non-positive reference modification error

ashutosh.pr · New User Joined: 13 Apr 2007 Posts: 36 Location: Pune

Hi,
We have to find if the contents of a character field (WS-VAR1 in example) are numeric and then accordingly proceed in the logic.
We are using the following logic:

---working storage---

WS-VAR1 PIC X(30)
LEN1 PIC 9(18) VALUE 0
LEN2 PIC 9(18) VALUE 0
--------------------------

Logic is as follows: -

*--- WS-VAR1 is a char of length 30 and we have to find the length of its contents--
INSPECT FUNCTION REVERSE(WS-VAR1)
TALLYING LEN1 FOR LEADING SPACES.
COMPUTE LEN2= 30 - LEN1

*-- then we are checking if the values in the field are numeric

IF WS-VAR1(1:LEN2) NUMERIC
MOVE 'XYZ' TO WS-VAR3
DISPLAY 'Is Numeric'
END-IF
DISPLAY 'Not Numeric'

The logic tested fine initially . However, now my peer has got the following error when WS-VAR1 is spaces. He says, that in case WS-VAR1 is spaces, LEN1 will be 30 and LEN2 will be 0.

The below line according to him is giving error:
IF WS-VAR1(1:LEN2) NUMERIC (*-- LEN2 being 0)
Error :- A non-positive reference modification length value of 0 on line
xxxx was found in a reference...

However when I run the program I get the following results through displays:

WS-VAR1 = ' ' (To test this, I passed spaces in this variable a little earlier in logic)

LEN1 = 000000000000000020

LEN2 = 000000000000000000

'Not Numeric'

So, I am not getting the same error, though the conditions are the same.

Can anyone help here? I tried searching for this kind of issue but could not find anything, except that value of length should be +ve (greater than 0. But I am not getting any error for this.

Am I missing something here?

Thanks,
Ashutosh

Girish Firke · New User Joined: 04 Feb 2009 Posts: 5 Location: Mumbai

Hi Ashutosh.

ashutosh.pr · New User Joined: 13 Apr 2007 Posts: 36 Location: Pune

OOPS!!
Thats a Typo..
These are the displays:

LEN1 = 000000000000000030

LEN2 = 000000000000000000

Thanks

dbzTHEdinosauer · Posted: Fri Sep 04, 2009 3:18 pm

you have provided nothing but pseudo-code.

you need to supply the data layout and the code of your module,
cut & paste.

another typo and you can solve this problem yourself.

obviously, had you the same code as your peer, you would have the same error.

you are not having the same error, and without your data definitions/code, we can not offer any solutions.

Terry Heinze · Posted: Fri Sep 04, 2009 6:25 pm

Not to stray off topic, but what is all this crap I've been seeing lately where -ve is supposed to mean negative and +ve is supposed to mean positive?

Have these bulletin boards become "chat speak" outlets for people or what? Does it really take that much extra time out of our lives to use proper terminology and spell words correctly?

Marso · Posted: Sat Sep 05, 2009 11:10 am

Terry, I agree with you. I've seen too a "-ve" post and had to scratch my head for a few seconds before I understood what it means.
However, in this case, it is apparently the way IBM has phrased its error message.

The Enterprise COBOL Language Reference says , about Reference modification:

Marso · Posted: Sat Sep 05, 2009 11:23 am

Oooh, OK, now I've seen a "+ve" too....

Anuj Dhawan · Posted: Mon Sep 07, 2009 5:48 pm

I believe this

ashutosh.pr · New User Joined: 13 Apr 2007 Posts: 36 Location: Pune

Apologies for using the short forms here on the board. I will take care of this from the next time.

Regarding the error that my colleague is getting, its in another program which is having exactly same logic (this was a change made in several of the programs for a business requirement). We currently do not have access to run that particular program, but the doubt that we have here is,
"IF LD-SRCH-BANK-REF-CD(1:SUB06) IS NUMERIC " is not giving an error when the value of SUB06 is Zero.

Regarding the actual code. Please find the code below and the Display spool , copy pasted from the code. We have delibirately moved spaces to
the variable LD-SRCH-BANK-REF-CD to get the error, but the program is not failing (moving spaces to LD-SRCH-BANK-REF-CD is not the part of logic).

----Working Storage--
05 SUB05 PIC 9(18) VALUE 0.
05 SUB06 PIC 9(18) VALUE 0.
05 LD-SRCH-BANK-REF-CD PIC X(20).

--- Code----

MOVE ZEROES TO SUB05,SUB06
MOVE SPACES TO LD-SRCH-BANK-REF-CD
INSPECT FUNCTION REVERSE(LD-SRCH-BANK-REF-CD)
TALLYING SUB05 FOR LEADING SPACES
COMPUTE SUB06 = 20 - SUB05
DISPLAY 'SUB06 IS :' SUB06':'
DISPLAY 'BANK-REF-CD :' LD-SRCH-BANK-REF-CD':'
IF LD-SRCH-BANK-REF-CD(1:SUB06) IS NUMERIC
DISPLAY 'IN NUMERIC'
MOVE ALL ZEROES TO WS-SRCH-BANK-REF-CD
ADD 1 TO SUB05
MOVE LD-SRCH-BANK-REF-CD(1:SUB06) TO
WS-SRCH-BANK-REF-CD(SUB05:20)
MOVE WS-SRCH-BANK-REF-CD TO LD-SRCH-BANK-REF-CD
END-IF
DISPLAY 'NOT IN NUMERIC'

----- display output ---
PROGRAM STARTED ON 2009-09-08
PROGRAM STARTED AT 02:27:13:30

SUB05 IS :000000000000000020:
SUB06 IS :000000000000000000:
BANK-REF-CD : :
NOT IN NUMERIC

Marso · Posted: Tue Sep 08, 2009 8:43 pm

Anyway, it shouldn't be too difficult to add

ashutosh.pr · New User Joined: 13 Apr 2007 Posts: 36 Location: Pune

Thanks Marso,
Yes, we have made the changes as the previous logic did not look very stable.
But I am still not sure why even explicitly setting the length value to zeroes has not given the reference modification error.
Another thing that we noted was that the error which was encountered was in an online CICS program ( with the same logic), but we did not get any errors in the above program that is a batch program.
Does anyone know if online or batch environments make a difference?