Re: Does Codd's view of a relational database differ from that ofDate& Darwin? [M.Gittens]

From: Paul <paul_at_test.com>
Date: Tue, 07 Jun 2005 11:22:37 +0100
Message-ID: <42a5756d$0$41940$ed2619ec@ptn-nntp-reader03.plus.net>

mountain man wrote:

>>Here's some discussion of a very concrete example of a NULL problem in
>>SQL: http://www.firstsql.com/iexist2.htm
>>
>>The author says that his RDBMS solves this problem though, and that is
>>just a problem with SQL and NULLs, not with the relational model and
>>NULLs.
>>
>>Are there any rebuttals of this claim that anyone has?

>
> Yes, in respect of his "problem with SQL and nulls" note that
> his concrete problem as expressed in the above article is the
> behaviour of a standard query *converted* to using an
> "EXISTS" clause.
>
> You'll note that there are absolutely no problems
> with the *standard* SQL query results, even though
> nulls exist in the base table.

"EXISTS" is standard SQL surely? There are many times when an EXISTS clause would make the logic clearer than a JOIN.

> Also, it should be pointed out that the author uses
> in the standard query (afterwards converted) sub-select
> statements, which are not necessary, and in fact
> probably exacerbate any problems.

Maybe they just used a simple example for pedagogical purposes, but the point is that any treatment of NULLs should be totally watertight and always work as expected, irrespective of however many subqueries you use.

But I was thinking more of a rebuttal of the claim that FirstSQL handles NULLs in a totally consistent and logical way. Can anyone give an example where FirstSQL's treatment of NULLs leads to problems? Maybe such an example doesn't exist? I'm thinking primarily logical problems rather than ease-of-use problems.

Paul. Received on Tue Jun 07 2005 - 05:22:37 CDT