Re: Hierarchies in the relational model

Vadim Tropashko <vadimtro_at_yho.cm> wrote in message news:<J4elb.11$5H6.80_at_news.oracle.com>...
> Instead of transitive closure operator can I alternatively suggest a
> distance function? Suppose we have
>
> table graph (
> tail integer,
> head integer
> )
>
> view nodes
> select tail as node from graph
> union
> select head as node from graph
>
> select a.node as head, b.node as tail from nodes a, nodes b
> where distance(a.node, b.node, graph) > 0

The distance operator above takes both data (the nodes) and meta-data (the graph relation) as parameters. Is it OK to mix data and meta-data like this?
Relational theory seems to maintain a strict separation between the two (because of the first-order nature of the logic I guess).  

> Note, that this is purely declarative definition, no inflationary
> semantics, etc.

Could someone give me or point me to a simple description of "inflationary semantics"? I can't understand at the moment how this "distance" operator is superior to the standard transitive closure operator. I've done some googling but I can't find a simple description in the context of relational database theory.

What I have gleaned is that it is also known as "fixpoint logic" and is something to do with intensional/extensional databases.

Is "inflationary semantics" just referring to the physical implementation of the operator i.e. what algorithm it uses?

What also is the motivation for it being called "inflationary semantics"?
Inflationary=growing, semantics=meaning, so does it mean that the "meaning" of the proposition is built up iteratively, maybe converging on the "real" meaning?

thanks,

Paul. Received on Tue Oct 28 2003 - 10:21:36 CET