MVSFORUMS.com

[abhikush]

Is there a way to achieve the following in sort
Input Records:
1
1
1
2
2
3
3
3
3
3

Output Record of File 1
1
1
1

Output Record of File 2
2
2

Output Record of File 3
3
3
3
3
3

The file may not always have these records the idea being splitting the files based on similar records.

[Frank Yaeger]

It's not clear what you're really asking for.

Do you mean that every time the key changes, you want to write to another output data set? If the key changes 1000 times, do you want 1000 output data sets? What is the actual maximum number of unique keys you can have? Are you going to supply a DD statement for each possible output file, or do you want to dynamically allocate the output files as needed, or ?
_________________
Frank Yaeger - DFSORT Development Team (IBM)
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

[abhikush]

Yes that is the idea. But what I am planning to do is split the file based on a key and if the keys are more than certain number say 15 then I will create 15 + 1 file. i.e. 15 files based on the key and then 1 file with the remaining keys.

I can then feed the new file created back to the same sort card and continue.

[Frank Yaeger]

Ok, let's assume your key is in positions 1-2, and that you will supply the 16 output DD statements needed. Here's a DFSORT job that will do what you want:

[abhikush]

Thanks Frank.

I will try it out.

[vishalknps]

[kolusu]

Vishal,

The following DFSORT/ICETOOL JCl will give you the desired results. If you have syncsort at your shop then use synctool as pgm in step0100.The first step reads the input dataset and creates dynamic control cards and also output files. The step0200 then submits another job to intrdr which actually splits the file into 'n' number of output files. You need to put your tid or the highlevel qualifier in place of USERID

For your understanding as to how the job works, remove the statement in step0200

[Frank Yaeger]

Clever solution, Kolusu. I added OPTION EQUALS to CTL2CNTL to make sure the records are sorted and summed in the correct order.

With DFSORT/ICETOOL, we can use SPLICE instead of the binary zero sum trick by replacing STEP0100 with the following:

[kolusu]

Frank,

Thanks for adding the option Equals parm. I always forget that because it is the installed default parm for our site. And I love to play with splice but unfortunately I can't. So I still have to go with good old binary zero sum trick

Kolusu
_________________
Kolusu
www.linkedin.com/in/kolusu