Loading SMALL INT Datatype into DB2

rarvins · New User Joined: 24 Jan 2007 Posts: 59 Location: Texas

Hi,

My requiremnt is that I have to produce a file with a COBOL program which can
be loaded into a DB2 table through the IBM load utility.

Say I have Table A with COL1 of datatype SMALLINT.

I am done with the processing logic and I write the output into a file
containing just one datatype S((04) COMP, so that i can load this file in to the
table.

But the problem is something like this:
I have moved a value 27797 to the S9(04) COMP.
But only 7797 is getting populated in the variable
resulting in data loss.

But when I run an insert query I able to insert 27797 into COL1 of TableA.

I dont understand where I am going wrong.

1) Cant S9(04) COMP hold a value as big as 27797 ?
2) Is there any other dataype or declaration that I have to use?

Can somebody help me out...

vini_srcna · Posted: Mon Apr 02, 2007 2:19 pm

What is the equivalent datatype given by DCLGEN..?
Are you not making use of DCLGEN..?
It is meant to get rid of this kind of situations.

rarvins · New User Joined: 24 Jan 2007 Posts: 59 Location: Texas

This is S9(04) COMP is the declarationg given by the DCLGEN

Phrzby Phil · Posted: Mon Apr 02, 2007 5:50 pm

A halfword binary can hold your value, but COBOL strictly honors the field definition, which says 4 digits, so the compiled COBOL code truncates the leftmost digit.

E.g., for PIC 9(1) COMP, although a halfword is used, only unsigned single-digit values are stored.

Since PIC 9(5) COMP thru 9(9) COMP allocate a fullword, I'm not sure how you can do this directly in COBOL.

I think it is unwise to store a 5-digit number in a halfword or SMALLINT, as it will not hold any 5-digit number. That's why PIC 9(1) COMP thru 9(4) COMP use a halfword - all 4-digit values can be accomodated.