Re: Externalizing input and output.

["James W. Kilgore" <qappdsn@xxxxxxxxxxxxx>]

Brian,

Using embedded SQL is a good idea, but if you do this within your
interactive application you will still suffer the wrath of CFINT.

If one wishes to externalize the I/O it can be done by passing data via
a service program.  The application would pass to the service program
the key fields and a record format to return.  Just as many of IBM's
APIs have different formats, so would your file server service program.

Now depending on your long term goals, you could bind this service
program to your application program, but your interactive sessions would
still face the wrath of CFINT. So, you bind for the batch jobs and use
data or message queues for your interactive jobs to a batch running
server program.

Here's another thought, if you don't want to go the SQL route, maybe
someone has already tried this:  Use logical views explicitly listing
the fields desired.
One question though, if you run a CHGPF after adding new fields at the
end, do you still need LVLCHK(*NO) even though nothing in the logical
has changed?

Brian Parkins wrote:
>
> Using embedded SQL to perform I/O offers a degree of isolation from the
> file layout.  This would be a first step.
>