Re: Temporal database - no end date

Bob Badour wrote:
> Apparently the DBMS_Dumber guy cannot comprehend the difference between
> a duration and an interval. An interval may not have 0.8 of a time
> quantum, but an average duration certainly can because it is an entirely
> different beast. Just as total fertility is an entirely different beast
> from a child.

  Have you read the quote I provided, Bob? Have you read the book at all?

  I am not the one who can't see the distinction between intervals and durations. I'm sure DD&L are also quite competent in this regard. My point it that in the DDL book, the author's emphasis on the fundamental nature of the 'time quanta' make it awfully hard to reason about this difference. I provided the quote where they _explicitly_ state their assumption, and an example of the kind of thing their model rules out. If you like I can also enumerate the reasons a quantum model simplifies matters, specifically the author's derivation of the PACK() and UNPACK() operators.

  But no one, unless I've missed a comment in this thread, is discussing a model where 'children' is a unit. And if they did, and made a claim such as 'The unit of the child is indivisible.', then we'd need to reject that model pretty darn fast. Besides - please stick to the propositions under discussion. Your fantasies regarding how the world ought to work aren't germaine.

  For a temporal data model to be useful, it must be useful not just for storing facts, but also for reasoning about them. The DDL book's emphasis on the paired interval is imo reasonable, and a contrast to the TSQL model. But by adopting the 'quanta' idea they complicate matters considerably. Received on Sun Jan 21 2007 - 22:12:56 CET