Re: NULLs: theoretical problems?

From: Jon Heggland <>
Date: Sun, 26 Aug 2007 10:28:25 +0200
Message-ID: <fardjd$b7i$>

Quoth Jan Hidders:
> On 24 aug, 14:42, Jon Heggland <> wrote:

>> But I believe these queries are equivalent:
>> SELECT * FROM t WHERE DEF t.a : t.a = 5
>> SELECT * FROM t WHERE           t.a = 5

> As Bob correctly remarked, the second would raise some kind of
> (static, if that applies) error.

I realise that; I meant the second query to be evaluated in normal SQL. What I was trying to say was that in most cases, the only difference between this DEF scheme and regular SQL is that the DEFs force you to handle NULLs explicitly (which I consider a good thing). You still have the excluded middle problem that V. J. Kumar complains about. JOG's apple/banana anecdote isn't quite apt, unless NULL/undefined should be interpreted as "not applicable" instead of "not known".

But anyway: Can potential NULL-ness always be discovered statically?

Received on Sun Aug 26 2007 - 10:28:25 CEST

Original text of this message