Constraints in RDBMS (was: Re: MySql vs. Oracle)

"Kristian Damm Jensen" <kristian-Damm.Jensen_at_REMOVEcapgemini.dk> wrote in message news:39F42996.6B425A5F_at_REMOVEcapgemini.dk...
> Let *me* be the one to decide that. If that truly is their philosophy,
> all the more reason to critize them. Views, transactions, subqueries and
> triggers are to me all essential parts of a good DBMS; mine to use, if I
> consider it prudent. Of course that requires me to know the price I may
> have to pay when using them - that's part of the deal.

As someone who's used databases both relational and (very) otherwise, I was _AMAZED_ to discover that MySQL purports to be an RDBMS but doesn't have transaction support etc. Why keep a database system and manage such things yourself? That's a return to the dark ages - might as well code it all in C and flatfiles... apart from which, MySQL may manage data in tables but if that were the only criterion then even xBASE would qualify as fully relational.

> (Okay, so triggers are a substitute for explicit declaration of
> constraints and referential integrities, but they are still the best
> tool on the market.)

I disagree. Triggers are pretty much the only tools the _database vendors_ have provided to do this kind of thing; they are not by any means the best or indeed the only tool on the market. I know this because I work for a company which specialises in database-independent business-rules-based systems. The approach taken is based on defining explicit constraints and behavioural rules (in SQL) to be managed by a coordinating rules engine independent of the underlying RDBMS.

But hey - if Oracle Corp (and the rest) can't be bothered to do it right, that's a nice niche for someone else to fill <G>.

Paul

[FX: donning nomex underwear as fast as possible...] Received on Fri Nov 03 2000 - 14:57:00 CET