Re: Artificial Primary keys

From: David Cressey <david_at_dcressey.com>
Date: Sat, 09 Feb 2002 10:54:30 GMT
Message-ID: <Gd798.72735$Nq6.14694_at_petpeeve.ziplink.net>


> The principle difference between the identity function and the
> invoice/SSN is that the IDENTITY value is not published outside the
> organisation that generates it.
>

Bernard,

I was thinking of replying to Joe, somewhat along the same lines as your reply. But I think you stated it better than I was about to. (sometimes, waiting does pay off).

Just to pursue the last point you made, a little further: The organization that generates a number, sequentially or not, to identify instances of some class of objects may have different levels of responsibility for the objects themselves.

Within an organization, the IT department might generate Ids for come class of objects. Or the department that "owns" the objects once they are generated might generate the Id. For example, new invoice numbers might be generated by the sales department, or they might be generated by some system that is run by IT. Or new invoice numbers might be generated by software that IT set up, and is responsible for, but being operated by the sales department, who might be responsible for gaps in the sequence. Ultimately, it makes no difference, as long as there are no duplicates, and as long as every issued number is catalogued somewhere.

In short, "generating" a new Id number inside a computer system is not fundamentally different from generating
an Id number by printing a different number on each preprinted invoice. I don't think I'm disagreeing with you here. I'm just making something explicit that might not have been completely explicit in what you said.

--
Regards,
    David Cressey
    www.dcressey.com
Received on Sat Feb 09 2002 - 11:54:30 CET

Original text of this message