Re: c.d.theory glossary - NULL
Date: Tue, 27 Apr 2004 11:05:53 -0400
Message-ID: <y5SdnQL6dKAF6xPd4p2dnA_at_comcast.com>
My attitude towards SQL is a compromise between theory and reality.
If I want to create a database tomorrow, I'm going to use SQL. Regardless
of its flaws.
Or maybe I'll use point and click on some wizard. But the wizard tends to
convert my commands into SQL anyway.
My understanding of the RDM may be quite different from yours. I'm willing to listen.
Here's where I differ from you: a deviation from true normalization is not necessarily, a "failure in modeling", IMO. Normalization is only one, among many, possible rules to follow, when designing a database.
As I said earlier, every NULL in a dtabase can be construed as the result of an outer join. From that, you may infer that every database where NULLS are allowed is not fully normalized. Not fully normalized isn't necessarily a "Failure".
Cool hand Leandro, what we have here is a failure to communicate. Received on Tue Apr 27 2004 - 17:05:53 CEST