Re: WWW/Internet 2009: 2nd CFP until 21 September

From: Mr. Scott <do_not_reply_at_noone.com>
Date: Sat, 15 Aug 2009 20:51:13 -0400
Message-ID: <cPydnVXfac6cxRrXnZ2dnUVZ_gSdnZ2d_at_giganews.com>


"paul c" <toledobythesea_at_oohay.ac> wrote in message news:0UChm.41410$PH1.8663_at_edtnps82...
> Mr. Scott wrote:
>> ...
>> What constitutes an "eligible" wff? ,,,
>
> Obviously the wff's that are representable in one of Codd's relations.
> This is not the set of all wff's.
>
>
> Remember that Codd endorsed the use of
>> null in the relational model. He devoted two complete chapters to them
>> in his book 'The Relational Model for Database Management, Version 2,'
>> contrasting the four-valued logic of RMv2 to the three-valued logic
>> needed in RMv1.
>
>
> Codd had at least two models, the 1970 one is the simplest. He should
> have written some chapters on the logic of expressing two different
> relations in one representation/table. Eg., how this affects projection.
> Walter gave a clear example of an ORDERS table that is an attempt to
> represent two different relations, eg., one that has attributes {OrderId,
> CustomerId} and one that might have attributes {OrderId, CustomerId,
> OrderDate}.

Pardon the pun, but your argument appears to be more applicable to inapplicable nulls. Each placeholder in a predicate requires that there be a value for that placeholder in every appearance of the predicate. So just the fact that there is a placeholder indicates that there should be a value. The presence of an inapplicable null indicates that there shouldn't be a value. Therefore, whenever there is an inapplicable null, there must be more than one predicate, one for when a value is knowable and one for when a value is unknowable. But the same is not true for applicable nulls. The presence of an applicable null is supposed to indicate that there should be a value but it is at present unknown.

The compound predicate, Pxy \/ Pxyz, is fine when inapplicable nulls are permitted for z because whenever an atomic formula is of the form Pab, z is unknowable. When only applicable nulls are permitted for z, an indication that z is unknowable would be wrong, and as a result Pxy \/ Pxyz could not be the predicate. In an atomic formula of the form Pabv, where a and b are constants and v is a variable, it is clear that the value for v is knowable but at present unknown, even though the predicate is the same as that in an atomic formula of the form Pcde, where c, d and e are all constants.

<snip> Received on Sun Aug 16 2009 - 02:51:13 CEST

Original text of this message