Re: OO and relation "impedance mismatch"

"Kenneth Downs" <firstinit.lastname_at_lastnameplusfam.net> wrote in message news:1jbvjc.nh5.ln_at_mercury.downsfam.net...
> This is a bummer. I am doomed to endless confusion when talking to Oracle
> people, cuz I'm not willing to stop using the term data dictionary.

This is endemic in the industry, not just with Oracle. When I cut over from DEC Rdb to Oracle in 1994, I had to adjust to the fact that both systems used terms like "Data Segment" and "Snapshot" to refer to completely different features of the database. There were others, but those two come to mind.

In the case you cited, the best bet is to just keep using terms like "system tables", "the system schema", or "metadata" and hope that their use of "data dictionary" in this sense gradually dies away.

Actually, in DEC Rdb, there is a feature of the metadata that it work very much like an actual data dictionary. You can (and should) create a bunch of domains that define all the the data types, formats and ranges that are used through out the database. These domains form the "data elements" of a classic data dictionary. Then, when defining columns, you can (and should) refer each column to a domain. When two columns refer to the same domain, it says something important about values in those two columns, if you do htings right.

You could add comments to the metadata and have a full blown data dictionary in the system tables, but I have rarely seen that done.

In Oracle, on the other hand, if two columns refer to the same Domain, you "just have to know" (unless it's been added in the latest version). Or it has to be documented in the off line data dictionary. Received on Wed Oct 06 2004 - 05:02:27 CEST