MVSFORUMS.com

[deepa12]

Hi
How to avoid s0c7 in comp fields since we cant do numeric check
Scenario:
suppose i have a LA-B in a file which is s9() comp
I read a record & have the following code
MOVE LA-B to WS-LA-B
The above move will fail if LA-B has some junk
So what to do
Thanks
deepa
_________________
deepa

[Terry_Heinze]

What prevents you from checking for numeric?
_________________
....Terry

[William Collins]

COMP fields are binary fields. The MOVE will not fail.

[deepa12]

Hi
Numeric check can be made only for X(5) sort of fields Not not COMP fields
Actually compiler will give error
_________________
deepa

[William Collins]

For NUMERIC you can check PIC X(n) fields (if the field is not too long), PIC 9(n) USAGE DISPLAY, USAGE COMP-3, USAGE PACKED-DECIMAL.

USAGE COMP, COMP-1, COMP-2, COMP-4, COMP-5, BINARY cannot be checked for NUMERIC simply because all bits may be relevant to the result, there is no "junk", there are no non-NUMERIC values (there are incorrect values, but unfortunately we can't test for that).

[Terry_Heinze]

Thanks, William. Sorry for the misleading answer, deepa12. Since all values are valid for COMP fields, one option would be to check for a valid numeric range, right?
_________________
....Terry

[kolusu]

[deepa12]

Actually i tried the folln but i get U4038 abend

ws-a pic s9(9)comp.
ws-b redefines ws-a pic x(4)
ws-c pic 9(9)
move 1234 to ws-a
compute ws-c= function numval(ws-x)

The requirement is i have to avoid soc7 in a comp field binary if it contains junk
LA-B S9(4)V9(4) COMP
WS-B S9(4)V9(4).
MOVE LA-B TO WS-B
_________________
deepa

[William Collins]

The requirement itself is junk.

There are no junk values in a binary field. Depending on how your field is defined, you will always get 0-maximum-picture or 0-maximum-binary value for an unsigned field, for a signed field your range will include negatives.

S0C7 only comes from one of the six decimal machine instructions. It is impossible that a decimal machine instruction will be used directly on a binary field, so impossible, impossible, can't be done, to get a S0C7 on a binary field.

NUMVAL will only work with "chjaracter" data. If you give it actual binary data, it will give you that abend.

Why use decimal places in a binary field?

[deepa12]

Thanks for this
Could you pls elaborate
on what are the six decimal machine instructions. Not clear about this

I dont know why thre record structure has been defined with virtual decimal in comp field
_________________
deepa

[William Collins]

You locate the Principles of Operation You search there. One is an Add Packed (AP). I think they are all dealt with together. There may be seven... let us know.

[kolusu]

[deepa12]

actually am an offshore developer It was suggested that we need to avoid s0c7 in this comp field
Am not having a real time scenario for this. i tried various combinations of non numeric value in WS-B E.g. ":,'%" , etc in all these combinations finally since its all in bits of 0,1 the subsequent instructions like add 1 to WS-A also works fine. So i think it wouldnt happen in live
_________________
deepa

[William Collins]

Well, I can 100% guarantee no S0C7.

If the analyst is so keen on this field, you could do the range-checking suggested by Terry. If there is no upper limit for the field, you could set your own as 9999.9999 and do something if the value is greater than that.

However, if using compiler option TRUNC(STD) this would be pointless as it would never find anything. If using compiler option TRUNC(OPT) it would be pointless unless someone is doing some "clever" coding with the filed (using REDEFINES, explicitly or imploicitly).

[deepa12]

thanks
_________________
deepa