Re: SQL Existence Check

[Vernon Hamberg <vhamberg@xxxxxxxxx>]

Thanks, Leif

At 06:25 PM 3/12/02 -0600, you wrote:
>From: Vernon Hamberg <vhamberg@attbi.com>
>
> > OK, got ya. I'm just looking this up to learn a little more. Thanks for the
> > eye-opener.
> > This looks like physical layer implementation. What would have to happen
> > for a database to be relational at this level? Is not the 400 relational at
> > the user level?
>
>there is also the aspect of simple hype. I sometimes hear the
>AS/400 being described as object-oriented at the OS-level.

Yeah, but IBM, AFAIK, has always used the term object-based, if anything. I
think that's fair, because you can talk about a certain degree of
encapsulation on the 400. And possibly even inheritance, at a stretch, as a
whole bunch of things are actually dataspaces, with additional attributes
and functionality for each.

But the confusion reigns everywhere.

>If you access the database with SQL I guess that you can say
>that the DB is relational at the user level, except of course that
>most users never see or use the SQL. What is a user?
>The data-entry clerk?, the programmer?, the system programmer?
>
>My answer would be that IBM has built a relational front-end
>to an ISAM database. Other implementations that I know of
>may be closer to a relational database at the low level, e.g.
>one where each attribute (column, field, ...) is stored in its
>own separate file.

Date's book on DB design talks of 3 levels, or interfaces (I'm on thin ice
here, I assure you). I think they are application, user, and physical? The
user here is, in my understanding, something like a DBA, that defines how
things are mapped from the physical representation to some logical
representation. (OK, ENOUGH ALREADY)

I think he'd say that it's possible to implement any application-level
interface with almost any physical implementation - some are just not as
handy. Many of us have thought of SQL as an afterthought on iSeries, right?
It's always used the same ODP structures as the physical/logical file
system. But new stuff from IBM is no longer using ODPs, I believe, for SQL.
No longer using the same cursor-based stuff.