Re: By The Dawn's Normal Light
Date: Fri, 22 Oct 2004 23:19:05 +0100
Message-ID: <4179875b$0$3262$ed2619ec_at_ptn-nntp-reader02.plus.net>
Laconic2 wrote:
> There are three ways a value might be "non atomic"
>
> The first is that the values might contain some substructure within their
> representation.
>
> In addition to LAST_NAME LIKE 'L%' we could have SUBSTRING(LAST_NAME, 1, 3)
> = 'ACO'
>
> In this case, I take the same view you do, that distinguishing between the
> relational engine and the type engine is useful, at least at the conceptual
> level.
> The second is that the values might themselves be tuples. > The third is that the values might be sets.
I see your second and third possibilities as just special cases of the first one. In fact I'd say all types (except for the bit type) have a substructure that is hidden to the relational engine.
I wonder if the distinction between the two engines would be useful on the DBMS implementation level as well? Maybe some DBMS designers could comment. It might be nice to have a relational engine totally separated from the type engine. Then you could maybe choose from a variety of type engines to fit your relational engine (or vice-versa), assuming they have a well-documented and open API.
Paul. Received on Sat Oct 23 2004 - 00:19:05 CEST