one temporary file remains. When
only one temporary file remains, this file is renamed or copied to the
output file.

SORT/MERGE Performance Considerations

6-3

Figure 6-2: The Operation of the SORT/MERGE Binary Tree
SORT PHASE
.

Inpl~.::I~: I v I w I x I y I MIN 1 0 I P I A I B I

I x Iy

I MIN

1I I I I
0

No. 2

No.1

TMP.l

IN.FIL I v I w

TMP.l

I,--V---,-'_ __

v 1wI

Key M < Key V, New Run Marker Used

IN.FIL

V I w

I I I
X

MIN I 0

I I I I
P

No.4

No. 3

TMP.l

I I
X

y I MIN

I I I I I
0

Ivlwixi

IN.FI L I V I w

NO.5

TMP.l

I Iw I I I

IN.FIL

I V I w I X Iy I MIN 101 P I A I

No. 6

TMP.l I V
TMP.2

IN.FIL I V I w I X I y I MIN 1 0 I P I A I B I

Iw I I
I X

M--,-I_ _ _ _ _ _ _ _ _----'
New Run Begins

,--I

3
TMP.l

I Iw I I y
V

TMP.2 1M 1 N I
Key A < Key N, New Run Marker Used

NOTE: Arrows point to the record(s) currently being read into the tree.
Superscripts preceding each letter are run markers.
MR-S-1721-B1

6-4

SORT/MERGE Performance Considerations

Figure 6-2 (Cont.): The Operation of the SORT/MERGE Binary Tree

No. 7

No. 8

2
TMP.1

I Iw I I

TMP.2

I MIN 1 0 1

I V Iw I x I y I
TMP.2 1M I N 10 I P I

No.9

IVIWlxlylAIB I
Run 1

TMP.2

TMP.1

y I

No. 10

TMP.1

I MIN 1 0
Run 2

TMP.1

IVlwlXlylAI

TMP.2

I MIN 1 0 1 P I

Run 3

I P I

New Run Begins

NOTE: Heavy vertical bars indicate run markers.
Arrows point to the record(s) currently being read into the tree.
Superscripts preceding each letter are run markers.
00

Special record used by SORT/MERGE to flush the tree of records.
Special run marker.
MH-S-1722-81

SORT/MERGE Performance Considerations

6--5

Figure 6-2 (Cont.): The Operation of the SORT/MERGE Binary Tree
FIRST MERGE

MERGE PHASE

Input Files: ,....1-:.....,---r--T""'"___......---..---------.

TMP.1

I Iw I I I I I
It I I
V

Run 1

TMP.2

MIN

TMP.1 I V I w I X I y

Run 3 _ _ _---,

A I B

P I

Run 2

No. 11

No. 12

TMP.3!L-M.-JI'---_ _ _ _ _ _ _-----'

TMP.1

I wI I I
V I

No. 14

TMP.3 I MIN I 0

TMP.3 I MIN I 0 I

TMP.1 I V

Iw I I y I
0 I I

TMP.2 I MIN I

A I B

TMP.1
TMP.2

I
I

V Iw

MIN

P I

I Iy I I I
I0 I
A

P I

No. 16

No. 15

TMP.3 I MIN

I I I wI
0

V I

MR-S-1723-81

6-6

SORT/MERGE Performance Considerations

Figure 6-2 (Cont.): The Operation of the SORT/MERGE Binary Tree

TMP.1 I v Iw I x I y I A IB I

TMP.1 I v Iw I x Iy I A I B I
TMP.2 1M! N! 0 ! PI

No. 17

No. 18

TMP.3 I MIN I 0 I PI v Iw I x I
FINAL MERGE

TMP.1 I v Iw I x I y I A I B I

TMP.1 I v I w I x I y IA I B I
TMP.3 I MIN 1 0 1 P1 v 1 w 1 x 1 y

TMP.3 I MIN I 0 I PI v w Ix I y I
1

No. 19

No. 20

OUT.FIL

A--LI_ _ _ _ _ _ _----'

1...-1

TMP.1 I v Iw I x I y I A I B I

TMP.1 I v Iw x I y I A I B I
1

TMP.3 I MIN 1 0 1 P1 v 1 w 1 x 1 y
No.

TMP.3 IMIN 1 0 1 P1 v 1 w 1 x 1 y

No. 22

MR·S·1724·81

SORT/MERGE Performance Considerations

6-7

Figure 6-2 (Cont.): The Operation of the SORT/MERGE Binary Tree

TMP.1 I v I w I x I y I A I B C-~
TMP.3 I MIN I 0 I PI v Iw I x Iy I
f

TMP.1 I v Iw I x I y I A I B I
TMP.3
No. 24

No. 23

TMP.1 I v I w I x I y I A I B I

TMP.3 I MIN I 0 I PI v Iw Ix Iy I

TMP.1 Iv Iw Ix Iy I A I B I

TMP.3 IMIN I0 I PI v Iw Ix I y I

No. 26

No. 25

TMP.1 Iv Iw I x I y I A I B I

------~

TMP.3 IMIN 10 I PI v I w I x Iy I

TMP.1 I v Iw I x I y I A I B I
TMP.3 I MIN I 0 I PI v I w I x I y I
f

No. 28

No. 27

OUT. F I L IA I B IMIN I0 IPI v Iw I x I y I
MR-S·1725-81

6-8

SORT/MERGE Performance Considerations

6.2 Performance Considerations
There are three major areas of concern In tuning the performance of
SORT/MERGE. They are:
1.

Tree size

Memory size (low segment size)

N umber of merge passes

The following sections describe all of these areas and give general guidelines for adjusting them to improve the performance of SORT/MERGE.
Variables such as installation memory size, system load, file size, file order,
key type and length, data type, and recording mode make it impossible to
give a simple, well-defined formula for improving the performance of a
particular sort. If you are concerned with performance, you should study
the guidelines given below and experiment with them.
If you are doing stand-alone sorts, then SORT/MERGE, after completion of
the sort, prints statistics concerning the sort. (COBOL and
FORTRAN-called sorts do not print these statistics.) Included in these
statistics are the tree size in leaves, the low segment size, and the number
of runs. Thus, you can study these values and decide which must be altered
for performance improvement. This is especially important in sorts that are
run using default values for tree size and memory size. The following illustrates the display of sort statistics:
[SRT>(PN Expanding to 48P]
Sorted 7 records
12 KEY COfTlParisonst 1.71 per record
4 0 r e cor dIe a \,1 e sin III e III 0 r y
o run s
0:00:00 CPU tillle t 18.43 MS per record
0:00:42 elapsed tifTle

6.2.1

Tree Size

During the sort phase, as the tree size increases, fewer runs and temporary
files occur, requiring fewer merge passes later. However, a tree that is too
large places a heavy burden on system resources and causes a subsequent
performance degradation of the sort. A tree that is too small causes excessive merge passes and increases the run time of the sort. The file itself
further affects performance. For a given tree size, a well-ordered file results in fewer (and longer) runs and therefore, fewer merge passes. A
backwards-sorted file results in many runs and merge passes.
The effect of the tree size on runs is as follows:
If N is the number of leaves in the tree, then:
1.

Worst case file -

each run except the last has N records

Random order file -

each run contains approximately 2:;:N records

SORT/MERGE Performance Considerations

&-9

Best case file input file

produces one run equal to the size of the entire

The bias is the ratio of the run to the tree size, and is a useful index of the
randomness of a file. The bias is printed only if at least one temporary file
is written. The following illustrates how to interpret the bias:
bias

Descending file sorted in ascending order (or vice versa)

bias

Randomly ordered files

bias

60% of the records are random, 40% are in order

bias

20% of the records are random, 80% are in order

For example, a bias of 7 indicates that 80% of the records in the file are in
order and each run contains approximately seven times the number of records which would fit in the tree.
SORT/MERGE's default memory algorithm uses the largest tree size larger
than 16 leaves that fits in the memory available.
You can specify the tree size yourself with the /LEA YES switch. However,
you should read Section 6.2.3 before using this switch.

6.2.2 Memory Size
SORT/MERGE's low segment appears as follows during the sort or merge
phase:
Impure
Code
Binary
Tree
Low Segment
Buffer
Space

MR-S-1727-81

Assuming that you have adjusted the tree size to an optimum value, you
should determine how many buffers SORT/MERGE generates for the sort.
A large number of 110 buffers per file causes a memory image bigger than
TOPS-IO can handle. Then the sort places a heavy burden on the system
and causes a subsequent performance degradation. An insufficient number
of 110 buffers causes inefficient 110 data transfer. As SORT/MERGE's default algorithm cannot always use the optimum memory size, the efficiency
of 110 buffer transfer can be improved if you specify a memory size that
causes SORT/MERGE to create a number of buffers more appropriate for
your job.

6-10

SORT/MERGE Performance Considerations

The minimum number of buffers is usually two buffers for each input and
output file. Two buffers allow double buffering; the monitor reads the second buffer while SORT/MERGE is processing the first buffer. For disks, a
good number of buffers is the cluster size of the disk plus one (five is usually adequate). Buffer lengths are as follows:
• disk files -

buffer length is 131 words

• tape files -

buffer length is equal to the size of a physical record plus
three

See the TOPS-10 Monitor Calls Manual for other devices.
Some calculation is involved before you can determine the approximate
number of buffers used in your sort. First, examine the following terms that
are used in the calculations. Unless otherwise noted, all values are
DECIMAL words.
1.

- number of leaves in the tree

- maximum record length in words

- 7680 words (15 pages). This is the length of SORT/MERGE impure code for stand-alone sorts.

- word length of extracted keys (see below)

- 4 words (3 words of tree pointers plus 1 internal count word)

- key length in bytes

- length of low segment in words

- length of low segment in pages

- length of impure code plus length of binary tree

10. b

- length of buffer space

11. t

- size of binary tree in words

12. cl

- disk cluster size plus 1 for nondisk devices, cl = 2

13. m

- maximum of 131 (size of a temporary file buffer) and the maximum buffer length for any input device

The following example explains the calculations you must perform. The
following assumptions are made:
1. The sort uses only disk files having a cluster size of 1.
2.

There are no extracted keys.

The sort does not use the /COLLATE switch.

SORT/MERGE statistics indicate that the tree size is 100 leaves and
the low segment size is 100 pages.

SORT/MERGE Performance Considerations

6-11

The tree size parameter has already been examined and determined to
be optimum.

The record size is eight words.

The first step is to convert the reported low segment SIze to words (in
algebraic form):
lw = (lp * 512)
lw = (100 * 512)
lw = 51200
The second step is to find the value of t, the size of the binary tree (in
algebraic form):
t = (n*(rl+el + p))
t = (100 * (8 I + 0 I + 4))
t = (100 * (12))
t = 1200
The third step is to find the value of s, the tree size plus impure code (in
algebraic form):

s = t + c
s
1200 + 7680
s - 8880
The fourth step is to find the value ofb. the buffer space (in algebraic form):
b = lw - s
51200 - 8880
b
b = 42320

The fifth step is to divide the buffer space by 2. This is because, during the
sort phase, SORT/MERGE. has exactly two files open: the current input file
and a temporary file:
42320/2 = 21160
The sixth step is to divide the buffer size by m. In this case, m

131.

21160/131 = 161.5 or 161 buffers (rounded downward)
Thus, SORT/MERGE used 161 buffers per file. This figure is high, as the
optimum number of buffers is usually two per file, or the cluster size plus
one for disk. The following computation (in algebraic form) gives you the
required buffer space, assuming two buffers per file:
b=cl*m*2
b = 2 * 2 * 131
b = 524 words
If you add this value to the value you previously calculated for s (impure
code pI us tree), you get:
8880 + 524 = 9404 words

6-12

SORT/MERGE Performance Considerations

Thus, lw (length of low segment in words)

9404.

To calculate lp (low segment in pages):
lp = lw 1 512
lp = 9404 1 512
lp = 18.3 rounded up to 19
Thus,

you

would

specify

the

following

switch

combination

SORT/MERGE:
ILEAVES:100/CORE:19P

NOTE
If, after attempting to run the sort with the new ICORE
value, you receive a message indicating that SORT/MERGE
did not have enough core, rerun the sort after adding one or
two pages to the ICORE value. Rounding errors can sometimes cause you to calculate a value that is slightly less than
the required value.
In the example above, you assumed no extracted keys. Thus, the value of e
was zero. If the data type switch is INUMERIC, IFORMAT, ICOMPU, or
ICOMP1, then e = 1 for single precision keys; and e = 2 for double precision
keys.
If you used the ICOLLATE: switch in the command string, then e is calculated by a different method. (Note that k = key length.)
If you used the ICOLLATE:ASCII switch, or the ICOLLATE:FILE: switch,
or the ICOLLATE:/LITERAL: switch, where the file or the literal contained
an alternative ASCII collating sequence, then:
e = (k + 4)/5
If you used the ICOLLATE:EBCDIC switch, or the ICOLLATE:FILE:
switch, or the ICOLLATE:LITERAL: switch, where the file or the literal
contained an alternative EBCDIC collating sequence, then:
e = (k + 3)/4
In summary, the various formulas that you used are as follows:
1.

Convert n pages to words:
n 1 512

Convert n words to pages:
n

* 512

Calculate the number of buffers per file for a given low segment value:

Ilw -

((n :1: (r I +el +

p))

+ c)1

(cl * m)

SORT/MERGE Performance Considerations

6-13

Calculate an optimized nun1ber of buffers:
cl :,: m :,: 2

6.2.3 Number of Merge Passes
The largest number of runs that SORT/MERGE can maintain before doing
a merge pass is an exponential function of the number of temporary files
available. For stand-alone sorts. the number of temporary files is fixed at
26. As merge passes are very expensive in terms of run time, it is important
to ensure that no more merge passes occur than are absolutely necessary.
You should observe the following figures:

RUNS

MERGE PASSES

0 (sort contained entirely in memory)

1 - 26 (26:'::"lJ

26 - 676 (26:'::':2)

676 - 17576 (26::::::~3)

17576 - 456976 (26:::::'4)

The values shown above that are exponential powers of 26 are the merge
threshold values. If the number of runs exceeds any given threshold limit,
then an additional 111erge pass is necessary. This is particularly unfortunate in cases where the number of runs is only 10 to 20(k greater than
the threshold limit. As an example, consider the case of a sort that uses 29
runs. This is only 3 runs above the closest threshold value (26). For the
sake of 3 runs, a very substantial performance degradation is incurred.
Thus, a slight increase in the size of the sort tree might reduce the number
of runs to, or below, the 26-run threshold value and eliminates the need for
an entire merge pass. However, a slight increase in memory usage gives a
substantial reduction in the run time of a sort. Therefore, the tree size is
specified with the ILEA YES switch. It is best to increase the size of the tree
until the number of runs equals, or drops below. the nearest threshold
value.
Alternately, you may be more concerned with memory usage than with run
time. If the number of runs is 10 - 20';' less than the nearest merge threshold value, you can effect a gain in system performance (at a slight expense
in run time for the sort) by specifying a smaller tree. Again, decrease the
tree size until the nUlllber of runs is equal to or less than the nearest merge
threshold value.

NOTE
The length of each run is data dependent. What works one
time may be incorrect the next time if the data is different.

6-14

SORTIMERGE Performance Considerations

Appendix A
Summary of SORT/MERGE Functions and Switches

This appendix has alphabetic lists of SORT/MERGE functions and
SORT/MERGE switches. Each command or switch is described briefly.
Complete descriptions of functions can be found in Chapter 2. Complete
descriptions of switches can be found in Chapter 3.
Each switch is given a brief description that includes its function, its type,
and any defaulting characteristics of the switch.

A.1

Functions
There are five major functions in SORT/MERGE:
• EXIT

- exit from SORT/MERGE and return to TOPS-IO command
level.
Use /EXIT

• HELP

- print the text of the help file.
Use /HELP or /HELP:SWITCHES

• MERGE - merge previously sorted files.
Use /MERGE with a SORT/MERGE command string
• RUN

- run a program.
Use /RUN:

• SORT

- sort the specified files.
Simply specify a SORT/MERGE command string

A-I

A.2 Switches (in Alphabetic Order)
IAFTER

Place the output record after the carriage-return/line-feed characters.
Function: File switch
Type:
Local
Default: IBEFORE switch
IALIGN

Word align all ASCII output records.
Function: File switch
Type:
Global
Default: Not word-aligned
Exceptions:
1. I ASCII/FORTRAN is word-aligned by default.

Line-sequenced ASCII input files are detected by SORT/MERGE, and
are word-aligned on output.

IALPHA

The key data type is alphanumeric.
Function: Key data type switch
Type:
Local
Default: If ISIGNED, IUNSIGNED, IFORMAT, INUMERIC, ICOMPU,
ICOMPl, or ICOMP3 not specified, this is the default data type.
IASCII

The recording mode of the data is ASCII.
Function: Recording-mode switch
Type:
Global
Default: See the explanation of this switch in Chapter 3 for a description
of the defaults.
IBEFORE

Place the output record before the carriage-return/line-feed characters.
Function: File switch
Type:
Local
Default: This is the default.

A-2

Summary of SORT/MERGE Functions and Switches

/BINARY
The recording mode of the data is binary.
Function: Recording-mode switch
Type:
Global
Default: See the explanation of this switch in Chapter 3 for a description
of the defaults.
/BLOCKED:n
The n is a decimal number indicating the blocking factor for a COBOL file.
Not meaningful for FORTRAN-created files.
Function: File switch
Type:
Modified position dependent
Default: Unblocked file
/CHECK
Check the record sequence of files that are being merged and generate an
error message if out-of-sequence records are found. This switch doubles the
number of comparisons and causes some increase in run time. However, it
protects against erroneously merging unsorted files. Valid only for the
merge function.
Function: Control switch
Type:
Global
Default: Don't check merge input files
/COLLATE:n
The argument to this switch (nl specifies the collating sequence to use
when sorting the associated files. The possible arguments are:
1.

ASCII

EBCDIC

FILE:filespec

LITERAL:/collating-sequence/

ADDRESS:address

Arguments 1 and 2 allow you to sort ASCII data according to the EBCDIC
collating sequence or vice versa. Argurnents 3 and 4 allow you to specify
your own collating sequence. Argument 5 is for FORTRAN users in the
format of ADDRESS:A n .
Function: Control switch
Type:
Global
Default: The collating sequence associated with the recording-mode
switch used.

Summary of SORT/MERGE Functions and Switches

A-3

/COMPU
The key data type is COMPUTATIONAL.
Function: Key data type switch
Type:
Key modifier
Default: If the recording mode is binary. this is the default data type.
/COMPI
The key data type is COMPI.
Function: Key data type switch
Type:
Key modifier
Default: No default characteristics
/COMP3
The key data type is COMP3.
Function: Key data type switch
Type:
Key modifier
Default: No default characteristics
/CORE:n
The n is the amount of the low segment to be allocated to SORT/MERGE.
Function: Control switch
Type:
Global
Default: Determined by SORT/MERGE's default core algorithm
/DENSITY:n
The n is a decimal number indicating the recording density to be used for
reading or writing a magnetic tape. This switch is ignored for other devices.
The possible values of n are:

200

556

800

1600

6250

Function: Tape switch
Type:
Position dependent (SCAN switch)
Default: Set by SET DENSITY monitor command

A-4

Summary of SORT/MERGE Functions and Switches

/EBCDIC
The recording mode of the data is EBCDIC.
Function: Recording-mode switch
Type:
Global
Default: See the explanation of this switch in Chapter 3 for a description
of the defaults.
/ERROR:addr
The addr is the absolute octal address that program control transfers to
when SORT/MERGE encounters a fatal error. This switch is designed for
FORTRAN users, who can specify an absolute address or (indirectly) a
symbolic address. See the description of this switch in Chapter 3 for further
details.
Function: Control switch
Type:
Global
Default: Stop the job on a fatal error
/EXIT
Exit to monitor command level.
Function: Control switch
Type:
Global (SCAN switch)
Default: Prompt for another SORT/MERGE command string
/FATAL:addr
The addr is the absolute octal address that the SORT/MERGE error code is
stored in when SORT/MERGE encounters a fatal error. This switch is designed for FORTRAN users, who can specify an absolute address or (indirectly) a symbolic address. See the description of this switch in Chapter 3
for further details.
Function: Control switch
Type:
Global
Default: Do not return a fatal error code to the program
/FIXED
The file contains fixed-length records.
Function: File switch
Type:
Global
Default: Depends on the recording-mode switch you specify. See the description of the appropriate recording-mode switch in Chapter 3
for more details.

Summary of SORT/MERGE Functions and Switches

A-5

IFORMAT:nPaw.d
The key data type is FORTRAN ASCII-NUMERIC.
nP =
a

w =
d

Scaling factor (positive or negative number)
D decimal floating point, double precision
E decimal floating point, single precision
F decimal floating point, single precision
G general
key length
number of decimal places in key

Function: Key descriptor switch
Type:
Local
Default: No default characteristics
IFORTRAN

The associated files are FORTRAN ASCII or FORTRAN binary files.
Function: File switch
Type:
Global
Default:- Not IFORTRAN
IHELP

Print the help file. IHELP:SWITCHES prints switches available to
SORT/MERGE users.
Function: Control switch
Type:
Global (SCAN switch)
Default: No default characteristics
IINDUSTRY
Read or write a magnetic tape in industry-compatible mode. This switch is
ignored for devices other than tape drives.
Function: Tape switch
Type:
Modified position dependent (SCAN switch)
Default: DEC format
IKEY:n:m:x

The n is the position of the first character of the key. The m is the length of
the key. The x is either ASCENDING or DESCENDING.
Function: Key switch (required)
Type:
Global
Default: Default switches are:
1. IALPHANUMERIC
if ISIGNED, IUNSIGNED,
INUMERIC, ICOMPU, ICOMPl, ICOMP3, or IFORMAT is
not specified.

A-6

Summary of SORT/MERGE Functions and Switches

/NUMERIC - if /SIGNED or /UNSIGNED is specified and if
/ALPHA, /COMPU, /COMP1, or /COMP3 is not specified.

/SIGNED - if /COMPUTATIONAL, /COMP1, /COMP3,
/NUMERIC, or /FORMAT is specified.

/LABEL:n
Specifies whether or not the file contains tape labels, and, if so, what type of
tape labels. This switch is ignored for devices other than tape drives. Possible arguments are:
1.

ANSI

DEC

IBM

NONSTANDARD

OMITTED

STANDARD

Function: Tape switch
Type:
Modified position dependent (SCAN switch)
Default: /LABEL:STANDARD
/LEAVES:n
Specifies the size of the binary tree that SORT/MERGE is to use during the
sort phase of a sort. The argument n is a decimal number that indicates
how many leaves (or terminal nodes) that the binary tree is to have. This
switch does not affect the merge phase of a sort, and is ignored when used
with the /MERGE switch.
Function: Control switch
Type:
Global
Default: Determined by SORT/MERGE's default memory allocation algorithm. See Chapter 6 for more details.
/MAXTEMP:n
Indicates the maximum number (n) of temporary files that can be used
during a sort or merge.
Function: Control switch
Type:
Global
Default: 26 temporary files
/MERGE
Merge the specified input files.
Function: Control switch
Type:
Global
Default: No default characteristics
Summary of SORT/MERGE Functions and Switches

A-7

IMESSAGE:arg
Controls how much of an error, informational or warning, message IS
printed. The possible arguments are:
1.

rN01FIRST - determines whether or not the text portion of the message is printed.

rN01PREFIX determines whether the
(i.e., SRTxxx) is printed with the message.

error

message

code

You can specify the arguments in combination. For example,
IMESSAGE:(NOPREFIX,FIRST) omits the error code, but prints the text of
the message.
Function: Control switch
Type:
Global
Default: IMESSAGE:(PREFIX,FIRST)
INOCRLF

Indicates that both input and output files are IFIXED IASCII records containing no carriage control characters.
Function: File switch
Type:
Global
Default: Both input and output records have a carriage-return line-feed
characters.
INUMERIC
The key data type is numeric.
Function: Data type switch
Type:
Key modifier switch
Default: Causes ISIGNED to
IUNSIGNED).

effect

(unless

you

specify

IOPTION:name
Reads switches from lines beginning with SORT:name in SWITCH.INI.
Function: Control switch
Type:
Global (SCAN switch)
Default: Read switches from lines beginning with SORT (unlabeled) in
SWITCH.INI.
IPACKED

The key data type is packed decimal (4 bits). This switch is identical to
ICOMP3.
Function: Key data type switch
Type:
Key modifier
Default: No default characteristics

A-8

Summary of SORT/MERGE Functions and Switches

/PARITY:n
Specifies the parity (n) to be used when reading or writing a magnetic tape.
The possible values for the parity argument (n) are:
1.

EVEN

ODD

Function: Tape switch
Type:
Position dependent '(SCAN switch)
Default: System default
/PHYSICAL
Suppresses local name assignments; only physical device names are
searched to find the data file.
Function: Control switch
Type:
Position dependent (SCAN switch)
Default: Search logical names first
/POSITION:n
Positions a magnetic tape before a file is read or written. The argument (n)
can be positive to skip a number of files or negative to backspace a number
offiles. This switch is ignored for devices other than magnetic tape drives.
Function: Tape switch
Type:
Local
Default: Position at the first file on the tape.
/PRIORITY:n
Sets the disk priority for your SORT/MERGE job. The argument (n) can be
a value from -3 to 3. An argument of + 3 is the highest priority; -3 is the
lowest priority.
Function: Control switch
Type:
Position dependent
Default: Normally o. The default can be changed with the DISK. monitor
call. (See the TOPS-J 0 Monitor Calls Manual.)
/RANDOM
The associated file is a random-access file. This switch IS identical to
/FIXED.
Function: File switch
Type:
Global
Default: Depends on the recording-mode switch you specify. See the description of the appropriate recording-mode switch in Chapter 3
for more details.

Summary of SORT/MERGE Functions and Switches

A-9

IRECORD:n

The n is a decimal integer specifying the record size.
Function: Record length switch (required)
Type:
Global
Default: None. This switch is required.
IREWIND

Rewind the tape before the file is read or written. This switch is ignored for
devices other than tape drives.
Function: Tape switch
Type:
Local
Default: Don't rewind the tape before use.
IRUN :filespec

Run the specified program. Three subordinate switches can also be specified with this switch:
1. IRUNCORE:n -

Run the program with a memory allocation of ~n'.
Run the program at the start address + n.

IRUNOFFSET:n -

ITMPFIL:nam:~string' -

Store the string in TMPCOR core file 'nam'.
The ~nam' can be a maximum of three characters.

Function: Control switch
Type:
Global
Default: No default characteristics
ISEQUENTIAL
The file is a sequential-access file. This switch is identical to IV ARIABLE.
Function: File switch
Type:
Global
Default: Depends on the recording-mode switch you specify. See the description of the appropriate recording-mode switch in Chapter 3
for more details.
ISIGNED

The key data is signed.
Function: Key data type switch
Type:
Key modifier switch
Default: In effect for INUMERIC, FORMAT, COMPU, COMPI, and
COMP3 unless IUNSIGNED is specified.

A-tO

Summary of SORT/MERGE Functions and Switches

/SIXBIT
The recording mode of the data is SIXBIT.
Function: Recording-mode switch
Type:
Global
Default: See the explanation of this switch in Chapter 3 for a description
of the defaults.
/STANDARD
The data should be read frqm or written to magnetic tape in
Standard-ASCII format. This switch is ignored if any other device is used.
Function: Tape switch
Type:
Modified position dependent
Default: Core-dump format
/SUPPRESS:n
Specifies which types of messages to suppress. The possible arguments
(n) are:
NONE

Suppress no messages.

INFORMATION

Suppress all information messages (those beginning
with TL

WARNING

Suppress all information messages, and all warning
messages (those beginning with ((If').

FAT AL

Suppress all information messages, all warning messages, and all fatal error messages (those beginning
with (?').

ALL

Same as FATAL.

NOTE
It is not possible to suppress messages beginning with ($'.

Function: Control switch
Type:
Global
Default: /SUPPRESS:NONE
/TEMP
Allows you to specify the device names for temporary file storage. For
stand-alone sorts, the number of device names is fixed at 15.
Function: Control switch
Type:
Local
Default: DSK:

Summary of SORT/MERGE Functions and Switches

A-ll

/UNLOAD
Rewind and unload the tape after the file is read or written. For multi-reel
files, this switch affects only the disposition of the last tape. All intermediate tapes are UNLOADed regardless of the setting of this switch. This
switch is ignored for any device other than a tape drive.
Function: Tape switch
Type:
Local
Default: Do not unload the last reel of a magtape file.
/UNSIGNED
The key is unsigned.
Function: Key data type switch
Type:
Local
Default: No default characteristics
/VARIABLE
The file contains variable-length records.
Function: File switch
Type:
Global
Default: Depends on the recording-mode switch you specify. See the description of the appropriate recording-mode switch in Chapter 3
for more details.

A-12

Summary of SORT/MERGE Functions and Switches

A.1 Switches (by Function)
Required Switches

File Switches

IRECORD:
IKEY:

IAFTER
IALIGN
IBEFORE
IBLOCKED:
IFIXED
IFORTRAN
INOCRLF
IRANDOM
ISEQUENTIAL
IVARIABLE

Key Data Type Switches
IALPHA
ICOMPU
ICOMPl
ICOMP3
IFORMAT:
INUMERIC
IPACKED
ISIGNED
IUNSIGNED

Control Switches

Recording Mode Switches
IASCII
IBINARY
IEBCDIC
ISIXBIT

Tape Switches
IDENSITY:
IINDUSTRY
ILABEL:
IPARITY:
IPOSITION:
IREWIND
ISTANDARD
IUNLOAD

ICHECK
ICOLLATE:
ICORE:
IERROR:
IEXIT
IFATAL:
IHELP
ILEAVES:
IMAXTEMP:
IMERGE
IMESSAGE:
IOPTION:
IPHYSICAL
IPRIORITY:
IRUN:
ITEMP

Summary of SORT/MERGE Functions and Switches

A-13

A.2 Switches (by Range)
Global Switches

Local Switches

IALIGN
IASCII
IBINARY
ICHECK
ICOLLATE:
ICORE:
IEBCDIC
IERROR:
IEXIT
IFATAL:
IFIXED
IFORTRAN
IHELP
IKEY:
ILEAVES:
IMAXTEMP
IMERGE
IMESSAGE:
IOPTION:
IRANDOM
IRECORD:
IRUN:
ISEQUENTIAL
ISIXBIT
ISUPPRESS:
ITEMP
IVARIABLE

IAFTER
IALPHA
IBEFORE
ICOMPU
ICOMPl
ICOMP3
IFORMAT:
INOCRLF
INUMERIC
IPACKED
IPOSITION:
IREWIND
ISIGNED
IUNLOAD
IUNSIGNED

Modified Position
Dependent Switches
IBLOCKED
IINDUSTRY
ILABEL:
ISTANDARD

Position Dependent Switches
IDENSITY:
IPARITY:
IPHYSICAL
IPRIORITY:

A-14

Summary of SORT/MERGE Functions and Switches

Appendix B
Collating Sequences and Conversion Tables

Table B-1 shows the ASCII and SIXBIT collating sequence and the conversions from ASCII to EBCDIC, SIXBIT to ASCII, and SIXBIT to EBCDIC. If
the ASCII character does not convert to the same character in EBCDIC, the
EBCDIC character is shown in parentheses next to the EBCDIC code. Note
that the first and last 32 characters do not exist in SIXBIT. Also, the characters in the first column (NULL, SOH, STX, etc.) are control characters,
which are nonprinting.

NOTE
All codes are specified in octal.

B-1

Table B-1:

ASCII and SIXBIT Collating Sequence and Conversion to EBCDIC

ASCII
Character 7-bit

NUL
SOH
STX
ETX
EOT
ENQ
ACK
BEL
BS
HT
LF
VT
FF
CR
SO
SI
DLE
DC1
DC2
DC3
DC4
NAK
SYN
ETB
CAN
EM
SUB
ESC
FS
GS
RS
US

B-2

000
001
002
003
004
005
006
007
010
011
012
013
014
015
016
017
020
021
022
023
024
025
026
027
030
031
032
033
034
035
036
037

EBCDIC
9-bit

000
001*
002*
003*
067
055*
056*
057*
026
005
045
013*
014*
025*(NL)
006*(LC)
066*(UC)
044*(BYP)
024*(RES)
064*(PN)
965*(RS)
004*(PF)
075*
027*(lL)
046*(EOB)
052*(CM)
031*
032*(CC)
047*(PRE)
023*(TM)
041 *(SOS)
040*(DS)
042*(FS)

Character

SIXBIT

ASCII
7-bit

Space

00
01
02
03
04
05
06
07
10
11
12
13
14
15
16
17
20
21
22
23
24
25
26
27
30
31
32
33
34
35
36
37

040
041
042
043
044
045
046
047
050
051
052
053
054
055
056
057
060
061
062
063
064
065
066
067
070
071
072
073
074
075
076
077

#
$

,
(

)

,
-

0
1
2
3
4
5
6
7
8
9
,

<
=
>
?

Collating Sequences and Conversion Tables

EBCDIC
9-bit

100
132
177
173
133
154
120
175
115
135
134
116
153
140
113
141
360
361
362
363
364
365
366
367
370
371
172
136
114
176
156
157

Table B-1 (Cont.): ASCII and SIXBIT. Collating Sequence and Conversion
to EBCDIC

Character SIXBIT
@

A
B
C
D
E
F
G
H
I
J
K
L
M
N

0
P
Q
R
S
T
U
V
W
X
Y
Z
[
\
]
A

40
41
42
43
44
45
46
47
50
51
52
53
54
55
56
57
60
61
62
63
64
65
66
67
70
71
72
73
74
75
76
77

ASCII
7-bit

EBCDIC
9-bit .

100
101
102
103
104
105
106
107
110
111
112
113
114
115
116
117
120
121
122
123
124
125
126
127
130
131
132
133
134
135
136
137

174
301
302
303
304
305
306
307
310
311
321
322
323
324
325
326
327
330
331
342
343
344
345
346
347
350
351
255*
340
275
137
155

Character
a
b
c
d
e
f
g
h
1

j
k
I
m

n
0

q
r
s
t
u
v
w

x
y

z
{
I

}

Delete

ASCII
7-bit

EBCDIC
9-bit

140
141
142
143
144
145
146
147
150
151
152
153
154
155
156
157
160
161
162
163
164
165
166
167
170
171
172
173
174
175
176
177

171
201
202
203
204
205
206
207
210
211
221
222
223
224
225
226
227
230
231
242
243
244
245
246
247
250
251
300*
117
320
241
007

* These EBCDIC codes either have no equivalent in the ASCII or SIXBIT character sets, or
are referred to by difference names. They are converted to the indicated ASCII characters
to preserve their uniqueness if the ASCII character is converted back to EBCDIC.

Collating Sequences and Conversion Tables

B-3

Table B-2 shows the conversion of ASCII code to SIXBIT code. The table
does not show ASCII codes 000 through 037 because they all convert to
SIXBIT 74 (\), except 11 (TAB) which converts to SIXBIT 00 (space).
Table B-2:

ASCII to SIXBIT Conversion

Character

ASCII
7-bit

SIXBIT

040
041
042
043
044
045
046
047

00
01
02
03
04
05
06
07

050
051
052
053
054
055
056
057

10
11
12
13
14
15
16
17

0
1
2
3
4
5
6
7

060
061
062
063
064
065
066
067

20
21
22
23
24
25
26
27

070
071
072
073
074
075
076
077

30
31
32
33
34
35
36
37

Space
!
((

$
%

,
(
)

+
,

,
<
=

>
?

B-4

Character

Collating Sequences and Conversion Tables

(it'

A
B
C
D

E
F
G

I
J
K
L
M
N

0
P

Q
R

S
T
U
V
W

X
Y

Z
[

\
]
~

ASCII
7-bit

SIXBIT

100
101
102
103
104
105
106
107

40
41
42
43
44
45
46
47

110
111
112
113
114
115
116
117

50
51
52
53
54
55
56
57

120
121
122
123
124
125
126
127

60
61
62
63
64
65
66
67

130
131
132
133
134
135
136
137

70
71
72
73
74
75
76
77

Table B-2 (Cont.):

ASCII to SIXBIT Conversion

ASCII
code

ASCII
character

SIXBIT
code

SIXBIT
character

140
141
142
143
144
145
146
147

74
41
42
43
44
45
46
47

50
51
52
53
54
55
56
57

H
I
J

60
61
62
63
64
65
66
67

P
Q
R

70
71
72
73
74
75
74
74

a
b
c
d
e
f
g

150
151
152
153
154
155
156
157

h
i
.j

160
161
162
163
164
165
166
167

170
171
172
173
174
175
176
177

k
I
m
n
0

q
r
s
t
u
v
w

y
z
{
I

}
-

Delete

B
C
D
E
F
G

K
L

M
N
0

S
T
U
V
W

Y
Z
I
\

I
\
\

Collating Sequences and Conversion Tables

B-5

Table B-3 shows the EBCDIC collating sequence and the conversion from
EBCDIC to ASCII. When conversion is from EBCDIC to SIXBIT, it is as if
the code was converted to ASCII and then from ASCII to SIXBIT.

Table B-3:

EBCDIC Collating Sequence and Conversion to ASCII

EBCDIC
code

EBCDIC
character

ASCII
code

ASCII
character

EBCDIC
code

000
001
002
003
004
005
006
007

NUL
SOH
STX
ETX
PF
HT
LC
Delete

000
001
002
003
024
011
016
177

NUL
SOH
STX
ETX
DC4
HT
SO
Delete

050
051
052
053
054
055
056
057

010
011
012
013
014
015
016
017

SMM
VT
FF
CR
SO
SI

134
134
134
013
014
134
134
134

\
\
\

060
061
062
063
064
065
066
067

020
021
022
023
024
025
026
027

DLE
DC1
DC2
TM
RES
NL
BS
IL

134
134
134
034
021
015
010
026

\
\
\

030
031
032
033
034
035
036
037

CAN
EM
CC
CUI
IFS
IGS
IRS
IUS

134
031
032
134
134
134
134
134

040
041
042
043
044
045
046
047

DS
SOS
FS

036
035
037
134
020
012
027
033

RS
GS
US

B-6

BYP
LF
ETB
ESC

VT
FF
\
\
\

FS
DC1
CR
BS
SYN
EM
SUB
\
\
\
\
\

DLE
LF
ETB
ESC

070
071
072
073
074
075
076
077
100
101
102
103
104
105
106
107
110
111
112
113
114
115
116
117

Collating Sequences and Conversion Tables

EBCDIC
character

ASCII
code

ASCII
character
\
\

ENQ
ACK
BEL

134
134
030
134
134
005
006
007

\
\
\
\

PN
RS
UC
EOT

134
134
134
134
022
023
017
004

CU3
DC4
NAK
SUB

134
134
134
134
134
025
134
134

\
\
\
\
\

SM
CUZ

Space

Cent
<
(

040
134
134
134
134
134
134
134
134
134
134
056
074
050
053
174

CAN
\
\

ENQ
ACK
BEL

DC2
DC3
SI
EOT

NAK
\
\

Space
\
\
\
\
\
\
\

\
\
\

<
(

Table B-3 (Cont.):

EBCDIC Collating Sequence and Conversion to ASCII

EBCDIC
code

EBCDIC
character

ASCII
code

ASCII
character

EBCDIC
code

120
121
122
123
124
125
126
127

046
134
134
134
134
134
134
134

&
\
\
\
\
\
\
\

170
171
172
173
174
175
176
177

130
131
132
133
134
135
136
137
140
141
142
143
144
145
146
147
150
151
152
153
154
155
156
157
160
161
162
163
164
165
166
167

*
)
A

,
o/c

>
?

134
134
041
044
052
051
073
137
055
057
134
134
134
134
134
134
134
134
134
054
045
137
076
077
134
134
134
134
134
134
134
134

\
\

!
$

*
)
A

\
-

I
\
\

\
\

\
\
\
\
\

,
7r

>
?
\
\
\
\
\
\

\
\

200
201.
202
203
204
205
206
207
210
211
212
213
214
215
216
217
220
221
222
223
224
225
226
227
230
231
232
233
234
235
236
237

EBCDIC
character

#
(u

=
"

a
b
c

d
e
f
g
h
i

k
1
m

n
0

p
q
r

ASCII
code

ASCII
character

134
140
072
043
100
47
075
042

134
141
142
143
144
145
146
147

150
151
134
134
134
134
134
134

h
i

134
152
153
154
155
156
157
160
161
162
134
134
134
134
134
134

Collating Sequences and Conversion Tables

#
(u

"
a
b
c

d
e
f
g

\
\

\
\
\
\
\

k
1
m

n
0

p
q
r
\

\
\
\
\
\

B-7

Table B-3 (Cont.):
EBCDIC
code

240
241
242
243
244
245
246
247
250
251
252
253
254
255
256
257

EBCDIC
character

t
u
v
w
x
y
z

EBCDIC Collating Sequence and Conversion to ASCII
ASCII
code

ASCII
character

EBCDIC
code

EBCDIC
character

ASCII
code

ASCII
character

134
176
163
164
165
166
167
170

310
311
312
313
314
315
316
317

H
I

110
110
134
134
134
134
134
134

H
I

171
172
134
134
134
133
134
134

y
z

260
261
262
263
264
265
266
267

175
134
134
134
134
134
134
134

270
271
272
273
274
275
276
277

134
134
134
134
134
135
134
134

300
301
302
303
304
305
306
307

B-8

A
B
C
D
E
F
G

173
101
102
103
104
105
106
107

t
u

v
w
x

\
\
\

I
\
\

\
\
\

\
\

\
\
\
\

1
\
\

A
B
C
D
E

F
G

320
321
322
323
324
325
326
327
330
331
332
333
334
335
336
337
340
341
342
343
344
345
346
347
350
351
352
353
354
355
356
357

Collating Sequences and Conversion Tables

J
K
L
M
N

0
P
Q
R

S
T
U

V
W
X
Y

175
112
113
114
115
116
117
120
121
122
134
134
134
134
134
134
134
134
123
124
125
126
127
130
131
132
134
134
134
134
134
134

\
\
\
\
\
\

J
K
L
M
N

0
P

Q
R
\

\
\
\
\

\
\

S
T
U

V
W
X

Y
Z
\
\
\
\
\
\

Table B-3 (Cont.):
EBCDIC
code

360
361
362
363
364
365
366
367

EBCDIC Collating Sequence and Conversion to ASCII

EBCDIC
character

ASCII
code

ASCII
EBCDIC
character
code

0
1
2
3
4
5
6
7

060
061
062
063
064
065
066
067

1
1
2
3
4
5
6
7

370
371
372
373
374
375
376
377

EBCDIC
character

ASCII
code

ASCII
character

070
071
134
134
134
134
134
134

9
\
\

\
\
\

Collating Sequences and Conversion Tables

B-9

Glossary

Term

Definition

ASCII

American Standard Code for Information Interchange. A 7-bit code
in which textual information is recorded. The ASCII code can represent 128 distinct characters. These characters are uppercase and
lowercase letters, numbers, common punctuation marks, and special
control characters.

Bias

The ratio of a run to the tree size. The bias should be approximately
2.0 if the file being sorted is truly random. See Run.

Buffer

A device or area, such as external storage devices, used to temporarily hold information being transmitted between two processes.

Data

A general term used to denote any or all information (facts, numbers, letters, and symbols) that refers to or describes an object, idea,
condition, or situation. It represents basic elements of information
that can be processed by a computer.

DSK

The generic device name for disk devices. It is translated into the
user's default path if a directory is not supplied, or into the structure
on which the user's default path resides if a directory is supplied.

Glossary-1

EBCDIC

Extended Binary Coded Decimal Interchange Code. An 8-bit translation code that can· represent 256 distinct characters. Because
TOPS-20 is a 36-bit system, EBCDIC disk files contain 9-bit bytes
with the 8-bits of data right-justified. Four 9-bit bytes make a
word. Standard 8-bit EBCDIC bytes are used when recording
industry-compatible tapes. All references to EBCDIC disk files in
this manual refer to the TOPS-20 implementation of EBCDIC.

File Format

The header words, control words, and control characters that are
used to format the data into separate records.

File Specification

A list of identifiers that uniquely speci(y a particular file. For
example,
dev: <directory >filnam.ext
The 'dev:' specifies the device where the file is stored, the
'<directory>' specifies the name of the directory where the file is
stored, the 'filnam' is the name of the file, and '.ext' is the file
extension.

Key

The part of the record being conlpared in order to sort the file into
the correct sequence.

Label

On magnetic tape, the leader or trailer record used to identify the
reel. On random-access devices, the directory block used by the
system.

Logical Record

The record that is created by some process, such as a program, but
has not yet been written to a storage device, such as disk or magnetic tape. See Physical Record.

Mixed-Mode
Binary

A binary file that is interpreted as containing any combination of
numeric, alphanumeric, fixed-point binary and floating-point
binary data. EBCDIC mixed-mode binary can include packeddecimal values; FORTRAN mixed-mode binary can include
FORMAT data.

Physical Record

The record that is written to a storage device such as disk or magnetic tape. Factors such as blocking and default physical record size
can generate a physical record consisting of many logical records.

Record

Glossary-2

A collection of adjacent data items treated as a unit.

Run

Scratch Device
Segment

The largest set of records ordered by key value that SORT/MERGE
can generate within its memory limits. (As memory increases,
fewer, but larger, runs occur.) For ascending (or descending) keys, a
new run is created if SORT/MERGE encounters a record whose key
is less (or greater) than the key value of the last record written to
the temporary file. Runs are always in monotonically increasing (or
decreasing) order.
A device used to store intermediate data during a sort.
A division or section of the user's virtual address space. There are
two types of segments:
Low Segment - starts at address zero and stops before address
400000. Program code stored in the low segment is not sharable.
High Segment - starts at address 400000 and uses as much of the
remaining address space required to hold the program code. This
code is sharable.
There are two advantages to segmenting the user's virtual address
space:

SIXBIT

Sort
Stable Sort
Standard Binary

Allows the program code to be shared.

Allows programs too large for the user's address space to be
divided into multiple high segments. When a particular portion
of code is required, it can be overlaid on (written over) the code
currently stored in the high segment of the user's address space.

A 6-bit code in which textual information is recorded. It is a compressed form of the ASCII character set.
The results of comparisons between groups of records.
A sort which leaves the relative order of equal keys unchanged.
A binary file that is interpreted as containing only fixed-point and
floating-point binary values.

Glossary-3

Index
36-bit, 1-13
Abbreviation,
collate, 3-17
IAFTER switch, 3-11, A-2
IALIGN switch, 3-11, A-2
IALPHA switch, A-2
ALPHANUMERIC data type, 3-5
IALPHANUMERIC switch, 1-12,3-5
Arguments,
ICOLLATE switch, 3-16
ILABEL switch, 3-24
ASCENDING order, 1-4
ASCII, Glossary-l
COBOL fixed-length, 4-5
COBOL variable-length, 4-9
fixed-length, 4-5
FORTRAN fixed-length, 4-6
FORTRAN variable-length, 4-10
variable-length, 4-8
ASCII file,
creating, 1-1
ASCII recording mode, 4-2
I ASCII switch, 1-12, 3-9, A-2
IBEFORE switch, 3-11, A-2
Bias, Glossary-l
Binary,
COBOL ASCII mixed-mode, 4-25
COBOL EBCDIC mixed-mode, 4-28
COBOL SIXBIT mixed-mode, 4-27
Binary file formats, 4-25
BINARY recording mode, 4-3
IBINARY switch, 3-10, A-3
Binary tree,
operation of, 6-4
Binary with LSCW,
FORTRAN random, 4-31
FORTRAN sequential, 4-35

Binary without LSCW,
FORTRAN random, 4-33
FORTRAN sequential, 4-37
Blocked fixed-length EBCDIC, 4-20
IBLOCKED switch, 3-12, A-3
Blocked variable-length EBCDIC, 4-22
Blocking files, 3-12
Buffer, Glossary-l
Calculating memory size, 6-11
Categories,
switch, 3-1
Character data,
sorting, 2-4
Characters in file,
tab, 2-3
ICHECK switch, 3-15, A-3
COBOL,
field descriptors for, 3-7
using SORT/MERGE from, 2-6
COBOL ASCII mixed-mode binary, 4-25
COBOL binary file formats, 4-25
COBOL EBCDIC mixed-mode binary, 4-28
COBOL fixed-length ASCII, 4-5
COBOL fixed-length EBCDIC, 4-17
COBOL fixed-length SIXBIT, 4-13
COBOL SIXBIT mixed-mode binary, 4-27
COBOL variable-length ASCII, 4-9
COBOL variable-length EBCDIC, 4-18
COBOL variable-length SIXBIT, 4-15
Codes,
error, 5-12
Collate abbreviation, 3-17
Collate completion, 3-17
Collate equivalence, 3-16
Collate functions, 3-16
ICOLLATE switch, 3-16, A-3
ICOLLATE switch arguments, 3--16

Index-l

Collating order,
key, 3-5
Collating sequences, B-1
Command files,
using, 2-5
Command formats, 2-4
COMP1 data type, 3-6, 3-8
ICOMP1 switch, 3-6, A-4
COMP3 data type, 3-6, 3-9
ICOMP3 switch, 3-6, A-4
Completion,
collate, 3-17
ICOMPU switch, A-4
COMPUTATIONAL data type, 3-6, 3-8
ICOMPUTATIONAL switch, 3-6
Considerations,
performance, 6-9
SORT/MERGE performance, 6-1
Control switches, 3-15
Control words in file, 2-3
Conversion tables, B-1
ICORE switch, 3-18, A-4
Creating ASCII file, 1-1
Creating multifield file, 1-1
Data, Glossary-1
sorting character, 2-4
sorting nonalphanumeric, 2-4
sorting nonnumeric, 2-4
Data type, 1-13
ALPHANUMERIC, 3-5
COMP1, 3-6, 3-8
COMP3, 3-6, 3-9
COMPUTATIONAL, 3-6, 3-8
FORMAT, 3-6, 3-8
NUMERIC, 3-5, 3-8
PACKED, 3-6, 3-9
Data types,
key, 3-5
IDENSITY switch, 3-23, A-4
DESCENDING order, 1-4
Descriptors for COBOL,
field, 3-7
Descriptors for FORTRAN,
field, 3-7
Determining key length, 1-2
Determining key position, 1-2
Determining record length, 1-2
DSK:, Glossary-1
EBCDIC, Glossary-2
blocked fixed-length, 4-20
blocked variable-length, 4-22
COBOL fixed-length, 4-17
COBOL variable-length, 4-18

Index-2

EBCDIC file formats, 4-17
EBCDIC recording mode, 4-2
IEBCDIC switch, 3-10, A-5
Equivalence,
collate, 3-16
Error codes, 5-12
Error messages, 5-2
SORT/MERGE, 5-1
IERROR switch, 2-8, 3-19, A-5
EXIT function, 2-4
IEXIT switch, 2-4, 3-19, A-5
IFATAL switch, 2-8, 3-19, A-5
Field descriptors for COBOL, 3-7
Field descriptors for FORTRAN, 3-7
File,
control words in, 2-3
creating ASCII, 1-1
creating multifield, 1-1
line-sequence numbers in, 2-3
sorting line-sequenced, 1-6
sorting multifield, 1-3
sorting multiline, 1-11
tab characters in, 2-3
File format, Glossary-2
File formats, 4-1, 4-3
File formats,
binary, 4-25
COBOL binary, 4-25
EBCDIC, 4-17
FORTRAN binary, 4-30
File specification, Glossary-2
File specifications, 2-5
File switches, 3-11
Files,
blocking, 3-12
merging, 2-5
sorting nontext, 1-12
using command, 2-5
Files containing tabs,
sorting, 1-8
IFIXED switch, 3-13, A-5
Fixed-length ASCII, 4-5
COBOL, 4-5
FORTRAN, 4-6
Fixed-length EBCDIC,
blocked, 4-20
COBOL, 4-17
Fixed-length SIXBIT, 4-12
COBOL, 4-13
Format,
IKEY switch, 3-4
message, 5-1
FORMAT data type, 3-6, 3-8
IFORMAT switch, 3-6, A-6

Formats,
binary file, 4-25
COBOL binary file, 4-25
command, 2-4
EBCDIC file, 4-17
file, 4-1, 4-3
FORTRAN binary file, 4-30
FORTRAN,
field descriptors for, 3-7
using SORT/MERGE from, 2-6
FORTRAN binary file formats, 4-30
FORTRAN fixed-length ASCII, 4-6
FORTRAN random binary with LSCW, 4-31
FORTRAN random binary without LSCW,
4-33
FORTRAN sequential binary with LSCW,
4-35
FORTRAN sequential binary without LSCW,
4-37
/FORTRAN switch, 3-14, A-6
FORTRAN variable-length ASCII, 4-10
FORTRAN-called sort restrictions, 2-7
FSORT program, 2-7
Function,
EXIT, 2-4
HELP, 2-4
MERGE, 2-4
RUN, 2-4
SORT, 2-4
switches by, A-13
Function summary,
SORT/MERGE, A-I
Functions, A-I
collate, 3-16
Getting started with SORT/MERGE, 1-1
Global switches, 3-1
HELP function, 2-4
/HELP switch, 2-4, A-6
How to use SORT/MERGE, 2-1
/INDUSTRY switch, 3-24, A-6
Information needed to sort files, 2-2
Introduction, 2-1
Key, Glossary-2
Key collating order, 3-5
Key data types, 3-5
Key length, 3-5
determining, 1-2
Key position,
determining, 1-2
Key sign status, 3-8

/KEY starting position, 3-4
/KEY switch, 3-3, A-6
/KEY switch format, 3-4
Keys,
sorting on two, 1-6
Label, Glossary-2
/LABEL switch, 3-24, A-7
/LABEL switch arguments, 3-24
/LEAVES switch, 3-20, A-7
Length,
determining key, 1-2
determining record, 1-2
key, 3-5
Line-sequence numbers in file, 2-3
Line-sequenced file,
sorting, 1-6
Local switches, 3-1
Logical record, Glossary-2
Lowercase text,
sorting, 1-11
LSCW,
FORTRAN random binary with, 4-31
FORTRAN random binary without, 4-33
FORTRAN sequential binary with, 4-35
FORTRAN sequential binary without,
4-37
/MAXTEMP switch, 3-20, A-7
Memory required, 2-2
Memory size, 6-10
calculating, 6-11
MERGE function, 2-4
Merge passes,
number of, 6-14
Merge phase, 2-1, 6-3
/MERGE switch, 2-4, A-7
Merging files, 2-5
Message format, 5-1
/MESSAGE switch, 3-21, A-8
Messages,
error, 5-2
SORT/MERGE error, 5-1
Mixed-mode binary, Glossary-2
COBOL ASCII, 4-25
COBOL EBCDIC, 4-28
COBOL SIXBIT, 4-27
Mode,
ASCII recording, 4-2
BINARY recording, 4-3
EBCDIC recording, 4-2
SIXBIT recording, 4-2
Mode switches,
recording, 3-9

Index-3

Modes,
recording, 1--14, 4-1
Modified position dependent switches, 3-2
Multifield file,
creating, 1-1
sorting, 1-3
Multiline file,
sorting, 1-11
INOCRLF switch, 3-14, A-8
Nonalphanumeric data,
sorting, 2-4
Nonnumeric data,
sorting, 2-4
Nontext files,
sorting, 1-12
Number of merge passes, 6-14
N umbers in file,
line-sequence, 2-3
NUMERIC data type, 3-5, 3-8
INUMERIC switch, 3-5, A-8

Operation of binary tree, 6-4
IOPTION switch, 3-21, A-8
Order,
ASCENDING, 1-4
DESCENDING, 1-4
key collating, 3-5
Overview,
performance, 6-1
PACKED data type, 3-6, 3-9
IPACKED switch, 3-6, A-8
IPARITY switch, 3-25, A-9
Passes,
number of merge, 6-14
Performance considerations, 6-9
SORT/MERGE, 6-1
Performance overview, 6-1
Phase,
merge, 2-1, 6-3
sort, 2-1, 6-2
Physical record, Glossary-2
IPHYSICAL switch, 3-21, A-9
Position,
determining key, 1-2
IKEY starting, 3-4
Position dependent switches, 3-2
modified, 3-2
IPOSITION switch, 3-25, A-9
IPRIORITY switch, 3-21, A-9
Program,
FSORT, 2-7

Index-4

Random binary with LSCW,
FORTRAN, 4-31
Random binary without LSCW,
FORTRAN, 4-33
IRANDOM switch, 3-14, A-9
Range,
switches by, A-14
Record, Glossary-2
Record length,
determining, 1-2
IRECORD switch, 3-3, A-10
Recording mode switches, 3-9
Recording modes, 1-14, 4-1
Records,
sorting variable-length, 1-10
Required,
memory, 2-2
Required switches, 3-3
Restrictions,
FORTRAN-called sort, 2-7
IREWIND switch, 3-26, A-10
Run, Glossary-3
RUN function, 2-4
IRUN switch, 2-4, 3-22, A-10
Running SORT/MERGE, 2-2
SCAN, 2-4
Scratch device, Glossary-3
Segment, Glossary-3
Sequences,
collating, B-1
Sequential binary with LSCW,
FORTRAN, 4-35
Sequential binary without LSCW,
FORTRAN, 4-37
ISEQUENTIAL switch, 3-15, A-10
Sign status,
key, 3-8
ISIGNED switch, A-10
SIXBIT, Glossary-3
COBOL fixed-length, 4-13
COBOL variable-length, 4-15
fixed-length, 4-12
variable-length, 4-14
SIXBIT recording mode, 4-2
ISIXBIT switch, 3-10, A-II
Size,
calculating memory, 6-11
memory, 6-10
tree, 6-9
Sort, Glossary-3
Sort files,
information needed to, 2-2

SORT function, 2-4
Sort phase, 2-1, 6-2
Sort restrictions,
FORTRAN-called, 2-7
SORT/MERGE,
getting started with, 1-1
how to use, 2-1
running, 2-2
starting, 1-3
SORT/MERGE error messages, 5-1
SORT/MERGE from COBOL,
using, 2-6
SORT/MERGE from FORTRAN,
using, 2-6
SORT/MERGE function summary, A-I
SORT/MERGE performance considerations,
6-1
SORT/MERGE switch summary, A-I
SORT/MERGE switches, 3-1
Sorting character data, 2-4
Sorting files containing tabs, 1-8
Sorting line-sequenced file, 1-6
Sorting lowercase text, 1-11
Sorting multifield file, 1-3
Sorting multiline file, 1-11
Sorting nonalphanumeric data, 2-4
Sorting nonnumeric data, 2-4
Sorting nontext files, 1-12
Sorting on two keys, 1-6
Sorting uppercase text, 1-11
Sorting variable-length records, 1-10
Specifications,
file, 2-5
Stable sort, Glossary-3
Standard binary, Glossary-3
ISTANDARD switch, 3-26, A-II
Started with SORT/MERGE,
getting, 1-1
Starting position,
IKEY, 3-4
Starting SORT/MERGE, 1-3
Status,
key sign, 3-8
Summary,
SORT/MERGE function, A-I
SORT/MERGE switch, A-I
ISUPPRESS switch, 3-22, A-II
Switch,
IAFTER, 3-11, A-2
IALIGN, 3-11, A-2
IALPHA, A-2
IALPHANUMERIC, 1-12, 3-5
IASClI, 1-12,3-9, A-2
IBEFORE, 3-11, A-2

Switch (Cont.),
IBINARY, 3-10, A-3
IBLOCKED, 3-12, A-3
ICHECK, 3-15, A-3
ICOLLATE, 3-16, A-3
ICOMPl, 3-6, A-4
ICOMP3, 3-6, A-4
ICOMPU, A-4
ICOMPUTATIONAL, 3-6
ICORE, 3-18, A-4
IDENSITY, 3-23, A-4
IEBCDIC, 3-10, A-5
IERROR, 2-8, 3-19, A-5
IEXIT, 2-4, 3-19, A-5
IFATAL, 2-8, 3-19, A-5
IFIXED, 3-13, A-5
IFORMAT, 3-6, A-6
IFORTRAN, 3-14, A-6
IHELP, 2-4, A-6
IINDUSTRY, 3-24, A-6
IKEY, 3-3, A-6
ILABEL, 3-24, A-7
ILEA VES, 3-20, A-7
IMAXTEMP, 3-20, A-7
IMERGE, 2-4, A-7
IMESSAGE, 3-21, A-8
INOCRLF, 3-14, A-8
INUMERIC, 3-5, A-8
IOPTION, 3-21, A-8
IPACKED, 3-6, A-8
IP ARITY, 3-25, A-9
IPHYSICAL, 3-21, A-9
IPOSITION, 3-25, A-9
IPRIOR.LTY, 3-21, A-9
IRANDOM, 3-14, A-9
IRECORD, 3-3, A-10
IREWIND, 3-26, A-10
IRUN, 2-4, 3-22, A-10
ISEQUENTIAL, 3-15, A-10
ISIGNED, A-10
ISIXBIT, 3-10, A-II
ISTANDARD, 3-26, A-II
ISUPPRESS, 3-22, A-II
ITEMP, 3-23, A-II
IUNLOAD, 3-26, A-12
IUNSIGNED, A-12
IVARIABLE, 3-15, A-12
Switch arguments,
ICOLLATE, 3-16
ILABEL, 3-24
Switch categories, 3-1
Swi tch format,
IKEY, 3-4

Index-5

Switch summary,
SORT/MERGE, A-I
Switches, A-2
control, 3--15
file, 3-11
global, 3-1
local, 3-1
modified position dependent, 3-2
position dependent, 3-2
recording mode, 3-9
required, 3-3
SORT/MERGE, 3-1
tape, 3-23
Switches by function, A-13
Switches by range, A-14
Tab characters in file, 2-3
Tables,
conversion, B-1
Tabs,
sorting files containing, 1-8
Tape switches, 3-23
/TEMP switch, 3-23, A-II
Text,
sorting lowercase, 1-11
sorting uppercase, 1-11
Tree,
operation of binary, 6-4
Tree size, 6-9

Index-6

Two keys,
sorting on, 1-6
Type,
data, 1-13
/UNLOAD switch, 3-26, A-12
/UNSIGNED switch, A-12
Uppercase text,
sorting, 1-11
Use SORT/MERGE,
how to, 2-1
U sing command files, 2-5
Using SORT/MERGE from COBOL, 2-6
Using SORT/MERGE from FORTRAN, 2-6
IVARIABLE switch, 3-15, A-12
Variable-length ASCII, 4-8
COBOL, 4-9
FORTRAN, 4-10
Variable-length EBCDIC,
blocked, 4-22
COBOL, 4-18
Variable-length records,
sorting, 1-10
Variable-length SIXBIT, 4-14
COBOL, 4-15
Words in file,
control, 2-3

TOPS-10
SORT/MERGE
User's Guide
AA-M063A-TB

READER'S COMMENTS
NOTE:

This form is for document comments only. DIGITAL will use comments submitted on
this form at the company's discretion. If you require a written reply and are eligible to
receive one under Software Performance Report (SPR) service, submit your comments on an SPR form.

Did you find this manual understandable, usable, and well-organized? Please mae suggestions for improvement.

Did you find errors in this manual? If so, specify the error and the page number.

Please indicate the type of reader that you most nearly represent.

o Assembly language programmer
o Higher-level language programmer
o Occasional programmer (experienced)
o User with little programming experience
o Student programmer
o Other (please specify)~~~~~~~~~~~~~~~~~~~~~_
Name

Date - _ _ _ _ _ _ _ _ _ __

Organization _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Telephone ~_ _ _ _ _ _ __
Street ___________________________________________
City _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __

State ____ Zip Code _____
or Country

I
I
1

I
1

- -- - - --

-~-.- _gDOmNotgTear FOa fie,re and Tape --- - - - --- --- - - - -- -- - id

-rll-ll1-------- ~~~~;:;~~ -~--:

if Mailed in the
United States

BUSINESS REPLY MAIL
FIRST CLASS PERMIT NO. 33 MAYNARD MASS.
POSTAGE WILL BE PAID BY ADDRESSEE

SOFTWARE PUBLICATIONS

200 FOREST STREET

MR1-2/L12

MARLBOROUGH, MASSACHUSETTS

01752

.- - - - - - - - -

Do Not Tear - Fold Here and Tape

____________________________________________ 1

I
I
I
I
I
I
I
I

I
1
1

I
1

(

I.!
I-

II
1

Ie:t
I

1.£
I-.!
I ";:

Ie:
I
I
1