Bill, I thought you just said you'd never seen this before on the mainframe!
:-)
I'm going to sidetrack this a bit to try to plug EXIT PERFORM. Yes, I know you
(Bill) don't like it. But I'm curious on others thoughts about the following
(taking your program and changing it):
----+----1----+----2----+----3----+----4----+----5----+----6----+----7--
ID DIVISION.
PROGRAM-ID. CALC1000.
DATA DIVISION.
FILE SECTION.
WORKING-STORAGE SECTION.
01 PIC X VALUE LOW-VALUES.
88 EXIT-OF-PERFORM VALUE HIGH-VALUES.
01 INPUT-DATA.
05 FILLER.
88 END-OF-INTERACTIVE-SESSION VALUE HIGH-VALUES
ZERO.
10 SALES-AMOUNT PIC 9(5)V99.
05 FILLER PIC X(73).
01 SALES-TAX PIC Z,ZZZ.99.
PROCEDURE DIVISION.
PERFORM UNTIL EXIT-OF-PERFORM
PERFORM READ-DATA
IF END-OF-INTERACTIVE-SESSION
EXIT PERFORM
END-IF
PERFORM CALCULATE-ONE-SALES-TAX
END-PERFORM
GOBACK
.
CALCULATE-ONE-SALES-TAX.
COMPUTE SALES-TAX ROUNDED = SALES-AMOUNT
* 0.0785
DISPLAY
SALES-AMOUNT
" SALES TAX = "
SALES-TAX
PERFORM READ-DATA
.
READ-DATA.
MOVE HIGH-VALUES TO INPUT-DATA
ACCEPT INPUT-DATA
.
The good (in my mind):
- There is one and only one "point of input". No explicit "priming read".
- It is more obvious that you are processing the input and then looping back to
read the next input. (And then exiting without processing (*) if there is no
more input.)
The bad (in my mind):
The COBOL standard (#) does not support:
- a comparison of two literals (IF 0 = 1)
- a "infinite/unbounded loop" (PERFORM UNTIL EXIT)
MicroFocus COBOL and GNU COBOL both have the "non-standard" PERFORM UNTIL EXIT
/ PERFORM FOREVER.
Fujitsu COBOL has the "non-standard" PERFORM WITH NO LIMIT.
IBM Enterprise COBOL has nothing similar and IBM refuses to implement one.
(Yes, I'm obsessive and upset about this.)
(*) The following might make it more apparent that we're not doing processing
if there is no more input:
PERFORM UNTIL EXIT-OF-PERFORM
PERFORM READ-DATA
IF END-OF-INTERACTIVE-SESSION
EXIT PERFORM
ELSE
PERFORM CALCULATE-ONE-SALES-TAX
END-IF
END-PERFORM
Frank
________________________________
From: IBM Mainframe Discussion List <[email protected]> on behalf of
Bill Woodger <[email protected]>
Sent: Sunday, July 10, 2016 11:24 AM
To: [email protected]
Subject: Error in a simple COBOL program
Here's one with the priming read, and a little trick for multiple ACCEPTs,
because ACCEPT doesn't give "end of file".
----+----1----+----2----+----3----+----4----+----5----+----6----+----7--
ID DIVISION.
PROGRAM-ID. CALC1000.
DATA DIVISION.
FILE SECTION.
WORKING-STORAGE SECTION.
01 INPUT-DATA.
05 FILLER.
88 END-OF-INTERACTIVE-SESSION VALUE HIGH-VALUES
ZERO.
10 SALES-AMOUNT PIC 9(5)V99.
05 FILLER PIC X(73).
01 SALES-TAX PIC Z,ZZZ.99.
PROCEDURE DIVISION.
PERFORM PRIMING-READ
PERFORM CALCULATE-ONE-SALES-TAX
UNTIL END-OF-INTERACTIVE-SESSION
GOBACK
.
CALCULATE-ONE-SALES-TAX.
COMPUTE SALES-TAX ROUNDED = SALES-AMOUNT
* 0.0785
DISPLAY
SALES-AMOUNT
" SALES TAX = "
SALES-TAX
PERFORM READ-DATA
.
PRIMING-READ.
PERFORM READ-DATA
.
READ-DATA.
MOVE HIGH-VALUES TO INPUT-DATA
ACCEPT INPUT-DATA
.
The trick is through the knowledge that when ACCEPT operates and there is no
remaining data, the target field (INPUT-DATA) remains unchanged. So set the
target-field to a known value which cannot exist (logically) in the input, and
you have an end-of-file test. In this case this is a catch-all for if the
"user" gets the "I don't want any more" response wrong when using a dataset for
input.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
