Re: The Practical Benefits of the Relational Model

Bernard Peek <bap_at_shrdlu.com> wrote in message news:<HB+Q7eNbHjm9Ewtt_at_diamond9.demon.co.uk>...

> >Ho ho. Rather misses the point that the Relational Model had a theoretical
> >background from day 1. Trying to formalise a whole mess of things after
> >the fact is non trivial and won't produce you a nice clean model, if
> >infact you can find a single model at all.

> I'm sure it's non-trivial, if it was easy it would already have been
> done. And you're right that there are a lot of different things labelled
> as object oriented databases. The relational model codified and
> rationalised a lot of what already existed in database systems, and we
> have built RDBMS on that theory.
>
> It's certainly possible that someone will develop a similar theory based
> on what we now call object oriented systems. It won't need to cover
> everything we now know as object oriented systems. It will just need to
> provide enough of a theoretical framework for the development of future
> systems.

The issues go well beyond whether or not various approaches have formal descriptions. When applied to data management, the OO paradigm looks very similar to a breed of DBMSes termed hierarchical. The hierarchical/network vs. relational debate has raged since the RM was first introduced. The debate continues to this day. In the end, the model debate boils down to which model is the simplest, complete one. At "The Great Debate" SIGMOD in 1975, Codd convincingly argued that the RM was such a model. Arguments against the RM then shifted to "well, nobody could ever build a fast relational system." This was obviously disproved by early SQL system which were able to do all kinds of acrobatics despite not even truly getting the RM right. But having the best "model" doesn't help anybody. In the end, a model is only as good as it's best implementation, which is why so many relational "purists" have been shouting for so many years that current relational attempts were forgoing a plethora of benefits by deviating from the model. To make things worse, many people have equated these problem areas as being problems in the model, therefore sending them off in search of a better model. It's a bit like a person using a poor calculator, heading off in search of a better arithmetic!

Regards,

--
Nathan Allan