Re: Proper way to do some file operations with SQL

Hi Jeff,

Adding to the suggestions you already received...

1) Update the control file row under commitment control, and don't commit
until you're done with the unit of work. This will hold a lock on the
control file row until you commit. Other programs have to be written to be
compatible with the presence of a temporary lock.

2) Consider deploying some SET based logic. When it comes time to write the
header and detail records, you likely have all the information you need to
write the header initially complete, without the need to update it later.
When it comes time to write the detail rows, try using an INSERT that adds
all of the rows as a set in one call.

Mike

On Tue, Apr 17, 2018 at 8:24 AM, Jeff Crosby <jlcrosby@xxxxxxxxxxxxxxxx>
wrote:

All,

I'm doing a small project where, for the first time, I'm trying to do /ALL/
the file ops via SQL. I have a couple of questions on the 'proper' way to
do this in SQL.

1) I'm generating a credit memo and need to use the next available credit
memo number. With RLA, one CHAINs to the control file, adds 1 to the last
credit memo number used, updates the control record, then uses the number
you just got. No one else can get this control record until you release it
so there are no duplicates. What's the proper way in SQL?

2) There's a header record and x number of detail records. The norm in RLA
is to write the header, write the detail records, then CHAIN back to the
header and update it with summary information from the detail records. I
assume the same sequence with SQL. Is there a slick way of doing this?

Thanks.


--


Jeff Crosby
VP Information Systems
UniPro FoodService/Dilgard
P.O. Box 13369
Ft. Wayne, IN 46868-3369
260-422-7531
direct.dilgardfoods.com