Re: 3 value logic. Why is SQL so special?
Date: Wed, 20 Sep 2006 18:35:50 GMT
Message-ID: <aSfQg.31537$9u.286622_at_ursa-nb00s0.nbnet.nb.ca>
peter koch wrote:
>>JOG wrote: >> >> >>>Bob Badour wrote: >>> >>> >>>>What is NULL if not the basis for 3 valued logic or 4 valued logic?
>
>
> My answer to that question is embedded below. I do not see why three-
> or four-valued logic should be forbidden.
I don't recall ever forbidding it. I ridicule it quite regularly, but I don't recall ever forbidding it.
Even less can i see how you
> can enforce users (database-designers) to create a four-valued type
> system if they should so desire.
Again, I have long espoused the idea to drop NULL and 3-vl in favour of an adequate data type system.
(Assuming the database has a
> sufficiently powerful typesystem: C++ comes in mind).
Ugh. C++ sucks as a typesystem. How about something better that is based on types instead of structures and pointers.
>>>An ugly hack to workaround the fact that most designers are either too >>>lazy or too lacking in pride to design their databases correctly? >> >>Um, well okay, but I guess I was asking for an answer from the folks who >>want to rationalize the ugly hack somehow.
>
> Well Bob.... being on usenet is not a full time pastime for me and
> currently I only have internet access from my job. My postings are thus
> bound to be somewhat sparse. But I do not see what NULLS have to do
> with database design.
Without an SQL dbms, who in their right mind would ever dream of using NULL?
I could perfectly well live with a situation
> where nulls were for some reason disallowed in "physical" tables.
Why? They are needed there even less than in the logical design.
(But
> do not see why they must be). The major problem of missing nulls is one
> of supporting the end users who query my base tables; here a left join
> is often quite nice to have.
NULL is not necessary for outer joins and introduces far more problems than it solves. Received on Wed Sep 20 2006 - 20:35:50 CEST