Finding spaces in numeric elementary item in a group variabl

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

Hi
In a group varibale, how I can find that any elementary numeric data type is having spaces without checking individually all the elementary numeric data types for IS NUMERIC etc.

say in a DCLGEN I have moved a record from a file. After move Numeric data item got spaces in it. How I can replace those spaces to zeros.

for e.g

Bill O'Boyle · Posted: Thu Oct 15, 2009 9:41 pm

A space (X'40') could very well be a valid byte (except for the last byte) in a packed-decimal field. But it would be invalid in a display-numeric field.

Bill

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

thanks for the prompt reply , Bill,

Bill O'Boyle · Posted: Thu Oct 15, 2009 10:06 pm

OK, let's say you have a 2-Byte packed-decimal field and it contains PL2'400' (X'400C'). So, you have a X'40' in the first-byte, which is legitimate, but it would be in error if it was in the second-byte.

Are you defining "decimal" as packed-decimal or display-numeric (a/k/a zoned-decimal) or both?

If you have a display-numeric field, such as PIC 9(02), with a value of X'40F1', then this would be invalid, because a X'40' in this type of definition (regardless of the byte-location) is invalid.

However, as strange as this may sound, if this value were to be packed, the result would be PL2'001' (either X'001C' or X'001F'), because a PACK instruction discards the zone-portion of each byte, except for the last-byte.

This is also true if the display-numeric field contains x'4040' as it would pack as PL2'000', but will most likely be resolved (by the compiler) to an 'F' sign-nibble.

Bill

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

Hi Bill
I am having the decimal field as packed decimal.

regards

dbzTHEdinosauer · Posted: Fri Oct 16, 2009 2:30 am

tomehta,

if your input is garbage you need to validate each field.
if you send garbage to DB2, DB2 will return an error.

and a db2 decimal field is BINARY.

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

Hi DBZ
Input is not garbage, but is always spaces for comp-3.

Is there a way I can change these spaces to 0000 without validating each elementary field ?

Thanks

dbzTHEdinosauer · Posted: Fri Oct 16, 2009 2:46 am

Bill O'Boyle · Posted: Fri Oct 16, 2009 3:32 am

If you're referring to packed-decimal, then finding a X'40' in any byte of the field, except the last-byte (that would be an error), is certainly possible and legitimate.

Example -

Craq Giegerich · Posted: Fri Oct 16, 2009 5:51 am

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

For packed I can check for NUMERIC class test, but what I can test for the BINARY data type, SMALLINT of DB2 maps to BINARY in cobol, what data type class test I can run for binary ?
Thanks all for your help.

Robert Sample · Posted: Fri Oct 16, 2009 8:09 pm

A binary field in COBOL is numeric -- always. You cannot do a NUMERIC test on a COMP (binary) field for this reason.

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

but say by mistake the input file had spaces for the comp field , which is a invalid value, but system will take it as valid value. ( binary equivalent of space).
How to handle this condition that the input value for comp field is valid numeric field ?

dick scherrer · Posted: Fri Oct 16, 2009 8:58 pm

Hello,

Validate the input field. . .

Robert Sample · Posted: Fri Oct 16, 2009 9:14 pm

dick scherrer · Posted: Fri Oct 16, 2009 9:23 pm

Hello,

Redefine this field as pic x(4) and compare to spaces (or x'40404040'). . .

It is not difficult, but it is not automatic - you have to do a bit of work.

tomehta · New User Joined: 18 Aug 2008 Posts: 98 Location: India

thanks Robert , Dick , Bill ....i got the point, bit late but point well taken...

dick scherrer · Posted: Fri Oct 16, 2009 10:05 pm

You're welcome - good luck

Someone will be here if there are other questions later.

d

Marso · Posted: Sun Oct 18, 2009 9:33 pm

Here is another way to look at the problem:

Your DCL contains both CHAR and DECIMAL fields.
If you consider your fields as a group, you won't be able to tell if the space you find belongs to a CHAR field (legal space) or a DECIMAL field (maybe illegal).