Re: Who conceptualized physical data independence before E.F. Codd, Chris Date and Michael Stonebreaker?

On Jan 26, 12:38 am, Bob Badour <b..._at_badour.net> wrote:
> David BL wrote:
> > On Jan 25, 11:04 am, paul c <anonym..._at_not-for-mail.invalid> wrote:
>
> >>On 24/01/2011 11:57 AM, knorth wrote:
>
> >>>Who conceptualized physical data independence before E.F. #Codd, Chris
> >>>#Date and Michael #Stonebreaker?
>
> >>>http://bit.ly/gibPP9
>
> >>What is this question about? No doubt Codd knew about Childs' work but
> >>what direct comparison can apply? Codd showed a possible connection
> >>between classical set ops and predicate logic. He didn't prescribe any
> >>particular data structure (nor any particular domains for that matter).
> >> The word 'physical' doesn't appear in his 1970 paper as far as I know.
>
> > In Codd's '70 paper he uses the term "data independence", and I
> > interpret it as physical data independence, where he talks about
> > "independence of application programs ... from ... changes in data
> > representation", or "... without superimposing any additional
> > structure for machine representation purposes".
>
> > What do you mean when you say Codd showed a possible connection
> > between classical set ops and predicate logic? That only sounds like
> > the idea of the predicate which is the indicator function of the set,
> > or the set which is the extension of the predicate. The idea of that
> > duality between sets and predicates has been around long before Codd.
> > E.g. the axiom of (restricted) comprehension in set theory is
> > essentially the idea that from a predicate one has an associated
> > extension which is a set.
>
> > In any case Child made it clear in his paper that a database could be
> > regarded as recording sets and relations in particular (to be "pointer
> > free") and set theoretic operators could form the basis for query on a
> > database, and doing so provides a nice way of achieving data
> > independence. I note as well that he defines operators on relations,
> > including a "relative product" which is similar to a join. Also the
> > duality between predicates and sets was implicit in his examples, such
> > as
>
> > M = { <x,y>: y is the mother of x }
>
> > I would say that Codd's insight was his idea to restrict the entire
> > database to just a set of named n-ary relations with *simple* domains
> > and his small number of elegant and sufficient operators on relations.
>
> Codd proved the equivalence between set algebra and predicate calculus
> in his 1972 paper. I am sure you can appreciate that apparent implicit
> duality is not a proof of duality.

Fair enough.

Expanding on that, Codd proved the equivalence between his RA and a particular predicate calculus which he defined in that paper. He named it the Relational Calculus. It is strictly less powerful than FOL. If I'm not mistaken it is the most powerful predicate calculus that is constrained by the requirement that under the intended interpretation tuple variables have a clearly defined, finite range. This is achieved by assuming the extensions of the monadic predicate constants are finite (because it is assumed they are the base relations in the database which must be finite), and there are restrictions on the use of negation and disjunction in WFFs. Received on Tue Jan 25 2011 - 20:38:55 CET