Re: So let me get this right: (Was: NFNF vs 1NF ...)
Date: 14 Feb 2005 15:29:29 -0800
Message-ID: <1108423769.730356.319030_at_g14g2000cwa.googlegroups.com>
Jan Hidders was widely reported to have said:
> "Related to the previous is the question whether domain values and
their
> associated operations should be allowed to be of arbitrary
complexity.
> From the beginning it was tacitly assumed that domain operations
would
> be simple and efficient and would not have to be optimzed. But if you
> allow them to get arbitrary complex then you might need something of
the
> complexity of a relational engine, complete with query optimization
et
> cetera, again at that level. Would that be worth it? Never mind what
> happens if allow nesting and unnesting and start mixing the
operations
> of the two levels. Does that really weigh against the benefits? Would
we
> really gain something over an approach where at the external level
> things can be arbitarily complex but at the conceptual level they are
> all mapped to relations with simple basic domains? "
Allow me to rant an' spit an drool and dribble on your shoes, for a minute, while my latest workload test runs.
The question before us is (ahem) related, in general, to the power of the relational model, and to its application as a general purpose programming system. On the one hand there are those--and I number Date & Darwen among this crowd, on account of their unaccountable preference for domains that have "equality"--who appear to believe that "domain operations would be simple and efficient". Based on admitedly scant evidence, I would further suggest that such folk are entirely happy to relegate the relational model to the basement of the information system stack while simultaneously insisting--quite rightly--that in architecture, foundations are everything, blah blah blah.
Standing in bewildered opposition to both schools of thought is the JaVa is KewL! tribe, usually found fueding furiously with the C-SHarp RulZ! tribe, both of whom are serviced by a small caravan of badly coifed cretins riding camels with bumper stickers that proclaim "XML is King!".
Suffice it to say I don't want to fix SQL, or replace it with Tutorial D, and I am completely underwhelmed by efforts to build a better DBMS, partly because I think such endeavours are doomed to dash on the rocks of commercial indifference, but most because the whole "DBMS as the persistence layer" idea is really, really dumb. But I also don't want to go beyond what can be safely said: Domains, Relations, Operators, and Constraints, all *programming* as values in relations. No 'domain operators'. No 'RVA's. No 'relvars' (or rather, not distinction between the relvar and the relation, if you like.) Set logics all the way, baby!
> SCREEN := (RETRIEVE U.X, U.Y, U.RGB FROM Images I, UncompressJPG U
WHERE I.Name = 'Fred' and U.Image = I.Image );
Crazy, you say? Hard to believe! Well, in keeping with the grand history of M. Fermat (and the developers of the TransRelational System), all I can say is that I have completely thought the issue through and I have a wonderful system, but this narrow margin, this paltry page in space and time, is inadequate for its exposition. (Plus, I'm bullshitting: I have no clue how to make this vision work.)
And now I see that my latest workload has completed, and I see that what I tried didn't make a damn bit of difference .... Received on Tue Feb 15 2005 - 00:29:29 CET