Re: Database I/O Modernization

On Fri, Sep 8, 2017 at 5:03 PM, Alan Campin <alan0307d@xxxxxxxxx> wrote:

Not really. You are limiting SQL. This allows a single table. What happens
if I need to join together three tables. With SQL, I just write the
statement or create a view.

You're right. The interface is limited to a single table or logical file.
The primary purpose for this type of interface appears to be individual
record inserts, updates, and deletes (e.g. CRUD functions). It doesn't
appear to be intended for queries. It wouldn't work for joins. Should that
alone invalidate it?

There is a way to write these things that is high efficiency and multiple
tables but again why bother if you have SQL.

Are you suggesting that SQL should not be encapsulated in CRUD procedures?

And the basic problem is here that if you are stuck in the past and still
using RLA instead of SQL why would you use procedures? You wouldn't. You
would avoid them like the plague. That's why IBM had to come up with OA.

Are you suggesting that RLA and procedures are mutually exclusive for IBM i
developers?

Anything you can do in OA you can do as good or a lot better in a procedure
and a service program but IBM knew that most RPG programmers would never
use a procedure so they created OA.

You appear to be rejecting a mind-set that you don't like. Should that
invalidate the interface? Or is there more to your objections?

Companies I used to work at I would write service programs but the
programmers demanded that I write a callable program wrapping the procedure
because they would not touch a procedure. Same thing were ever I have been.

They must have been stuck in an RPG III rut.