Re: A question for Mr. Celko

jcelko212_at_earthlink.net (--CELKO--) wrote in message news:<18c7b3c2.0407171721.b0b4e9c_at_posting.google.com>...
> >> Joe Celko said that? Interesting, because I usually get the same
> reaction
> when talking to colleagues of mine that actually do research in the
> field of temporal databases. <<
>
> See? I'm smarter than I look!

I didn't say you were dumb, I asked you to clarify a statement you made and my post went unanswered. I realize you are a busy person, but so am I, and I think this is an *extremely* important issue, one deserving some discussion, especially by someone who is educating the next generation of database professionals, and who many people look to as an expert in the field of databases. For you to claim that the Temporal book is just weird extensions to the relational model without justifying your claim is unprofessional and unfair.

> I could almost live with the use of a "set of time points" in one
> column as a poor way to model a continuum; when Date Packed a
> "continuum of salesman numbers" into one of these things, I just could
> not take it.

The Temporal book is not attempting to model a continuum, and never claimed to. The book proposes the use of intervals (discrete intervals specifically) as a solution to the problem of representing temporal data in a relational database. If a given set of values has a total ordering, such as salesman numbers, why shouldn't I be allowed to create an interval over it?

> If you go to the back of the book and look at the diagrams (which I
> wish had appeared sooner to make the model easier to see), he has many
> redundant ways to represent the same data.

Correct me if I am wrong, but I assume you are referring to the three different designs which are termed Current Relvars only, Historical Relvars only, and Both Current and Historical Relvars. One of the main points of the book is that the first design, current relvars only cannot adequately capture historical data, the second design, historical relvars only, cannot adequately capture current data, and therefore the third design, both current and historical relvars is the only viable solution to the problem. It seems hard to justify calling the three different designs redundant when they are incapable of representing the same information.

> But wasn't one of Date's
> objections to SQL that we have many redundant ways to represent the
> same data in SQL?

Despite this, Date's objections to SQL are not that data can be represented redundantly, this would be an artifact of a bad database design. Date's objections to SQL are based on the fact that there are many different ways to express the same query. There is redundancy in the language itself.

> The final thing was the claim that his extensions were still inside
> the Relational Model. He meant his version, not Codd's, not SQL's and
> I am not even sure about that.

The Temporal book never referred to them as extensions, that's your addition. Indeed, as I pointed out in an earlier post, the book makes the explicit point that it is *not* extending the relational model. Please justify your claim that the book contains extensions to the relational model.

P.S. I note that the book has three authors: Chris Date, Hugh Darwen, and Nikos Lorentzos. It is unfair to the other two authors to refer to the book simply as Date's book. Received on Sun Jul 18 2004 - 09:15:30 CEST