MVSFORUMS.com

[misi01]

Since I'm working (or not) as a consultant without a contract at the moment, I'm not able to even try testing this on the mainframe. I'm sure an answer is out there, but I've googled phrases such as "dfsort sort key variable position" without success. Okay, here's what I would like to sort.

Assume I've done an unload of a DL/1 database (using a Cobol program and not a utility) containing segments A, B and C.
A is the root segment and B and C are children to A.
Assume the key field in A is account (number), and that the unload (via a Cobol program rather than an IMS utility) has ensured that the account from A has ALSO been written to
segment records B and C. The resulting file might look something like this:-

[Nic Clouston]

As it is an extract and not an unload why not have the extract add an extra field at the beginning of each record to hold the account number, sort on that and remove it in the output phase of sort?
_________________
Utility and Program control cards are NOT, repeat NOT, JCL.

[misi01]

Hmmmm, that seems like a simple, good idea.

The logic behind it is that I want to be able to perform regression testing in a DL/1 to DB2 conversion where both databases are updated by the same transaction. On that basis, I want to be able "dump" the DL/1 segments in such a manner that they "resemble" the dumped DB2 rows (I'll leave it at that without going into more details, since that would probably only muddy the waters).

Thanks Nic. (sometimes you can't see the wood for the trees)

If I ever get a new contract and can test it, I will.
_________________
Michael

[kolusu]

misi01,

It is quite simple. All you have to do is checking for the segment name and put the sorting key in a common place. Since the input is a VB file, you cannot add the temp field at the end of the record as it will ruin the Varible length concept.

We would overcome that issue by adding the temp fields right after the RDW. So by doing that we actually moved the contents to the right by the size of the key. In this case I assumed the account number to be 8 bytes.

I added temp field(8 spaces) using WHEN=INIT right after the RDW and by doing so the original contents are moved to position 13.

For "A" type records the account number would be at 24+8 = 32
For "B" type records the account number would be at 39+8 = 47
For "C" type records the account number would be at 54+8 = 62

Now using WHEN=A we overlay position 5 with contents from position 32.

Similarly you do the same for other types.

Now it is just a matter of SORTING the keys on the SEGMENT and Account number at position 5.

After SORTING you remove the temp sort field and build the record from position 13 so that you retain the original Variable length records as is

The following is a DFSORT JCL which will give you the desired results

[Nic Clouston]

Kolusu

I was suggesting adding it in the program that created the data in the first place so it would be one pas of both programs as is your solution.
_________________
Utility and Program control cards are NOT, repeat NOT, JCL.

[misi01]

Hey guys, I'm grateful for both suggestions.

Nic's because it was so simple and Kolusu's because I'm damn sure I'll need the concept at some time in the future.
_________________
Michael