Facilitate use of COBOL records in EXEC SQL multi-row operations.

We have looked into this RFE and have determined that this cannot be done as part of the COBOL compiler. It would need to be done via DB2. As a result, can you please open an RFE with the DB2 team via the following link:
https://ibm-data-and-ai.ideas.aha.io/

We will be rejecting this RFE from a COBOL point of view. Please let us know if you have any issues opening an Aha request via the above link. Thanks.

I suggest to manage the generation of occurs on the elementary variables by adding a statement "SQL TYPE IS ROWSET FOR n ROWS" at the level of the structure including these elementary variables.

In source before compilation :

01 X SQL TYPE IS ROWSET FOR 100 ROWS.
05 Y1 ...
05 Y2 …

After pre-compilation :

01 X .
05 Y1 … OCCURS 100.
05 Y2 … OCCURS 100.

Due to processing by IBM, this request was reassigned to have the following updated attributes:
Brand - Servers and Systems Software
Product family - Programming Languages
Product - COBOL Compilers

For recording keeping, the previous attributes were:
Brand - Rational
Product family - Design & development
Product - COBOL Compilers

>> multi year strategy. << ? I think that the "dimacross" option should be available _as soon as possible_! We want to tune our applications and using rowset is a big point. But if you show how to code that in cobol today noone wants to think about using it.

Please think about that point. Thanks.

PS: Supporting 24 bit (RFE 36563) should have much lower priority!!!

This is in-line with our multi year strategy. We will update this request once we plan it for a future compiler version.

An explanation to the code of http://www.cps4it.de/wikifiles/Schulungen_cps4it/ROWSET_COBOL_U441RLF.txt. That was the program to test the performance of fetch rowset vs. fetch row. Therefor the different sections and the loops . . .

Thanks.
Ralf

Hi IBM_Rational_Developer,
I don't understand your question. It should be clear how to code a sql fetch with rowset. ;-))

If not - have a look to http://www.cps4it.de/wikifiles/Schulungen_cps4it/ROWSET_COBOL_U441RLF.txt . That's an example I tested at AXA in 2009 / 2010. The code hat both versions of fetching - row and rowset.
Please have also a look on MR051111354.

A SQL INSERT works in the same matter.

The problem is: You can't use DCLGEN for generation the copybook, write a level number above the structure with occurs. Actually you must code the occurs clause for each field seperately. That's crazy!

Another possibility to look for code is the example in the SQL-Book from IBM. ;-))) That example should work at IBM machines.

. . . or haven't I understood your question?

Hi DieterHarms,

Can you please explain how users code multi-row fetch with SQL and COBOL today? It would be good to get an example.

Also, can you please elaborate a bit more on the Use Case ie "All EXEC SQL operations which might become more effective by using multi-row mode." ?

Thanks.

This is a duplicate to GGADL11014 / MR051111354. Title: COBOL: Improved Support for SQLs 'FETCH WITH ROWSET'

Text in that requirment:
It's very useful to have an option like DIMACROSS in PL/1 to define the
OCCURS-clause in the higher leveled variable like:
05 struc OCCURS n ROWSET.
10 Field-A pic anything.
10 Field-B pic anything.
etc.
Explanation:
Fetching DB2 rows with the option "with ROWSET" needs to define the varia
like e.g.:
05 struc.
10 Field-A pic anything OCCURS n.
10 Field-B pic anything OCCURS n.
etc.
To be able to use copybooks for that strukture it's very useful to have an optio
like DIMACROSS in PL/1 to define the OCCURS-clause in the higher levelle
variable like:
05 struc OCCURS n ROWSET.
10 Field-A pic anything.
10 Field-B pic anything.
etc
In that case it's possible to use a copybook with
10 Field-A pic anything.
10 Field-B pic anything.
etc.
the option ROWSET writes the variables in that sequence like the definition
before and the fields of the strukture can easily be used for fetch with rowset.