Re: O'Reilly interview with Date

"Paul" <paul_at_test.com> wrote in message news:42ff15c5$0$17493$ed2e19e4_at_ptn-nntp-reader04.plus.net...

> And another similar table for the available sizes.
>
> Now you might want to return a list of all available (colour,size)
> combinations for each product code. Assuming the product code is part of
> the primary key of the two child tables, you could avoid a join to the
> main table by joining the two directly.
>
> Hmm maybe the optimizer would be able to do this internally though.
>

Two tangents from the above discussion.

First the partition of a (floating point) measure into a countable number of disjoint intervals.

Example: clothing sizes. You can take a person's measurements, and determine (within some uncertainty) which clothing size will fit best.

Second, the join between two tables whose common parent exists in the universe of discourse, but not in the database.

Example: Colleges and students tables each have columns called "CITY" and "STATE" in them, as part of the address of the student or college respectively. (Let's disregard the comments of the normalization police.) Now CITY and STATE, taken together are unique in the universe of discourse. There could be a table called "CITIES" with "CITY" and "STATE" as the primery key. But there is no such table in the database, for whatever reason.

You can still to a join between students and colleges based on those two columns. Received on Sun Aug 14 2005 - 12:54:36 CEST