Re: NULLs: theoretical problems?
Date: Wed, 08 Aug 2007 14:20:37 -0000
On 8 aug, 14:26, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
> sinister wrote:
> > Many discussions point out one deficiency of NULLs: that they collapse
> > multiple, distinct concepts into one ("no value possible," "value missing,"
> > "value not available at this time", etc).
> > What are the other theoretical problems? My impression from skimming some
> > threads in this ng is that some anomalies might occur, maybe having to do
> > with NULLs and joins, or NULLs and keys composed of more than one field, but
> > I'm not sure.
> The ultimate theoretical problem is a complete lack of any theory
> underpinning NULL.
Just to avoid any misunderstandings: there has of course been lots of theory on certain interpretations of null values, such as the work by Raymond Reiter and by Joachim Biskup, but not on the specific meaning (if you can call it that) that they were given in SQL. Whether that is necessarily a big problem is IMO not so easy to say.
- Jan Hidders