Temporal operations

"Paul G. Brown" <paul_geoffrey_brown_at_yahoo.com> wrote in message news:57da7b56.0309021140.e506e98_at_posting.google.com... [snip]
> And note that, in order to support temporal operations of the kind used
> in Q3 elegantly, DD&L[2] find themselves obliged to add an entirely new set
> of relational operators to the data model (T_UNION, T_JOIN etc).

You mean U_UNION, U_JOIN, etc and these are not (as far as I can see) 'an entirely new set of operations', not at the logical level anyway. E.g. U_JOIN is short hand for two UNPACKs, a JOIN then a PACK.

At the implementaoin level these would be new operators. Well you would want to reconise the UNPACKs, JOIN, PACK sequence anyhow and optimise them into some physical operations.

> (BTW: from a theory perspective this is an admirable work and the book
> ought to be read.) But I would only observe that the widely quoted D&D
> diktat "The relational model needs no extension, no correction, no
> subsumption -- and above all, no perversion! -"[1], needs to be qualified
> with "except by us!".

I guess it depends on what you consider to be an extension of the model. I can see how their statement would rather too strong for some tastes. I for one certainly don't think the model is perfect and want to improve/expand/clarify/deepen the model.

D&D discuss the "U_" operators under the chapter heading "Generalising the Relational Operators". So maybe they allow generalisation but not extension in their 'religion'.

> And this feels less like science to me and more
> like religon.
>

Nice post otherwise though.

Regards
Paul Vernon
Business Intelligence, IBM Global Services Received on Fri Sep 05 2003 - 11:52:02 CEST