Easytrieve Sequential File Processing

Declaring Sequential Files

1
2
3
4
5
6
7
8
9
10
11
* Default sequential — fixed blocked
FILE PAYROLL FB(150 1800)
 
* Explicit sequential with status checking
FILE TRANSACT SEQUENTIAL FB(80 27920)
 
* Variable blocked extract
FILE LOGFILE VB(256 8192)
 
* Output sequential dataset
FILE ERRORS FB(150 1800) CREATE

File-name must be unique in the program. The first three characters should differ from the site Easytrieve work file prefix (often EZT) to avoid colliding with sort work datasets. SEQUENTIAL keyword is optional but documents intent and enables FILE-STATUS. FB means fixed-length records blocked together; VB means variable-length with block structure. Match numbers to IDCAMS or catalog listing for the real DSN in JCL.

Record Format Parameters Explained

DEFINE field positions assume a logical record layout after the access method strips block overhead. If LRECL is 80 but DEFINE extends to column 90, trailing fields read wrong bytes from the next field or filler. Use consistent copybook layouts from the upstream system that creates the extract.

Automatic Read With JOB INPUT

The simplest sequential batch pattern declares FILE and DEFINE, then starts a JOB activity with JOB INPUT PAYROLL. Each time the job loop cycles, Easytrieve issues the equivalent of a forward GET before your statements run. When the file exhausts, the activity terminates normally and FINISH procedures execute. You write IF conditions against current buffer fields, PRINT detail lines, accumulate totals with ADD or assignment, and PUT to output files. You must not code GET PAYROLL in the same activity while PAYROLL is automatic input—that duplicates reads and fails compile or runtime checks.

1
2
3
4
5
6
7
8
9
10
11
12
13
FILE PAYROLL FB(150 1800)
FILE HIGHPAY FB(150 1800)
 
DEFINE PAYROLL
  EMP-NO  9  1
  SALARY  P  9.2  10
 
JOB INPUT PAYROLL
  IF SALARY GT 100000
     MOVE EMP-NO TO HIGH-EMP
     MOVE SALARY TO HIGH-SAL
     PUT HIGHPAY
  END-IF

Controlled GET Loops

Use JOB INPUT NULL when you need explicit loop control: read until EOF inside WHILE or repeated GET, process multiple passes, or interleave GET with complex PERFORM structure. After each GET test IF EOF file-name before using fields. Optional STATUS sets FILE-STATUS for branching on I/O errors without aborting the step. GET PRIOR reads the previous record when the access method supports backward retrieval; if position was never established, Broadcom places the last record in the buffer. POINT and GET PRIOR interact with restrictions—do not GET PRIOR after POINT or GET after POINT PRIOR without reading the POINT chapter.

Writing Sequential Output

Output files are sequential unless you declare otherwise. PUT file-name writes the current buffer contents for that file to the dataset—typically after MOVE fields into an output DEFINE layout. JCL must allocate DISP=NEW, MOD, or CATLG per operations standards. CREATE on FILE signals the product may build a new file when disposition allows. Printer output uses FILE name PRINTER instead of disk sequential; REPORT activities format lines to SYSOUT. For extract jobs, one input sequential and one output sequential is the canonical teaching example before introducing keyed masters.

QSAM Versus VSAM ESDS as Sequential

QSAM datasets are plain sequential files on disk or tape. VSAM ESDS is entry-sequenced: records are ordered by insertion, browsed forward with sequential access. Easytrieve FILE SEQUENTIAL with ESDS JCL works for browse-only reporting. Random access by key requires INDEXED and KSDS—not this page. ESDS as sequential suits audit trail scans and log replay. Confirm SHAREOPTIONS and buffer space in IDCAMS match job concurrency; DISP=SHR on input is standard for read-only reports.

FILE-STATUS and Error Handling

When FILE specifies SEQUENTIAL explicitly, FILE-STATUS is available for that file. GET or PUT with STATUS also populates status for branching. Typical pattern: GET INPUT STATUS; IF FILE-STATUS NE 0 handle error. Without STATUS, a bad operating-system status often raises a diagnostic and ends the step. Sequential jobs on tape may hit volume switches; operations handle mount messages outside Easytrieve while your program continues forward reads after remount.

Performance and Buffering

Blocked FB files reduce EXCP counts versus small unblocked records. BLKSIZE should align with track capacity for disk (multiples of LRECL fitting half or full track per site standards). Reading entire file with JOB INPUT is efficient because the runtime buffers aggressively. Writing many small PUT records benefits from sensible block size on OUTFILE. SORT activities spill to work files when data exceeds memory; sequential FILE declarations on sort input and output follow the same FB and VB rules.

Sequential Versus Indexed Decision

Choose sequential when every record must be processed in arrival order, when file size fits one pass, or when the feed is already sorted by needed keys. Choose indexed when random lookups by key dominate—employee master by ID during transaction edit. Hybrid jobs use JOB INPUT sequential transactions and READ on INDEXED master inside job PROC—a pattern covered in multiple-files and indexed pages.

Common Sequential Mistakes

• Wrong LRECL or BLKSIZE in FILE versus actual dataset.
• GET on file that is automatic JOB INPUT.
• No EOF test after manual GET in WHILE loop.
• PUT without populating output buffer fields first.
• Using INDEXED READ against QSAM sequential DD.
• Assuming PRIOR works after POINT without checking rules.

Explain It Like I'm Five

A sequential file is a stack of cards numbered from top to bottom. You take one card off the top, read it, then take the next. JOB INPUT means the computer hands you the next card automatically each time. GET means you reach for the card yourself and must remember to stop when the stack is empty. PUT is putting a new card on another stack you are building. FB means every card is the same size; VB means some cards are longer than others.

Exercises

1. Code FILE, DEFINE, JOB INPUT for 80-byte FB input filtering on one field.
2. Rewrite the job with JOB INPUT NULL and GET with IF EOF guard.
3. Add PUT to write matching records to an output FILE.
4. Compare when you would use VB instead of FB for an interface file.
5. Write JCL DD for DISP=SHR input and DISP=NEW output sequential datasets.