unable to move pic 9(4) comp value to pic s9(4) comp

ms0033436 · New User Joined: 20 Nov 2008 Posts: 15 Location: Pune

Hi

My requirement is :
There are 2 variables X and Y.

X - IMS segment variable - pic 9(4) comp value 0001
Y is DB2 variable (DCLGEN variable ). - pic s9(4) usage comp.
My program uses cobol-db2 where the below is happening:
move X to Y.

There is a select query (Dynamic cursor) which uses Y variable in the where clause.

After executing the program it abends with SQL CODE -104 as it is expecting in this form '+0001' in DB2 table

can anybody has idea how to move pic 9(4) comp to pic s9(4) usage comp?

Robert Sample · Posted: Fri Jan 28, 2011 9:04 pm

Craq Giegerich · Posted: Fri Jan 28, 2011 9:09 pm

Deleted by poster.

Akatsukami · Posted: Sat Jan 29, 2011 1:51 am

Bill O'Boyle · Posted: Sat Jan 29, 2011 1:59 am

Compile the program using the LIST,NOOFFSET options (generate Assembler) and post the Assembler expansion of this move.

Bill

dick scherrer · Posted: Sat Jan 29, 2011 2:20 am

Hello,

I suspect that the actual move is not the problem but rather the definitions of the data in the database versus the program are not acceptable to db2 as presented.

Methinks Akatsukami is on target. . .

And the requested info needs to be posted for us to better help.

Bill O'Boyle · Posted: Sat Jan 29, 2011 2:38 am

Dick,

Remember the recent posting where the OP insisted that COMP was Packed-Decimal? The compiler should execute this COMP-to-COMP move in one MVC.

But, maybe Akatsukami is on to something?

Bill

david jackson · Posted: Sat Jan 29, 2011 2:39 am

dick scherrer · Posted: Sat Jan 29, 2011 2:55 am

Hi Bill,

Yup, i do recall. . . Unfortunately, we have very many with a complete mis-understanding of how data is actually stored for the various data types. . .

And it is almost like pulling teeth to get the useful info posted . . .

Maybe it will follow soon

d

Bill O'Boyle · Posted: Sat Jan 29, 2011 2:57 am

After further review, perhaps the compiler is loading the unsigned-halfword into a register and then issuing a "LOAD POSITIVE REGISTER"?

This would ensure any negative value becomes positive.

After the LPR of itself (or another register), it probably issues a STH (or maybe an STCM) to the signed-halfword and we're done?

Yeah, an MVC would not completely do the trick, because you're going from an unsigned-halfword to an signed-halfword and the receiving field needs to be a signed-value.

But, I agree, this MOVE should work as-is....

Bill

david jackson · Posted: Sat Jan 29, 2011 6:35 am

Bill O'B,
In the direction of a move of an unsigned to a signed field, a straight MVC will be fine as the from field is by implication positive.
The opposite move however of a signed filed to an unsigned, will as you indicate result in a LH, followed by a LPR and a STH to ensure the result is positive.
None of this helps the OP however...!

Kjeld · Posted: Mon Jan 31, 2011 8:17 pm

ms0033436 · New User Joined: 20 Nov 2008 Posts: 15 Location: Pune

[S9(4) Comp is a half word binary field and would contain x'0001' if positive (or for example x'FFFF' if negative 1).
What exactly is the form that DB2 is expecting this in as you say it is expecting '+0001'? Can you clarify that point?
]

David,
i am getting the below message UNEXPECTED SQLCODE OF -104 ENCOUNTERED IN program CAUSING U3920 ABEND.
SQLCODE = -104, ERROR: ILLEGAL SYMBOL " ". SOME SYMBOLS THAT MIGHT BE
LEGAL ARE: + .

SQLSTATE = 42601 SQLSTATE RETURN CODE
SQLERRP = DSNHPARS SQL PROCEDURE DETECTING ERROR
SQLERRD = 3 0 0 -1 250 502 SQL DIAGNOSTIC INFORMATION
SQLERRD = X'00000003' X'00000000' X'00000000' X'FFFFFFFF'
X'000000FA' X'000001F6' SQL DIAGNOSTIC INFORMATION
BAD SQLCODE ON PREPARE QUERY AGAINST DB2 TABLE PARA
PROGRAM ABNORMALLY TERMINATED
PTOM DUMP OUTPUT 383
Due to this i am thinking that + is need in that variable. but when i display all the values used in the select query are coming correctly
this value comes with out a sign - ie. 0001

Bill O'Boyle · Posted: Wed Feb 02, 2011 5:54 pm

If a halfword contained X'0001' (H'1'), the high-order sign-bit is "Off" and therefore, the value is positive, so it could be defined to an unsigned or signed halfword picture clause and utilized as such.

As you've said, a negative halfword one is X'FFFF' (H'-1') and the high-order bit is "On".

There's something we're missing here, because your code (based upon what you've posted), should work as-is.

Have you posted all there is to this problem?

Just a WAG, but perhaps these halfwords need to be aligned (a/k/a COMP SYNC)?

Bill

dbzTHEdinosauer · Posted: Wed Feb 02, 2011 6:10 pm

The OP's problem is that his generated dynamic sql SUCKS

the error message ---ILLEGAL SYMBOL " "
says it all.
the OP's refuses to post the generated slop that he feeds DB2 and
only responds to posters (like David) that provide a potential excuse
not to be responsible for his code.

Bill O'Boyle · Posted: Wed Feb 02, 2011 7:14 pm

Dick,

Yes, absolutely, Kjeld's post said it all, but I missed it completely < sigh >

Bill

Kjeld · Posted: Wed Feb 02, 2011 7:29 pm

I think OP has confused the format of host variables that are populated by the query and host variables that introduces dynamic literals in the where statements, and thats why we need to see the SQL-code and host variable contents that causes the syntax error.