[OT] Obvious Is In The Mind Of The Beholder - JCL illustrIllustration

docdwarf_at_panix.com
Date: 11/15/04 

Date: 15 Nov 2004 13:37:34 -0500

All righty... so there's some processing being done that requires some
translation. What comes in is a file with an SSN and a job number... due
to some whacky accounting requirements there is billing that is done under
a 'pseudo-SSN' which is assigned to this combination. There's another
file which contains all three of these elements; given that the number of
records is moderate (about 70,000) the way to deal with this has been
decided to be a VSAM (indexed) lookup-file, with real-SSN and job number,
concatenated, as the key and the pseudo-SSN (which can be the same as the
real) delivered as the result.

Since this is 'this week's solution' I didn't put too much into it...
USER02 supplies a flat file and I put together some JCL to make a VSAM out
of it. Testing revealed a small problem... there were some records with
the same real-SSN and job number but different pseudo-SSNs associated with
them; the Corner-Office Guy said 'If we try to chase down why this was
done we'll be here until next month... to hell with it, just grab one
real-SSN/job number combination and use that, for now, we'll look at the
problem of dupes When We Have Time.'

So... I cobbled together a jobstream, nothing too pretty but it works...
and it looks like this:

//STEP000 EXEC PGM=IEFBR14
//DD01 DD DISP=(MOD,DELETE),
// UNIT=DISK,
// DSN=USERID01.PSEUDSSN.FLAT
//*---------------------------------------------------------------*
//STEP010 EXEC PGM=ICEMAN
//SORTLIB DD DSN=SYS1.SORTLIB,DISP=SHR
//SYSOUT DD SYSOUT=*
//SORTIN DD DISP=SHR,
// DSN=USERID02.FLAT.SSN
//SORTOUT DD DSN=USERID01.PSEUDSSN.FLAT,
// DISP=(,CATLG,CATLG),
// UNIT=FILE,
// SPACE=(CYL,(10,5),RLSE),
// DCB=(RECFM=FB,LRECL=26,BLKSIZE=0)
//SYSIN DD *
 SORT FIELDS=(1,9,CH,A,86,8,CH,A,12,9,CH,A)
 OUTREC FIELDS=(01:01,9,10:86,8,18:12,9)
 END
//*
//STEP020 EXEC PGM=ICEMAN
//SORTLIB DD DSN=SYS1.SORTLIB,DISP=SHR
//SYSOUT DD SYSOUT=*
//SORTIN DD DISP=SHR,
// DSN=USERID01.PSEUDSSN.FLAT
//SORTOUT DD DSN=USERID01.PSEUDSSN.FLAT,
// DISP=SHR
//SYSIN DD *
 SORT FIELDS=(1,17,CH,A)
 SUM FIELDS=(NONE)
//*
//STEP030 EXEC PGM=IDCAMS,COND=(0,NE)
//SYSIN DD *
  DELETE (USERID01.VSAM.PSEUDSSN) SCRATCH
  IF MAXCC NE 0 -
      THEN SET MAXCC=0
  DEFINE CLUSTER (NAME (USERID01.VSAM.PSEUDSSN) -
                  KEYS (17 0) -
                  VOLUMES (FILE25) -
                  CYLINDERS (5 5) -
                  RECORDSIZE (26 26) -
                  FREESPACE (15 20) -
                  REUSE -
                  SHAREOPTIONS (2 3))
/*
//*
//STEP040 EXEC PGM=IDCAMS,COND=(0,NE)
//INPUT1 DD DSN=USERID01.PSEUDSSN.FLAT,DISP=SHR
//OUTPUT1 DD DSN=USERID01.VSAM.PSEUDSSN,DISP=SHR
//SYSIN DD *
  REPRO INFILE(INPUT1) OUTFILE(OUTPUT1)
/*
//*

... and for those unfamiliar with the phenomenon... STEP000 deletes a
previously-made flat file, STEP010 reads the input and grabs the real-SSN
from position 1, the job number from position 86 and the pseudo-SSN from
position 12 and writes all of these to another flat file, STEP020 goes
through this newly-created file and removes the duplicates based on
real-SSN and job number, STEP030 deletes the former VSAM file and defines
a new one and STEP040 loads the de-duped flat file into the VSAM.

Nothing special, nothing pretty... but it works... for 'this week's
solution'.

Anyhow... another programmer stopped by my cube and asked about my
processing, seems like he'll be doing the reverse of what I'm doing and
trying to look up a real-SSN based on a pseudo-SSN; I said that this was
what I'd tossed together and he was welcome to it. Then... I felt kinda
bad about dumping so ugly into his lap and sent him an email, apologising
for the unattractiveness of what I'd given him and the lack of explanatory
comments in it.

His response was: 'No reason for apologies, it looked like a playboy
centerfold compared to some of my stuff and it was pretty self
explanatory.'

'Obvious' is in the mind of the beholder.

DD