Re: thinking about UPDATE

"Jan Hidders" <jan.hidders_at_REMOVETHIS.pandora.be> wrote in message news:pan.2004.07.23.07.26.28.218478_at_REMOVETHIS.pandora.be...
> On Fri, 23 Jul 2004 01:43:23 +0000, D Guntermann wrote:
> >
[snip]
>
> CK {a,b,c} corresponds to FD {a,b,c}->{}

I understand this...

> CK {a,b} corresponds to FD {a,b}->{c}

I understand this...
> CK {a} corresponds to FD {a}->{b,c}

I understand this...
> CK {} corresponds to FD {}->{a,b,c}
>
I *don't* understand this....

If you have a set of attributes that are candidates for a key, but you forego any of them for the empty set, then it seems that the empty set is being treated as an attribute/column versus an empty set. As indicated in previous posts, I could only reasonably understand an empty set as the key in cases where the relation was defined with no attributes.

> Note that the last line does not say FD {{}}->{a,b,c} because that
> wouldn't make any sense; FDs hold between sets of attributes and the empty
> set is not an attribute.
>
> Does this compute? :-)

Yes, in some ways; but I have lingering doubts. I will have to take your word for it Professor. ;-)

I don't know if you have your Date *_Intro to database systems_* book handy (7th Ed), but if you do, perhaps you could check out problem 10.7 on page 341 (the answer is on page 344). The question is phrased as, "List all the FDs satisfied by the shipments relvar SP."

To give an indication, part of the answer goes something like this:

{S#} -> {S#}
{S#} -> {}

{P#} -> { P#}
{P#} -> {}

{ QTY } -> { QTY }
{ QTY } -> {} {} -> {}

The last is the only time an empty set is on the LHS and is has one and only one result, the empty set.

What do you think?

>
> -- Jan Hidders

Regards,

• Dan