Re: deductive databases

"alex goldman" <hello_at_spamm.er> wrote in message news:25978468.7Z4LEkyPSR_at_yahoo.com...
> Nilsson's book [1] talks about using first-order logic (and its subsets)
as
> the language for deductive databases. The book is 10 years old, and yet,
> AFAIK the expressive power of modern state-of-art database software like
> Oracle and PostgreSQL still falls far behind first-order logic: it
> essentially doesn't have functors or recursion. Does anyone know why?
>
> [1] http://www.ida.liu.se/~ulfni/lpp/

What do you mean by "the expressive power of ...database software" ? I assume you meant "the expressive power of data (sub)language".

1. functors They are eliminated in the first step of the "normalization" process. See Codd 1970 paper ( http://www.scism.sbu.ac.uk/~rmkemp/codd1970.pdf ) The semantic in Prolog and deductive databases is based on the Herbrand universe and LFP (least fixpoint). Existence of functors mean infinite domains, which mean quantifiers cannot always be eliminated.
2. recursion There are many places for recursion: - in a language like Java, C/C++, Pascal, Cobol , etc. used in conjunction with the DBMS - in the language of stored procedures (PL/SQL, Java, etc.) - in the data (sub)language, in the definition of a view for example.