Comparing comp 3 field with numeric field

deepak.kec · New User Joined: 26 Feb 2007 Posts: 71 Location: bangalore

Please let me know how can we compare the comp 3 with numeric field using DFSORT or ICETOOL.

PeterHolland · Posted: Tue Feb 26, 2013 7:48 pm Post subject:

What did the manual say ?

Robert Sample · Posted: Tue Feb 26, 2013 7:52 pm Post subject:

Using COND maybe?

Really, are you asking HOW to do the comparison -- if so, there are plenty of examples on this forum and the Internet that you could easily find yourself. Furthermore, if you are going to ask on a professional forum for work product (such as code), be prepared to pay the going rate for such code (typically on this forum you will pay $500 to $1000 US per day or fraction of a day spent coding).

Or, if you are asking whether or not such a comparison can be done at all -- READ THE MANUAL!

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7312

Yes, in the usual way. There is no problem comparing a PD to a ZD. Is that all you want to know,? As Peter and Robert have indicated, it is fully documented. There are links to the manuals in this forum and on the top line of the pages for the site.

deepak.kec · New User Joined: 26 Feb 2007 Posts: 71 Location: bangalore

Please let me know if there are any keywords which can used to get the information. I am trying to search in all the manual but it is taking lot of time.
I want to compare a pack decimal to numeric field in jcl.

You expert advise will be very helpful to me.

Robert Sample · Posted: Tue Feb 26, 2013 11:54 pm Post subject:

Skolusu · Posted: Wed Feb 27, 2013 12:03 am Post subject:

deepak.kec · New User Joined: 26 Feb 2007 Posts: 71 Location: bangalore

I am sorry for my ignorance and shall have given you more information.

I need to compare the two file, one is having data in packed decimal and other one is having in numeric. Need to compare them and write first file to output file when the matching record is found.
Both the files are Fixed block and are of different length.

daveporcelan · Posted: Wed Feb 27, 2013 2:44 am Post subject:

Five questions were asked and two of them were answered half way.

1. Half an answer given.
2. No answer
3. No answer
4. Half an answer (sort of)
5. No answer

Why does Skolusu or anybobdy ever bother?

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7312

We had no clue you wanted to compare between files.

Here is a sample. The thing for JOINKEYS with numerics is that they have to be in the same format, they have to be "normalised". JOINKEYS is just operating on "binary" data, not a field-type, so your data-fields have to be of the same type for it to work. All JOINKEYS key field "pairs" (one on file 1, one on file 2) have to be the same length, but can start at different positions.

I have chosen here to make the ZD into a PD, as it will keep the records shorter.

You do this in a JNFnCNTL dataset, with the "n" being 1 or 2 depending on which file has the ZD field (you haven't mentioned, so I've done it on file 2). The PD is created at the "end" of the Fixed-length record. My records are 80-bytes, so the new field starts at position 81.

The "default" for JOINKEYS is just to select the matched records. You only want file 1 data, so there is a REFORMAT statement to create the records from file 1 only. These records will be written to SORTOUT as part of the COPY operation (in this case) which runs as an integral part of the JOINKEYS processing, after the join has been made.

If your data is in the sequence of the numbers you are using, then you can add SORTED,NOSEQCK to the JOINKEYS statements. Otherwise a SORT, with EQUALS, will be performed for each file.