Re: compound propositions

From: Nilone <>
Date: Wed, 17 Mar 2010 03:21:09 -0700 (PDT)
Message-ID: <>

On Mar 17, 4:01 am, paul c <> wrote:
> I would say tuples are abstractions, also that they are not
> propositions, just partial representations.

That makes no sense to me. If relations in a database are predicates, then the tuples in those relations represent propositions, don't they?

The view which is generally promoted in db design and programming courses is to view tables as containing entities or parts of entities, which is a faulty mapping from the entity-relationship model to the relational model. Is that what you mean?

> >> If those relations were 'exclusive', I think the proposition would be
> >> something like ( 'C1' is a customer and 'C1' is not a client ) OR ( 'C1'
> >> is not a customer and 'C1' is a client ).
> > I'm sorry, I don't follow.  Would you mind rephrasing, please?
> That is a proposition which satisfies an 'exclusive-or' predicate, which
> I think is the assumption you had in mind when you mentioned 'exclusive'
> in the previous message and 'prevent the conjunction' in this one.  By
> giving the 'xor' proposition, I was just trying to point out that it was
> different from the original one I had given and that I hadn't made that
> assumption about disjoint values.

My apologies, it seems I jumped to conclusions.

> I think one reason people interested in db theory spend so much time on
> predicates is because of the practical problem of expressing union (and
> negation), eg., this proposition might be more clear if expressed by
> tuples with four attributes, but the storage to materialize that is
> usually impractical.  So instead most if not all dbms' expect us to use
> what Codd called 'union-compatible' relations for a result that has only
> one attribute.

The reason db theory people spend so much time on predicates is because a relational database is a representer of facts. Predicates and propositions are fundamental to the whole model and theory.

I'm not sure what you mean in the rest of the paragraph, since I don't see a difference between tuples and propositions and I don't see what storage has to do with it.

> I don't object to people talking about entities instead of the actual
> 'objects' of an algebra, after all Codd talked about them.  But I think
> he did that to help his exposition, not to suggest that an algebra or
> FOL could ever operate on entities (at least in his papers before 1974
> or so, not sure about later nor his book).

I agree.

I think the whole DSL movement is an attempt to recreate the algebraic abilities lost in the assumption of the entity-relationship model.

>  I assume my original
> question is at fault for leading to this 'exclusive' assumption because
> it was probably too fuzzy.

It's very likely that I'm somewhat fuzzy too. ;) Received on Wed Mar 17 2010 - 11:21:09 CET

Original text of this message