Host variables declared in File descriptor of FILE SECTION

rakesh17684 · New User Joined: 08 Oct 2006 Posts: 61 Location: San Diego

Hi,
I am trying to use a host variable in a query for a COBOL DB2 program.
The variable is declared in the file section

Bill Woodger · Posted: Mon Feb 20, 2012 12:41 pm

What does your manual say?

How do you feel this "reduces the number of variables" and why would you want to do that?

dbzTHEdinosauer · Posted: Mon Feb 20, 2012 12:59 pm

GuyC · Posted: Mon Feb 27, 2012 8:10 pm

If I recall correctly : addresses within file section vary within a block

Bill Woodger · Posted: Mon Feb 27, 2012 10:17 pm

Generally, yes. Read a record, new address. Write a record, new address.

So, no address before the first program instruction, and then a changing address.

I'd even suspect it tellls you in some manual somewhere not to even try this.

don.leahy · Posted: Mon Feb 27, 2012 11:05 pm

Yes, I am pretty sure it is documented, but in IBM-speak. Years ago, when I first read that section my reaction was "WTF?" After some research, I figured out what they meant and boiled it down to a simple rule of thumb: Always define DB2 Host Variables in Working Storage. Never use the FD area, never use Linkage. (Local Storage is okay too, but who uses that?).

My other rule of thumb: Where possible, always use the host variables defined in the DCLGEN. You can't go wrong that way.

enrico-sorichetti · Posted: Mon Feb 27, 2012 11:29 pm

quoting from the manual

dbzTHEdinosauer · Posted: Mon Feb 27, 2012 11:39 pm

for linkage section, you need to add a line of code to your program to move '1'
(or space, i forget which,
but you can tell how to turn the switch back on for each call
by looking at the compiler listing)
to INIT-FLAG as the first instruction in your program.
that way, each time the program is CALLed,
db2, again, goes thru the code which passes the addresses to db2.

because one needs to know what is going on in the computer,
it is generally safer to tell everyone only to use working/local storage.

linkage requires resetting of the INIT-FLAG,

Bill Woodger · Posted: Mon Feb 27, 2012 11:55 pm

And presumably don't get clever with SET ADDRESS, just because it is allowed in the Linkage Section.

dbzTHEdinosauer · Posted: Tue Feb 28, 2012 1:02 am

a little further clarification.

you have a db2 module that is CALLed by everybody to select some rows.
first time it is CALLed by prg-1 USING PGM-1 WORKING-STORAGE.
db2 module provides the addresses linkage section variables to db2, which are prg-1 working-storage.

then prg-2 CALLs db2 module with prg-2 WORKING-STORAGE.

if the db2 module does not have the INIT-FLAG reset logic,
which would cause db2 to be CALLed for addressing,
all the sql would use pgm-1 data areas for host-variables.
even though pgm-2 CALLed the db2 module (with pgm-2 using pgm-2 working-storage)

sad but true.

don.leahy · Posted: Tue Feb 28, 2012 1:46 am

In an IMS online environment, it can get even worse because the subroutine may remain in storage between transactions. Even if only one program calls the subroutine, that calling program may not be loaded in the same area of storage from one transaction to the next. It *usually* works okay, but there are no guarantees. I have had some very interesting debugging sessions caused by this issue.

Rather than try to explain all this to, er, less technically minded developers, I usually just trot out the rule of thumb. The smarter ones will ask why; there's no point in trying to explain it to the others.