Converting PD to ZD in VB file

r2k1984 · New User Joined: 21 Jun 2005 Posts: 69 Location: chennai

Hi,

My mainframe shop is having SYNCSORT and I'm having the following requirement.

1. Convert all the packed decimals in the file to zone decimals.
2. Input file is VB of 1000 record lenght.
3. In 210,225,245,235,255 lenght there are few packed decimals. But only for few records , details will be available after 210.
4. So when ever I try to convert the record after 210, the records for which no details are available it is not converting since the values for the fields short records are null.

Due to that I'm getting short record error.

Below is the JCL I was trying t use.

Bill Woodger · Posted: Wed Nov 09, 2011 12:32 am

If you have different record-types on the file, test the field which differentiates them and only attempt conversion of the fields on that record-type.

If you have occurs, with different numbers of entries, then test the count of the number of entries.

If you have both, do both.

If you have none of the above 1) you should 2) you could try to use the record-length.

EDIT: Are those decimal-points you are putting in most of the fields? That makes your field Numeric Edited not Zoned Decimal. Is that what you want?

dick scherrer · Posted: Wed Nov 09, 2011 12:46 am

Hello,

Surely not what you are looking for, but why does someone believe it is a good idea to change the data from packed to zoned? This will waste space as well as require more resources to process.

If some data needs to be downloaded to a pc for example, that data could be formatted to be pc-friendly without making a mess of the entire data file.

r2k1984 · New User Joined: 21 Jun 2005 Posts: 69 Location: chennai

We are migrating data from one platform to another platform.So we need to migarate history data also.

So at any cost we need to do this.

Akatsukami · Posted: Wed Nov 09, 2011 1:14 am

dick scherrer · Posted: Wed Nov 09, 2011 1:24 am

Hello,

Craq Giegerich · Posted: Wed Nov 09, 2011 2:18 am

There is nothing an auditor whould like more then going back to history info and not being able to tell debits from credits.

r2k1984 · New User Joined: 21 Jun 2005 Posts: 69 Location: chennai

Thanks for all your inputs .

I tried with VLTESTI=2 and 1 and also added VLSHRT. Job abends if I'm removing the include condition .

If I add the include condition ,job is working fine. I have a big occurs field and there can be random records for the fields.

Please provide inputs.

Below is the JCL I was using with include condition.

Bill Woodger · Posted: Wed Nov 09, 2011 2:59 am

I agree about the signs. Unless the there is something, external to the amount field which tells you with absolute certainty what is what, I can't see how you can go without them.

'd have done this in Cobol. Checking everything I can for logical consistency, a big report, printed cross-reference, cross-reference file... headers and trailers on everything.

Going out with a SORT doing a blind field conversion seems to be going out with a whimper, instead of a BANG! Have the users saying "that old system, it only had green screens, but boy that history report never lets us down".

Not my call, though, is it?

dick scherrer · Posted: Wed Nov 09, 2011 3:37 am

Hello,

You might try:
OPTION VLSHRT
SORT FIELDS=COPY

instead of:
OPTION COPY,VLSHRT
SORT FIELDS=COPY

for the run without the INCLUDE. Untested, but may be worth a shot. . .

r2k1984 · New User Joined: 21 Jun 2005 Posts: 69 Location: chennai

Thanks Dick for the inputs.

I tried both the options .Both are not working.

Any other ideas?

Bill Woodger · Posted: Wed Nov 09, 2011 4:52 am

In general the VLSHRT sort of stuff is going to affect SORT fields, INCLUDE/OMIT fields and SUM fields, I believe.

Consult the documentation to verify if there is some way to use it for this purpose.
Also check on the output lrecl message you are getting.

gcicchet · Posted: Wed Nov 09, 2011 6:50 am

Hi,

you could change your include to something like this this

r2k1984 · New User Joined: 21 Jun 2005 Posts: 69 Location: chennai

Issue with Short record is solved but I didnt get any reocrds in output.

Could you please explain the condition?

gcicchet · Posted: Wed Nov 09, 2011 11:18 am

Hi,
it's selecting records longer or equal to 240 bytes
sorry it should be

r2k1984 · New User Joined: 21 Jun 2005 Posts: 69 Location: chennai

I tried it. Now job is not getting abended and no records are found in output.

dick scherrer · Posted: Thu Nov 10, 2011 1:04 am

Hello,

Suggest you hex-dump a few blocks of data and make sure your displacements are correct. . .

Possibly there are no records that meet the first part of the INCLUDE?

Bill Woodger · Posted: Thu Nov 10, 2011 5:27 am

Is this a file with variable-length records but only one type of record?

If so, somewhere there must be a count of the number of occurences.

You'd have to do all your static fields, then do some test to know how many of the coversions to do for the variable fields.

If you proceed with what you have now (setting aside for the moment the fact that it does not work) won't you only be changing the long records, and leaving all the others with their original packed data?

kratos86 · Posted: Thu Nov 10, 2011 2:31 pm

Try using VLFILL parameter, which i believe would help in your case.