Easytrieve SORT Statement

Full Statement Shape

1
2
3
4
5
6
7
SORT input-file-name TO sorted-file-name +
  USING (sort-key-field-name [D] ...) +
  [COMMIT ([ACTIVITY|NOACTIVITY] [TERMINAL|NOTERMINAL])] +
  [SIZE record-count] +
  [WORK number-of-work-data-sets] +
  [BEFORE proc-name] +
  [NAME sort-name]

Each piece has a job. input-file-name is the FILE you read—must support sequential processing. sorted-file-name receives output in the same record layout unless BEFORE changes lengths. USING parenthesizes keys major-to-minor. D after one key requests descending for that key only. COMMIT mirrors JOB commit semantics for recoverable online or SQL environments. SIZE tells the sort how many records to plan for when volume is not obvious. WORK controls sort work data set count on z/OS. BEFORE names a prescreen procedure. NAME documents the activity for EXECUTE and program maps.

SORT as an Activity Block

Implied PROGRAM runs SORT and JOB in source order for batch programs without an explicit PROGRAM block. You may EXECUTE a named SORT only when a flag is set. SORT never replaces JOB—it prepares input JOB will read.

Complete Sort-Then-Report Example

1
2
3
4
5
6
7
8
9
10
11
12
13
14
FILE PERSNL FB(150 1800)
%PERSNL
FILE SORTWRK FB(150 1800) VIRTUAL
COPY PERSNL
 
SORT PERSNL TO SORTWRK USING (REGION, BRANCH, DEPT) NAME PAY-SORT
 
JOB INPUT SORTWRK NAME PAY-JOB
  PRINT PAY-RPT
 
REPORT PAY-RPT LINESIZE 132
  CONTROL REGION BRANCH DEPT
  TITLE 01 'PERSONNEL BY REGION BRANCH DEPT'
  LINE 01 REGION BRANCH DEPT EMP# NAME GROSS

PERSNL is the nightly extract. SORTWRK is VIRTUAL with COPY PERSNL so field positions match without maintaining two layouts. SORT orders REGION major, BRANCH minor, DEPT more minor—all ascending here. JOB reads SORTWRK explicitly; naming INPUT avoids ambiguity when Library declares other files. REPORT CONTROL uses the same key order as USING so breaks fire when values change. Mismatch between USING and CONTROL is the top reason sorted jobs still look wrong on spool.

INPUT and TO File Rules

INPUT must be defined on FILE with sequential access—QSAM, VSAM ESDS-style sequential, VIRTUAL, and other organizations your release documents. Keys must be fixed fields from DEFINE on that file. TO is usually a separate FILE; classic guidance discourages TO equaling INPUT for VSAM or ISAM rewrite in place. VIRTUAL TO with COPY of INPUT is the beginner-safe pattern: same picture, new order, no catalog clutter. Permanent TO files need JCL DD with SPACE and DISP appropriate for retention.

USING — Quick Key Rules

• Keys must be defined in Library on the input file layout.
• Major-to-minor: first key dominates sort groups.
• Maximum key length under 256 bytes; no varying-length keys.
• Invalid key types include V varying, K, and M per Broadcom.
• D suffix applies only to the key immediately before it.
• Installation caps how many keys one SORT may list.

Deep treatment of key choice lives on the Sort keys and Multi-key sorting pages in this chapter.

BEFORE Prescreen and SELECT

1
2
3
4
5
6
7
SORT PERSNL TO SORTWRK USING (DEPT, EMP#) BEFORE DEPT20
 
DEPT20. PROC
  IF DEPT EQ 20
     SELECT
  END-IF
END-PROC

Every input record enters DEPT20 before the sort utility sees it. SELECT keeps department 20 rows; others drop. Invalid in sort procedures include GET, PUT, READ, and PRINT—I/O belongs elsewhere. DISPLAY to SYSPRINT is allowed for debugging counts. STOP ends prescreen early; use carefully in production. Broadcom documents no AFTER on SORT—post-sort work belongs in JOB.

SIZE, WORK, and COMMIT

SIZE record-count helps the sort allocate workspace when Easytrieve cannot infer volume from a prior activity. WORK 0 means your JCL supplies sort work DDs; WORK 1–31 requests dynamic work data sets. Values are site-specific—coordinate with operations. COMMIT ACTIVITY or NOACTIVITY pairs with TERMINAL options when online recoverable sorting matters; batch-only jobs often omit COMMIT entirely.

Chaining SORT to JOB

After SORT completes, the next JOB in source order typically consumes TO output. Explicit JOB INPUT sort-file-name NAME job-name documents intent for maintainers. If INPUT is omitted and the prior activity was SORT, default automatic input is the SORT output file. Multiple SORT or FILE declarations break that default—always name INPUT when the program is not trivial. FINISH on JOB after empty sort output should still behave cleanly when BEFORE filtered everything out.

SORT Versus REPORT SEQUENCE

SORT reorders the entire input file for any downstream consumer. REPORT SEQUENCE reorders only spooled report rows after PRINT for one REPORT. Use SORT when JOB logic, multiple reports, or file extracts need shared order. Use SEQUENCE when one report's print order is the only requirement and a full-file sort would waste CPU. The sorting overview compares both paths in a table.

Common SORT Statement Mistakes

• Coding SORT inside JOB body instead of as its own activity.
• Key field offset mismatch with FILE layout—garbled order.
• BEFORE proc without SELECT on kept rows—empty TO file.
• USING order unlike REPORT CONTROL—breaks look random.
• Variable-length field as key—compile or sort failure.
• Forgetting JOB INPUT when multiple files exist in Library.

Testing Your SORT Statement

1. Compare input and output record counts without BEFORE filtering.
2. DISPLAY first and last key values on sorted file in JOB.
3. Verify CONTROL report breaks align with USING keys.
4. Test BEFORE SELECT with a known subset count.
5. Run empty input—confirm clean job end without ABEND.

Explain It Like I'm Five

SORT is a teacher who collects all the homework papers and stacks them in order before anyone reads them aloud. USING tells the teacher: sort by last name first, then by first name. The TO pile is the neat stack. JOB is the next class that reads only from the neat stack. BEFORE is a helper who throws away papers that fail a check unless they get a SELECT sticker.

Exercises

1. Write SORT with two ascending keys and NAME parameter.
2. Add VIRTUAL COPY layout for TO file.
3. Chain explicit JOB INPUT to SORT output.
4. Write BEFORE proc that SELECTs one region only.
5. List when you would choose SEQUENCE instead of SORT.