Re: Newbie question about db normalization theory: redundant keys OK?

"Tony Rogerson" <tonyrogerson_at_torver.net> wrote in message news:fk47j4$7oc$1$830fa79d_at_news.demon.co.uk...
> David,
>
> Stop trying to divert; you got it wrong; just leave it there.

I'm surprised if you assume there is a single right or wrong answer. I generally try to keep an open mind which is why I posed my responses as questions rather than claim that anything you said was wrong.

> ROWVERSION has nothing at all to do with this I really don't understand
> why you are trying to deflect the thread to that.

I agree ROWVERSION has nothing to do with it. I didn't introduce the topic. Brian did.

>
> The simple matter of the fact is that on revisting the database with the
> natural key you obtained (which should not have changed) has changed and
> points to completely different data.
>
> That is why we need an immutable key.

That's a straw man argument. If you design a solution on the assumption that data will not change then you will certainly have problems. I agree with you. I just don't advocate designing systems that way. Now if the rows referenced by your "immutable" key have been deleted, what result do you intend in that case? You don't assume that they will always exist do you? Of course you don't - you determine the right response and design the appropriate solution on the basis that the data may have changed.

>
> And... I'm still waiting for you to show me how to put history into
> celko's schema that would fix the problem where the natural key changes -
> why oh why are you STILL ignoring that request?
>

I already explained.

-- 
David Portas