Identifying Packed Decimal data in a dataset

sojivarkey · New User Joined: 07 Dec 2019 Posts: 1 Location: USA

Hi,

I'm working on a project where we have to move around 300K datasets to a Cloud location. Many of these datasets contain packed Decimal data but identifying the copybooks for all these 300K datasets is going to take forever. My question is , is there any tool or utility available that will tell me if a particular dataset has any COMP-3 data in it. If I can have the datasets with COMP-3 data identified, I can just focus on those datasets alone to pick a copybook rather than looking for 300K datasets.

Any help is much appreciated.

Thank You

Robert Sample · Posted: Sat Dec 07, 2019 9:10 am

There is no way to tell if data is packed decimal without an external reference. X'97994D' could be "pr(" if it is character data but it could be packed decimal -97994. Which is correct? Only a program or copy book can tell you.

Phrzby Phil · Posted: Sun Dec 08, 2019 1:58 am

A few ideas here from a retiree who can therefore no longer test these ideas.

Perhaps a (small?) number of datasets could be confirmed as to not having packed decimal data if they have no nybbles containing valid rightmost nybble values in the same positions of all records.

Datasets with variable length records would be a problem.

I don't remember if all low-values (so rightmost byte x'00') is handled as zero for certain arithmetic operations. If so that would be a problem.

Spaces (x'40') in what should be packed decimal fields would also cause an analysis problem if programs know that can happen and check for numeric before using such fields, so some records might have x'40' in a rightmost byte where other records show packed decimal.

sergeyken · Posted: Mon Dec 09, 2019 9:02 pm

sergeyken · Posted: Mon Dec 09, 2019 9:09 pm

In general, it is possible only to detect the "text-only uppercase only" datasets.

All other types of data can be anything; cannot detect without a copybook.