Re: The fable of DEMETRIUS, CONSTRAINTICUS, and AUTOMATICUS

"Marshall Spight" <mspight_at_dnai.com> wrote in message news:VZwed.236726$wV.133_at_attbi_s54...

> This is an intriguing statement, and I would like to ask you to expand
> on it. I expect I am some years earlier in the process than you are;
> I have seen evidence for this, but I'm still thinking in terms of having
> the catalog in the database. (But I suppose this is the "reference copy"
> you mention.)

Like you, I haven't been thinking about this for as long as Kenneth has. And like you, I'm intrigued.

Let me just say this: it is absolutely necessary that the catalog be synchronized with the database itself. It is likewise necessary that the catalog be available whenever the data is available. These two can be accommodated by putting the catalog inside the database. It's also one of the 12 rules that the same query language be usable with the catalog and with the data. This argues for storing the catalog in the form of metadata.

But it's not necessary that the catalog in the database be the "system of reference" with regard to data definitions. There are some very interesting consequences if the database definitions are derived from a subset of the enterprise wide data dictionary (or whatever you want to call it). If you do things this way, and you don't make mistakes, you can guarantee that two database will be able to "talk to each other".

Interestingly enough, you even do this if one of the databases is relational and the other is not, provided there's a bridge between them. Received on Sat Oct 23 2004 - 20:37:08 CEST