Re: In an RDBMS, what does "Data" mean?

"Anthony W. Youngman" <wol_at_thewolery.demon.co.uk> wrote in message news:<qZztNxCLLizAFwnY_at_thewolery.demon.co.uk>...
> Now compare the amount of *metadata* available to Pick and/or
> relational. It doesn't matter what your database is, the data in it is,
> as far as the dbms is concerned, a meaningless "blob". To optimise
> performance, storage, whatever, the only thing available of any use to
> the dbms is *metadata*. Which Pick has in abundance.
>
> That's why I describe Pick as a superset of relational - it can convert
> metadata into data and present it to the app. It can also USE the
> metadata to optimise itself. Relational can only store this sort of
> information as *data*, and as such the information is not available to
> the dbms for its internal use.

False. A relational database contains a lot of metatdata: primary/unique keys, foreign keys, other constraints. All of these are available to the RDBMS for optimisation purposes. To take your invoice example, the RDBMS "knows" that a given invoice has 14 invoice lines just as surely as Pick does. Received on Tue Jun 15 2004 - 11:38:31 CEST