Re: Temporal database - no end date
Date: 19 Jan 2007 16:18:07 -0800
Message-ID: <1169252287.769277.7770_at_l53g2000cwa.googlegroups.com>
Bob Badour wrote:
> Kev wrote:
> > Thanks for taking the time to reply Bob.
> >
> > For everyones benefit I assume you are talking about the book
> >
> > Temporal Data & the Relational Model,
> > First Edition, C.J. Date, Hugh Darwen, Nikos Lorentzos
> > (Morgan Kaufmann, 2002);
> > ISBN 1-55860-855-9
> >
> > I will take a look.
> >
> >
> > Does anyone know if I might find formal answer to my original question
> > in this book, or anywhere even?
> >
> > Thanks all.
> >
> > Bob Badour wrote:
> >
> >
> >>If you are satisfied with SQL kludges, then I suppose Snodgrass is okay.
> >>If you are interested in something more robust, I suggest you google
> >>Lorentzos and check out the Date, Darwen, Lorentzos material.
>
> A number of ways exist to model the problem of open intervals. In the
> Date/Darwen/Lorentzos book, they opt to make all intervals inclusive. To
> model an open interval in that system, one would use the largest
> representable value as the end of the interval, which for practical
> purposes is the same as the other methods.
>
> One might choose to separate the starts and the ends of intervals into
> separate relations. In that model, a start without a subsequent end or
> an end without a prior start would indicate open intervals. However, one
> immediately encounters a number of problems. For example, one will
> generally have to resort to a lot of special cases to make use of the
> open intervals, and one will find it difficult--if not impossible--to
> express many common constraints.