Re: 3 value logic. Why is SQL so special?

From: J M Davitt <jdavitt_at_aeneas.net>
Date: Thu, 21 Sep 2006 00:49:58 GMT
Message-ID: <WklQg.7334$L15.2173@tornado.ohiordc.rr.com>

Chris Lim wrote:
> Roy Hann wrote:
>

>>It's worse than that.  Suppose one user can't be bothered to find out if the
>>customer is insolvent and another user decides the null means insolvency or
>>otherwise isn't a relevant attribute.

>
>
> I don't see the problem with NULLs. Yes you have to be careful when
> using nullable columns in queries, but it's not the mess that you guys
> make out (at least not in practical terms). It's certainly easier (when
> writing queries) to deal with NULLs than to handle separate tables for
> every optional attribute, which is the proposed solution (considering
> how many NULLs there are in a 'typical' database, the number of extra
> tables is enormous).

Nope. Not at all.

It is, though, to paraphrase a poser, "certainly easier when designing a database to incorporate NULLs than to handle separate tables."

> This is one argument where the theory just does not seem to have a
> case, at least until SQL supports a better way of combining all the
> optional attributes of an entity together to make querying easier.

WTF does SQL have to do with "the theory?"

  I
> mean, with the whole surrogate key vs natural key debate I can see both
> sides to the argument, and even though I wouldn't do it personally, I
> can see how a database with purely natural keys could work. But a
> database without NULLs? It might be theorectically correct, but it
> would be a nightmare to write queries against.

Nope. Not at all. Received on Wed Sep 20 2006 - 19:49:58 CDT