I don't want a subscribe button in the thread. But I can't articulate why, so carry on. Bureaucracy announcements seem fine.

Oh and John H. In terms of putting a primary non-meaningful key on each table. If you need some formal arguments to back up your view point google on "Surrogate Key" and "Date" (Of Codd and Date who invented relational theory).

Date has always been my favorite database theorist. Codd is the guy who made the big breakthrough - databases would work better if they were designed to conform to certain mathematical prinicples . But then he kept on adding more and more requirements without stopping to think what made sense.

It it was Date who said stuff like. OK - we need to deal with null in relational databases. But that doesn't mean we need true mathematical nulls, because the introduces four valued logic which is actually less powerful than two valued logic. Why don't we just have user defined null handling...? Codd? Codd? NOOO... DON'T PRESS THE BIG RED BUTTON!

I hereby solemnly promise to go away, read up on this, come back and find that funny, Gar!

uhh OK. So I'm a geek. But Nulls aside, if you need it, Codd totally has your back on the non-meaningful primary key thing.

We were discussing non-meaningful primary keys? When?

John H. mentioned as an aside that someone at his work is opposed to non-meaningful primary keys. John H is (rightfully in my opinion) strongly in favor of them. I was just mentioning that Date (of Codd and Date) gives strong theoretical support for surrogate keys - in case the guy might be impressed by that sort of cite. And then I made a joke about an old Date+Codd dispute over nulls, which exposed as having an unneccesarily obscure sense of humor. OK,not everyone has a favorite database theorist...

Sorry. I guess I skimmed. I thought it was in reference to the Phoenix.

Yeah - my fault for nattering in what is supposed to be a natter resistant zone...

It's not that the guy at work was in favour of them, more that he kept coming up with meaningful data that seemed to him like he'd found the right key, and I was trying to sell the concept of the arbitrary key to him.

I'm one step ahead of him, but roughly seventy-three steps behind you guys, it seems.

I've even run into this with fellow database analysts (who are supposed to bnow the theory). They claim to understand the concept, but see all these cases for exceptions - which makes it seem to me that they don't really understand the theory.