Re: Advanced SQL

On 7 sep, 16:19, "David Cressey" <cresse..._at_verizon.net> wrote:
> "Jan Hidders" <hidd..._at_gmail.com> wrote in message
>
> news:1189173585.627344.49900_at_19g2000hsx.googlegroups.com...
>
>
>
> > On 7 sep, 15:43, Jon Heggland <jon.heggl..._at_idi.ntnu.no> wrote:
> > > Quoth Evan Keel:
>
> > > > "David Portas" <REMOVE_BEFORE_REPLYING_dpor..._at_acm.org> wrote in
> message
> > > >news:pZSdnfih_oRy5H3bRVnyigA_at_giganews.com...
> > > >> Call me picky but I am not entirely comfortable with that U.Texas
> site,
> > > >> even though it's a commendable effort overall.
>
> > > >> "The definition of second normal form states that only tables with
> > > >> composite primary keys can be in 1NF but not in 2NF"
>
> > > > They have it right here. All non-key attributes must be dependent on
> the
> > > > full key. Only applies to tables with keys composed of multiple
> columns>
>
> > > No. There may be dependencies where the left side is empty.
>
> > In that case the declared candidate key is not really a candidate
> > key. Under the assumption that the declared candidate keys are indeed
> > candidate keys, the claim is correct.
>
> I'm really confused by the above. First off, if there were a dependency
> where the left side is empty, wouldn't the same value have to exist the
> dependent column, for all of the rows?

Yes. But as Brian correctly pointed out, what I said in the above was wrong. In fact my remark about binary relations in BCNF is also not correct. For example R(A,B) with FD {}-->B is not in BCNF. Not my day, apparently. :-(

• Jan Hidders