Re: Codd and many-valued logics

Op vrijdag 3 juni 2016 09:51:14 UTC+2 schreef Nicola:
> On 2016-06-01 02:12:15 +0000, -CELKO- said:
>
> > I also do not remember Codd referencing prior multivalue logics.
> > However, D David McGovern spend some time on this topic. It will be
> > worth googling it. I also posted things on this decades ago. David's
> > work is much much better than mine. He points out of a full logic has
> > to have conference rules.
>
> Thanks, I have found his Nothing from Nothing series of articles.
>
> > In SQL we have search conditions and resolution rules for the searches;
> > not the same thing exclamation
>
> Sure. From a theoretical point of view, though, one may legitimately
> wonder about the relationship between SQL with nulls and 3VLs in the
> same way as we know the relationship between SQL without nulls and
> classical logic, at least for a fragment of SQL. Both McGoveran and
> Date (well, mostly the former, since the latter is just repeating
> McGoveran's arguments) are too dismissive in my opinion, e.g., they do
> not even consider first-order 3VLs. There are at least a couple of
> papers formalizing the semantics of SQL with nulls by extensions of
> relational calculi or algebras, so the question of the expressive power
> of such extensions compared to the 3VLs in the textbooks is well
> defined (the problem of giving an *intuitive* interpretation of such
> logics is another matter).
>
> Coming back to my initial question, I am wondering whether Codd
> deliberately avoided citing works about many-valued logics for fear
> that his proposal would be considered too abstract for serious
> consideration by implementors.
>
> Nicola
>

There are [at least] two questions that a DBMS must be able to answer for itself and which are undeniably 2VL questions : (1) am I going to include this tuple in the result set or not (2) am I going to accept this update or not (is any constraint violated, or not).

If a "mapping" is going on under the covers, such as is the case in SQL, then understanding results correctly becomes harder to do for the users. Received on Mon Jul 25 2016 - 12:07:36 CEST