Re: [xml-dev] Caught napping!

["Gary Stephenson" <garys@ihug.com.au>]

Hi Tom,

> Wow... Glad I'm not in his list of Bozos.  For as long as he's been
> ranting about this stuff, I still haven't been able to figure out what
> his point is, and what kind of bug crawled up his butt.
>
> Any ideas?

I'm sure many others on this list have a much better grip on all this than I,
but, leading with my chin as always.... :-)

I believe FP's point is basically the same as Date and Darwen's in their
"Third Manifesto" work(s).  That is, the relational _model_ is _theoretically
sound , and more than capable of supporting all the use cases where the OODB
proponents, and now also the XML-DB proponents, have been claiming that RDBMS
systems won't cut it.

Unfortunately, the relational _model_ has not yet been properly implemented
(for reasons that are not entirely clear).  Date and Darwen (and FP) propose
that rather than throwing the baby out with the bathwater in rejecting RDBMS's
in favour of some other model, it would be far better to make your RDBMS more
faithfully adhere to the underlying (sound) relational theory - whereby you
will (supposedly) get "the best of both worlds", or perhaps "the best of all
worlds".

Personally, I think the point that they have failed to properly address is -
Why hasn't the relational model ever been properly implemented?   I"ve heard
it said that of Codd's original twelve rules, nobody ever managed to implement
more than 5 or 6 concurrently!  Assuming this to be true,  why?  If the theory
is not implementable, then it is essentially useless - yes?  Or worse, it is
positively counter-productive!  FP has been making noises to the effect that
he will soon be making available a _real_ RDBMS, but unless and until he, or
anybody else for that matter, does so, methinks the argument to be somewhat
hollow.

gary