IBM Mainframe Forum

by **arla** » Thu Sep 25, 2008 10:22 pm

First post, so be gentle

I'm having a problem with using XSUM in a sort routine, until today (at least a year) this sort has worked just perfectly, and then today seemed to start giving "random" results as to which record it was eliminating.

Essentially the Key to my file is a 16 character field at position 26

I'm sorting two files together using this

Select all

//*-------------------------------------------------------------------
//* NOW SORT THE ACH INTO IT
//*-------------------------------------------------------------------
//S020 EXEC PGM=SORT
//SORTWK01 DD SPACE=(CYL,1000)
//SYSPRINT DD SYSOUT=*
//SYSOUT DD SYSOUT=*
//$ORTPARM DD DISP=SHR,DSN=TCTAA.CNVTKIT.SYSIN(GIN1001A)
//SORTIN DD DISP=SHR,DSN=KCTAA.COC.BKP.CPOL.OUTPUT(0)
// DD DISP=SHR,DSN=TCTAA.COC.ACHTEST
//SORTOUT DD DISP=(NEW,CATLG),DSN=TCTAA.TKIT.J129S010.CNC300S(+1),
// SPACE=(CYL,(10,10),RLSE),
// VOL=(,,,40)
//SYSIN DD *
SORT FIELDS=(26,16,CH,A,
42,4,CH,A,
46,4,CH,A,
19,4,CH,A,
69,5,CH,A,
50,4,CH,A)

As you can see, it sorts with the various sort fields, this always ends up with the record I am interested in is first, and the record I want elimited as a duplicate is second (verified repeatedly in file output)

For the XSUM part I'm using

Select all

//*-------------------------------------------------------------------
//* NOW SORT THAT ELIMINATING DUPLICATES (BASED ON POLICY NUMBER)
//*-------------------------------------------------------------------
//S030 EXEC PGM=SORT
//SORTWK01 DD SPACE=(CYL,1000)
//SYSPRINT DD SYSOUT=*
//SYSOUT DD SYSOUT=*
//$ORTPARM DD DISP=SHR,DSN=TCTAA.CNVTKIT.SYSIN(GIN1001A)
//SORTIN DD DISP=SHR,DSN=*.S020.SORTOUT
//SORTOUT DD DISP=(NEW,CATLG),DSN=TCTAA.TKIT.J129S010.CNC300S(+2),
// SPACE=(CYL,(10,10),RLSE),
// VOL=(,,,40)
//SORTXSUM DD DISP=(NEW,CATLG),DSN=TCTAA.TKIT.J129S010.CNC300S(+3),
// SPACE=(CYL,(10,10),RLSE),
// VOL=(,,,40)
//SYSIN DD *
SUM FIELDS=NONE,XSUM
SORT FIELDS=(26,16,CH,A)
//*-------------------------------------------------------------------

Which I believe should always eliminate the second record, and yet, sometimes it's eliminating the first record?

Anyone have any ideas?

by **dick scherrer** » Fri Sep 26, 2008 2:44 am

Hello Arla and welcome to the forums,

Anyone have any ideas?

My guess is that it is not now deleting the "first". What is happening is that records do not always sort into "first" and "second", so when the order changes, the "other" record gets deleted.

If you include EQUALS, this will keep the records within a sort key in the "same" order, so the first record remains the first record

by **arla** » Fri Sep 26, 2008 3:45 am

Okay, that seems to work, thanks a bunch, my $ORTPARM was

Select all

DYNALLOC=(SYSDA,16)
CORE=MAX

Followup question though, any idea why my code might have worked just fine for the last year or so, just luck of the draw?

by **dick scherrer** » Fri Sep 26, 2008 5:04 am

Hello,

any idea why my code might have worked just fine for the last year or so,

What is the chance that the first 2 became shuffled and not caught somewhere along the line?

just luck of the draw?

Possibly. That or something in the process caused the difference in how the records were manipulated. As IBM said about many things. . . "The results may be unpredictable"

You can put the EQUALS in the sort control info if you don't already have/don't want a $ORTPARM. I believe it may be set as a site option also (not positive on that). I check with the system programmer who supports Syncsort on your system.

IBM Mainframe Forum

Problem with XSUM and Syncsort

Problem with XSUM and Syncsort

Re: Problem with XSUM and Syncsort

Re: Problem with XSUM and Syncsort

Re: Problem with XSUM and Syncsort