You define control fields to identify the information you want DFSORT to sort or merge. When thinking of the contents of your data sets, you probably think of names, dates, account numbers, or similar pieces of useful information. For example, when sorting your data sets, you might choose to arrange your records in alphabetical order, by family name. By using the byte position and length (in bytes) of the portion of each record containing a family name, you can define it as a control field to manipulate with DFSORT.

DFSORT uses the control fields you define as keys in processing. A key is a concept, such as family name, that you have in mind when you design a record processing strategy for a particular application. A control field, on the other hand, is a discrete portion of a record that contains the text or symbols corresponding to that information in a form that can be used by DFSORT to identify and sort, or merge the records. For all practical purposes, you can think of keys as equivalent to the control fields DFSORT uses in processing.

To arrange your records in a specific order, identify one or more control fields of your records to use as keys. The sequence in which you list the control fields becomes the order of priority DFSORT uses to arrange your records. The first control field you specify is called the major control field. Subsequent control fields are called minor control fields, as in first, second, third minor control fields, and so on.

If two or more records have identical values for the first control field, they are arranged according to the values in the second. Records with identical values for the first and second are arranged according to the third, and so on, until a difference is found or no more control fields are available.

Records with identical values for all the control fields specified retain their original input order or are arranged randomly, depending upon which of the two options, EQUALS or NOEQUALS, is in effect. You can direct DFSORT to retain the original input order for records with identical values for all control fields by specifying EQUALS.

Control fields may overlap, or be contained within other control fields (such as a three-digit area code, within a 10-digit telephone number). They do not need to be contiguous but must be located within the first 32752 bytes of the record (see Figure 1).

Figure 1. Control Fields. Control fields may overlap, or be contained within other control fields.

DFSORT offers several standard collating sequences. You can choose to arrange your records according to these standard collating sequences or according to a collating sequence defined in the active locale. Conceptually, a collating sequence is a specific arrangement of character priority used to determine which of two values in the same control field of two different records should come first. DFSORT uses EBCDIC, the standard IBM collating sequence, or the ASCII collating sequence when sorting or merging records. If locale processing is in effect, DFSORT will use the collating sequence defined in the active locale.

The collating sequence for character data and binary data is absolute; character and binary fields are not interpreted as having signs. For packed decimal, zoned decimal, fixed-point, normalized floating-point, and the signed numeric data formats, collating is algebraic; each quantity is interpreted as having an algebraic sign.

You can modify the standard EBCDIC sequence to collate differently if, for example, you want to allow alphabetic collating of national characters. An alternate collating sequence can be defined with the ALTSEQ installation option, or you can define it yourself at run-time with the ALTSEQ program control statement. You can also specify a modified collating sequence with an E61 user exit or with an EFS program.

You can specify the LOCALE installation or run-time option to use an active locale's collating rules.