Updated: 11 December 1998

OpenVMS User's Manual

This chapter describes how to use the OpenVMS Sort/Merge utility (SORT/MERGE). The Sort/Merge utility performs two operations:

• Sorts records from one or more input files according to the fields you select and generates one reordered output file
• Merges up to 10 (high-performance Sort/Merge utility supports up to 12) input files that have been sorted previously according to the same key fields and generates one output file.

On Alpha systems, you can also choose the high-performance Sort/Merge utility. This utility takes advantage of the Alpha architecture to provide better performance for most Sort and Merge operations. Refer to Section 11.1 for information.

This chapter describes:

• High-performance Sort/Merge
• Sorting files
• Specifying collating sequences
• Running Sort as a batch job
• Merging files
• Entering records from a terminal
• Using a Sort/Merge specification file
• Optimizing a Sort or Merge operation
• Summary of Sort/Merge qualifiers

For additional information, see the following:

• For information on commands used in this chapter, refer to the OpenVMS DCL Dictionary.
• For information on how a system manager can improve efficiency when using the Sort/Merge utility, see the OpenVMS System Manager's Manual.

On Alpha systems, you can also choose the high-performance Sort/Merge utility. This utility takes advantage of the Alpha architecture to provide better performance for most Sort and Merge operations.

The high-performance Sort/Merge utility uses the same command line interface as SORT/MERGE. Any differences between the high-performance Sort/Merge utility and SORT/MERGE are noted throughout this chapter.

Use the SORTSHR logical to select the high-performance Sort/Merge utility. Define SORTSHR to point to the high-performance sort executable in SYS$LIBRARY as follows:

$ define sortshr sys$library:hypersort.exe

To return to SORT/MERGE, deassign SORTSHR. The SORT/MERGE utility is the default if SORTSHR is not defined.

The behavior of the high-performance Sort/Merge utility is the same as SORT/MERGE, except as shown in Table 11-1.

If you attempt to use an unsupported qualifier or assign an unsupported value to a qualifier, the high-performance Sort/Merge utility generates an error.

Table 11-1 High-Performance Sort/Merge: Differences in Behavior

Feature	High-Performance Sort/Merge Behavior
Key data types	The H-FLOATING and ZONED decimal data types are not supported. The size of a BINARY data type key must be 1, 2, 4, or 8 bytes. A 16-byte binary key is not supported. (Implementation of this feature is deferred to a future OpenVMS Alpha release.)
Collating sequences	The National Character Set (NCS) collating sequences are not supported. (Implementation of this feature is deferred to a future OpenVMS Alpha release.) Do not specify the name of an NCS collating sequence for the /COLLATING_SEQUENCE qualifier. The ASCII, EBCDIC, and MULTINATIONAL collating sequences are supported. The default is ASCII. You cannot define or modify your own collating sequence through the use of a specification file. (Implementation of this feature is deferred to a future OpenVMS Alpha release.)
Specification files	Specification files are not supported. (Implementation of this feature is deferred to a future OpenVMS Alpha release.) Do not use the /SPECIFICATION qualifier.
Internal sorting process	Only the record sort process is supported. (Implementation of this feature is deferred to a future OpenVMS Alpha release.) You can specify /PROCESS=RECORD or omit the /PROCESS qualifier. The TAG, ADDRESS, and INDEX values for the /PROCESS qualifier are not supported.
Statistical summary information	The following statistics are currently supported: Records read Records sorted Records output Input record length The following statistics are unavailable: Internal length Output record length Sort tree size Number of initial runs Maximum merge order Number of merge passes Work file allocation Full implementation of this feature is deferred to a future OpenVMS Alpha release.

11.2 Sorting Files

To sort files, use the DCL command SORT. Specify the names of the files to be sorted, separated by commas, followed by the name of the ordered output file to be created.

Optionally, you can specify a key for each field on which you want to sort. Each key includes the following information:

• Starting position of the key field in a record (required)
• Size of the key (required)
• Data type of the key
• Order in which the records are sorted
• Priority of the key

If you do not specify any keys, Sort assumes there is only one key and that this key field:

• Begins in the first position of a record
• Includes the entire record
• Contains character data
• Is sorted in ascending order

The following two examples use the default key.

1. In this example, the file NAMES.LST is sorted in ascending order:

   $ SORT NAMES.LST BYNAME.LST

   
   This command creates the ordered output file BYNAME.LST, as shown in Figure 11-1.

   Figure 11-1 List Sorted in Ascending Order

   ---

   
2. In this example, the files NAMES.LST and NAMES2.LST are sorted into the ordered output file BYNAME.LST. Sort treats the files as if they were one large file:

   $ SORT NAMES.LST,NAMES2.LST BYNAME.LST

See Section 11.9 for a complete list of SORT qualifiers.

11.2.1 Defining a Key

Table 11-2 describes the five elements that comprise a key.

Table 11-2 /KEY Qualifier Values

Key Element	Value	Description
Key position	POSITION: n	The position of the first byte of the key field within the record. The first byte in a record is position 1. POSITION: n is required.
Key size	SIZE: n	The length of the key field. SIZE: n is required except for floating-point data. The data type you specify for the key determines what values are acceptable when specifying size. The following table lists the possible values for each type of data and the units used to specify the size of the key. Data Valid Range Units Character 1 through 32,767 Characters Binary 1, 2, 4, 8, or 16 (For the high-performance Sort/Merge utility, the size of a binary data type key must be 1, 2, 4, or 8 bytes. Support of a 16-byte binary key is deferred to a future OpenVMS Alpha release.) Bytes Decimal 1 through 31 Digits Floating-point No value is necessary. For decimal data, if the decimal sign is stored in a separate byte, that byte is not counted toward the size of the data. If you specify a key that extends beyond the end of a record, Sort treats the missing characters as null characters.
Data type	CHARACTER	Character data. CHARACTER is the default data type.
	BINARY	Binary data. SIGNED --- Signed binary or decimal data. SIGNED is the default for binary and decimal data. UNSIGNED --- Unsigned binary or decimal data.
	F_FLOATING	F_FLOATING format data.
	D_FLOATING	D_FLOATING format data.
	G_FLOATING	G_FLOATING format data.
	H_FLOATING	On VAX systems, H_FLOATING format data. (Not currently supported by the high-performance Sort/Merge utility.)
	S_FLOATING	On Alpha systems, IEEE S_FLOATING format data.
	T_FLOATING	On Alpha systems, IEEE T_FLOATING format data.
	DECIMAL	Decimal data. TRAILING_SIGN --- Trailing sign decimal data. TRAILING_SIGN is the default for decimal data. LEADING_SIGN --- Leading sign decimal data. The leading sign must be in the first position of the field and the field must be left zero padded. OVERPUNCHED_SIGN --- Overpunched decimal data. OVERPUNCHED_SIGN is the default for decimal data. SEPARATE_SIGN --- Separate sign decimal data.
	ZONED	Zoned decimal data. (Not currently supported by the high-performance Sort/Merge utility.)
	PACKED_DECIMAL	Packed decimal data.
Sort order	ASCENDING	Orders the sorting operation in ascending alphabetical or numerical order. ASCENDING is the default order.
	DESCENDING	Orders the sorting operation in descending alphabetical or numerical order.
Key priority	NUMBER: n	Specifies the order of priority of each key if you do not list multiple keys in the order of their priority. A value of 1 to 255 can be specified.

If the data in the key fields is not character data, you must specify the data type. The following data types are recognized by the Sort/Merge utility:

BINARY, [SIGNED]
BINARY, UNSIGNED
CHARACTER
DECIMAL, LEADING_SIGN, SEPARATE_SIGN [SIGNED]
DECIMAL, LEADING_SIGN, [OVERPUNCHED_SIGN, SIGNED]
DECIMAL [,SIGNED, TRAILING_SIGN, OVERPUNCHED_SIGN]
DECIMAL, [TRAILING SIGN], SEPARATE_SIGN, [SIGNED]
DECIMAL, UNSIGNED
D_FLOATING
F_FLOATING
G_FLOATING
H_FLOATING
S_FLOATING, IEEE (Alpha systems only)
T_FLOATING, IEEE (Alpha systems only)
PACKED_DECIMAL
ZONED

The items in brackets are defaults and need not be specified.

In Figure 11-2, each record in the file EMPLOYEE.LST consists of three fields: (1) a department name, (2) an account number, and (3) a customer name.

Figure 11-2 Record Fields in a List

The following examples illustrate how to sort the records in EMPLOYEE.LST both with, and without, a key field:

1. In this example, EMPLOYEE.LST is sorted by account number, using the /KEY qualifier to describe the account number field:

   $ SORT/KEY=(POSITION:5,SIZE:4,DECIMAL) EMPLOYEE.LST BILLING1.LST

   
   This command specifies that the key field (the account number) starts in position 5, is 4 characters long, contains decimal data, and should be sorted in ascending order (the default). Figure 11-3 shows the results of this Sort operation.

   Figure 11-3 Sorting by Key Field

   ---

   
2. This example shows how to sort the file EMPLOYEE.LST without specifying a key field:

   $ SORT EMPLOYEE.LST BYDEPT.LST

   
   Because no key is specified, Sort assumes the default characteristics. Figure 11-4 shows the result of this Sort operation.

   Figure 11-4 Sorting with Default Key Records

   ---

   
   Sort treats each record in EMPLOYEE.LST as one key of character data. In this example, each record includes a department name, an account number, and a customer name. If Sort finds a duplicate department name, it sorts the names by account number. If it then finds a duplicate account number, it sorts by customer name. Note that the account number is part of the record. Unless you specify otherwise, it is treated as character data.

You can sort with more than one key (up to a limit of 255 keys). You can specify multiple keys in order of their priority with the primary key first, the secondary key next, and so on. Alternately, you can specify a key's priority using NUMBER:n. Each key can be ascending or descending.

In the following example, the file EMPLOYEE.LST is sorted by the customer name key first and then (where there are identical names), by the account number:

$ SORT /KEY=(POSITION:10,SIZE:15,CHARACTER) - _$ /KEY=(POSITION:5,SIZE:4,DECIMAL) EMPLOYEE.LST BILLING2.LST

Figure 11-5 shows the results of this Sort operation.

Figure 11-5 Sorting with Multiple Key Fields

In the following example, records are sorted first by the department name in descending order, then by the customer name in ascending order:

$ SORT/KEY=(POSITION:1,SIZE:3,DESCENDING) - _$ /KEY=(POSITION:10,SIZE:15) - _$ EMPLOYEE.LST BILLING3.LST

Figure 11-6 shows the results of this Sort operation.

Figure 11-6 Sorting with Multiple Key Fields (Ascending and Descending Order)

By default, Sort/Merge keeps records with identical key fields but does not necessarily maintain the same order in which they appeared in the input file. To control the way in which records with identical keys are sorted, specify one of the following qualifiers:

• /STABLE
  Maintains the input order of records with identical keys. If you use this qualifier when sorting multiple input files, on output, records with equal keys in the first file precede those from the second file and so on.
• /NODUPLICATES
  Retains only one copy of records with identical keys. If you want to specify which duplicate record to keep, invoke Sort at the program level and specify an equal-key routine.

The /STABLE and /NODUPLICATES qualifiers are incompatible. You cannot specify both qualifiers on the same command line.

In the following example, records with duplicate account numbers are eliminated from the file EMPLOYEE.LST:

$ SORT /KEY=(POSITION:5,SIZE:4)/NODUPLICATES EMPLOYEE.LST BUDGET.LST

Figure 11-7 shows the results of this Sort operation.

Figure 11-7 Sorting with Identical Key Fields

If you sort records that contain items other than character data, specify the data type of each key. In addition, take care in calculating starting positions and sizes because the items being compared can occupy more than 1 byte.

If you are sorting a file that contains 20 characters followed by 3 floating-point numbers in F_floating format, the positions are as follows:

• The character data occupies positions 1 to 20 (20 characters).
• The first F_floating-point number occupies positions 21 to 24.
• The second F_floating-point number occupies positions 25 to 28.
• The third F_floating-point number occupies positions 29 to 32.

To sort the file by the third floating-point number, specify the key field as follows:

$ SORT/KEY=(POSITION:29,F_FLOATING) STATS.RAW STATS.SOR

You do not need to specify the size of the floating-point number because it is fixed at four bytes.

11.2.5 Output File Organization

By default, Sort produces an output file with the same file organization as that of the first input file. To specify a different output file organization, include one of the following qualifiers after the output file specification on the Sort command line:

• /FORMAT (record format)
  When you use this output qualifier, you can define the file record format, length, and block size.
• /INDEXED_SEQUENTIAL
  Using this qualifier, you can define the output to have indexed sequential file organization. If you specify indexed sequential as the output file organization, you must also do the following:
  • Before you perform the Sort operation, create an empty file to be used as the output file. Sort requires an output file that already exists and is empty.
  • Include the /OVERLAY qualifier after the name of the output file on the SORT command line. The /OVERLAY qualifier indicates the existing file is to be overlaid with the sorted records of the input file.
• /RELATIVE
  Using this qualifier, you can define the output to have relative file organization.
• /SEQUENTIAL
  Using this qualifier, you can define the output to have sequential file organization.

In the following example, a sequential file is produced after the indexed sequential file EMPLOYEE.LST is sorted:

$ SORT/KEY=(POSITION:10,SIZE:15) - _$ EMPLOYEE.LST BYNAME.LST/SEQUENTIAL

11.2.6 Sorting Process

Sort arranges files using one of the internal processes: record, tag, address, or indexed. (The high-performance Sort/Merge utility supports only the record process. Implementation of tag, address, and index processes is deferred to a future OpenVMS Alpha release.) The process you specify can affect the efficiency of the Sort operation. Refer to Section 11.8 for information about optimizing a Sort or Merge operation.

The following table describes the four types of process. Use the /PROCESS=type qualifier to specify the sort process.

Sort Process	type	Description
Record	RECORD	Keeps records intact while sorting and produces an output file consisting of complete records. Record is the default sorting process.
Tag	TAG	Sorts the key fields only and then rereads the input file to produce an output file of complete records. The net result is the same as for a complete record sort. A tag sort is useful if disk space is low because it typically uses less work file space during the sorting. In most cases, a tag sort is slower than a record sort because it requires extra time to reread the input file.
Address	ADDRESS	Sorts the key fields only and produces an output file that is an index of record file addresses (RFAs) in binary format. An address sort is faster than a record sort but you must write a program to associate the record addresses with the records of the input file.
Indexed	INDEX	Sorts the key fields only and produces an output file of keys and RFAs (in binary format). As with an address sort, an index sort is faster than a record sort, but you must write a program to associate the record addresses with the records of the input file.

Copyright © Compaq Computer Corporation 1998. All rights reserved. Legal

6489PRO_024.HTML