Re: Network databases

On Sat, 15 Jan 2005 21:42:16 -0600, "Dawn M. Wolthuis" <dwolt_at_tincat-group.comREMOVE> wrote:

>So, the relational model is one where parts of it can become outdated? ;-)

No, it is a model where interpretations of it can become outdated.

>Is it more accurate to say that some of Codd's rules have been shown to be
>unnecessary, incomplete, or incorrect, rather than "outdated"?

Indeed, but I am a lazy typer :)

>> We can find a modern interpretation of the Relational Model
>> in "The Third Manifesto".
>
>And what criteria could we use to show that "The Third Manifesto" does not
>also have parts that have become outdated?

The Third Manifesto is not perfect. The second edition is near to be outdated by the third edition.

> In fact, Date has made
>modifications since that time, particularly related to temporal databases
>and maybe even related to non-1NF (I forget precisely what is in TTM
>compared to the 8th Edition of Intro to Databases).

No, they were extensions.

>If a database management system stores & retrieves data in a way that is
>logically based on mathematical relations (as defined and largely agreed
>upon by the mathematical community rather than one of the many odd
>definitions given by some one person working with databases) and includes
>operations on those relations, then it seems to me that it may claim to be
>an RDBMS.

But SQL is not based on mathematical relations. It is based on pseudo-multisets.

Try to find the words "relational" and "relation" in the SQL specs.

>Of course, soon enough companies will be clamouring to call their DB's
>something other than (just) relational, I suspect.

They have been doing that for years.

Regards Received on Sun Jan 16 2005 - 13:07:06 CET