Re: Examples of SQL anomalies?
From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Tue, 08 Jul 2008 10:31:44 -0300
Message-ID: <48736c43$0$4046$9a566e8b_at_news.aliant.net>
>
>
> Not that I would expect a change in the standard for these things, but
> wouldn't it be better to have two different markers for "no data" such
> as:
>
> - NULL for "unknown yet, but should have a value lateron" which
> behaves like 1 + NULL + 1 = NULL
> - N/A for "not applicable" what will be omitted in any expression and
> function such as 1 + N/A + 1 = 2
>
> Normally a database should be designed a way N/A can not occurr (the
> column should not be there), but having this would give the
> possibility to use it where needed and to clear up the inconsistent
> behaviour, I think.
>
> brgds
>
> Philipp Post
Date: Tue, 08 Jul 2008 10:31:44 -0300
Message-ID: <48736c43$0$4046$9a566e8b_at_news.aliant.net>
Philipp Post wrote:
>>Nulls are often but not exclusively used to mean unknown value. Failure to distinguish between different uses of null is the basis of most of these, rather pointless, arguments. <
>
>
> Not that I would expect a change in the standard for these things, but
> wouldn't it be better to have two different markers for "no data" such
> as:
>
> - NULL for "unknown yet, but should have a value lateron" which
> behaves like 1 + NULL + 1 = NULL
> - N/A for "not applicable" what will be omitted in any expression and
> function such as 1 + N/A + 1 = 2
>
> Normally a database should be designed a way N/A can not occurr (the
> column should not be there), but having this would give the
> possibility to use it where needed and to clear up the inconsistent
> behaviour, I think.
>
> brgds
>
> Philipp Post
Codd already suggested as much in the early 80's, and Lee Fesperman's FirstSQL actually implements both NULL markers. Sadly, this does not solve the problem. Important identities like SUM(A) + SUM(B) = SUM(A+B) still break. Received on Tue Jul 08 2008 - 15:31:44 CEST