Re: BCNF: superkey or candidate key ?
Date: 27 Sep 2006 08:21:38 -0700
Message-ID: <1159370490.890451.6630_at_k70g2000cwa.googlegroups.com>
David Cressey wrote:
> "Jan Hidders" <hidders_at_gmail.com> wrote in message
> news:1159351205.410007.70750_at_h48g2000cwc.googlegroups.com...
> >
> > Both splitting methods can be generalized. So you see that the price of
> > splitting off a nullable key-columns comes either in the form of
> > exclusion dependencies or embedded functional dependencies.
>
> This is the sort of thing I was looking for. Thanks, Jan.
> It's going to take me a little while to digest it.
You are very welcome.
> In the meantime, I have a question. You say:
>
> > and R2(A, B, D) contains all tuples in R projected on {A,B}
>
> Why isn't that all tuples in R projected on {A,B,D} ??
Uhm, because I wanted to test if you had actually read and understood it? :-)
When writing it I had to get back to my real work quickly. When I started writing the example there first was no D column, but that had some annoying side effects such as FDs with empty left-hand sides. So at the last moment I inserted the D column in the example, but in my haste I may have missed a few spots. I also seem to have forgotten it in the inclusion dependencies (the SUBSET constraints).
- Jan Hidders