relationship help

Jeff Boyce wrote:

Your questions/statements about "what about A...?" , (...B?, ...C?,
....D?),
your "one size fits all" comment, and your reference to ISO standards
together suggest the possibility that you are looking for a single
model
that fits all real world situations.

Oh dear, we are in a muddle aren't we g? I thought the same of you
i.e. assume Jeff type voice: 'I did a subclassed database for a
client once and now I try to apply it to healthcare scenarios.' I can
now see that isn't the case.

I'll be clear: I am familiar with the subclass approach. I am currently
working for a client with an extensively subclassed database i.e. an
Officer is an Individual is an Entity (I didn't name the elements g).
It works really well in this data model because the client is in the
business of modelling company structures. While it would be possible to
leverage this approach to, say, the Northwind example, I wouldn't
recommend it because there would be costs (e.g. attributes would be
split between entities which would have to be rebuilt using JOINs) and
no/few benefits.

In summary, I am trying to steer the OP away from the subclassed
approach suggested by Mike Sherrill ('When your doctor gets sick...')
because I think the level of detail isn't required. I'm done on the
subject, apart from asking you to please state whether you recommend
your subclassed approach for the OP of this thread (they'll be no come
back from me).

one size fits all

That's a no from me (although I can see scope for industry standard
schemas, as is happening in the XML world).

Your questions/statements about "what about A...?"

You got it right when you asked in another thread, 'is this your
learning style?' My answer is yes. Do I learn by asking questions?
Yes. Do I learn through discussion? Yes. Do I learn by asserting a
position and seeing if I can defend it? Of course. I can do this
because I retain an open mind. Is it a good learning style?
Questionable g.

I was going to ask whether you think you could learn something from me
but this in an unfair question. So I'll end with a warning: don't end
up like other MVPs with a closed mind. Take the following thread as an
example: do a google groups search for the exact phrase, "this should
have a bit of tolerance". I'm still waiting for my thanks for
introducing the guy to CHECK constraints g but equally I'm without a
reply to my question about whether the tolerance is required (I was
wondering whether NOW() could fail the CHECK if the clock ticked over
to the next second while processing). Should you wish to pick up on
that thread yourself ...?

Me, I have to go and eat humble pie dished out by an MVP in a thread
where I thought you couldn't INSERT a value to an autonumber column. I
think I prefer it when I'm wrong because I learn something.

Cheers,
Jamie.

--