Garret,

>>>Forced use of generated PKs
>>>I think you'll find that's not just a Fox position: I first read about
>>Surrogate Keys in Databased Advisor back in '94 or so.
>>Nope, this is not the case.

>Oh, so I _didn't_ read about Surrogate Keys in Databased Advisor. Thank you for kindly informing me of this....

I don't own that issue, but does it promote the forced use of surrogate keys ? if it does, is it in the same issue where the use of DELETED() tag is promoted ?

>http://www.elementkjournals.com/dbm/9708/dbm9781.htm

Both intelligent and surrogate keys pose their own problems. Some people belong to the "surrogates everywhere" camp, while others believe that intelligent keys are the only way to go. We believe in making the decision on a case-by-case basis."

It seems i'm not the only one not believing in the force use of surrogate keys.

>http://www.teleport.com/~bstonier/devhbook/databasedesign.html

A surrogate key is a single, meaningless, numeric column acting as the primary key in lieu of the columns that would otherwise be the primary key. I highly recommend the practice of using them in most situations."

>http://www.dbpd.com/vault/9805xtra.htm

A true promotion of the FORCED use of PKs. I found the following reference that directly violates the relational model:

Third, the primary key must be absolutely stable. It must be protected in all cases from change. This does not mean protected from likely change; rather, it must be guaranteed not to be allowed to change. This third principle is always violated when an "intelligent" key is used as the primary key

Hmmm, I wonder why cascading updates are a part of the relational model ?

>http://www.bcarter.com/intsurr1.htm

Besides the fact that Chris Date likes surrogate keys and Joe Celko likes intelligent keys, each alternative has a number of real-world advantages and disadvantages.

>Not an xBase article among them. They don't all argue for Surrogates exclusively, but some do....

My standpoint is that it is simply wrong to force the use of surrogated pks and agree with much of Joe Celkos standpoint. I would be more than satisfied if these references where used in a thread where this is an issue rather than a reply "Don't re-use primary keys, use generated pks".

Walter,