Re: Surrogate Key vs Production Key

DA Morgan <damorgan_at_x.washington.edu> wrote in message news:<1098159843.519191_at_yasure>...
> >
> > My take is, when Noons appears agitated, there's probably something
> > very interesting, worth looking at carefully. Usually something where
> > the general IT usage is incorrect or sub-optimal.

Shewt, if Joel starts paying attention to what I say, I know I'm in. Deep in the proverbial...

> Often. But in this case the issue of natural keys vs. surrogate keys
> is often more of a matter of religious fervor than anything else. I

No it most definitely is not! The ones in the wrong just need to be "liberated"! :)

> have seen many very heated discussions on the subject and many that
> turn quite unpleasant. Date and Cobb

Codd, as in the liver oil.

> were rather plain speaking on the
> subject but many people make assumptions about what is, and is not, a
> natural key.

My take is the definitions are too flaky to start with. Hence the lack of a unified approach. If I look at Codd's materials on this subject I have no doubt whatsoever he had not made up his mind on which way was right. Or rather: he had. But because no database maker implemented the full relational model, he had to consider a compromise. Pragmatic. Something he always was.

So Codd provided a set of guidelines, rules if you prefer, for BOTH camps. Those shielded us from getting too hang up in the labyrinth of data modelling. The arrangement has worked perfectly and I certainly can't see a reason to change it.

Regardless of what Oracle now might want us all to use to avoid problems with RAC. I'd rather Oracle fixed it to make it work no matter what. Just like they fixed the bind variables issue: at the root.

> In this case Noons and some others among which group I
> count myself ... more or less respectfully ... disagree.

Absolutely. And I certainly won't allow that to get between our mutual appreciation of single-malted infusions! Received on Tue Oct 19 2004 - 07:27:22 CDT