Re: Foreign key problem

From: paul c <>
Date: Thu, 15 Jun 2006 18:41:04 GMT
Message-ID: <4Rhkg.32844$iF6.7247_at_pd7tw2no>

Jay Dee wrote:
> x wrote:

>> "Jay Dee" <> wrote in message
>> news:svbkg.68231$
>>> Hmm.  A _name_ /would/ be necessary to permit your example..
>> I know. But it shouldn't be.
>>> But what is the meaning of the overlapping constraints
>>> with different names?
>> Overlapping as in the example, or overlapping as identical ?

> As in the example. I interpreted the 'foreign_keys(...)'
> stanzas to be some sort of expressions of some sort of super
> constraint in which two of three referring columns had the
> same referents. What are those expressions? I don't know.
> What are super constraints? I don't know.
> Did you have any specific reason for re-specifying those?
>> How about this:
>> t1(a,b)
>> t2(c,d)
>> (a,b) references (c,d)
>> (b,a) references (c,d)

> Funky. But fine by me.
> not(cardinality(t1) > cardinality(t2{c} intersect t2{d})), right?

It all seems like fiddlesticks to me. If two input tables or relations aren't comparable I doubt if there is any logical way to compare them, so to envisage that there is such a way if it is given that they may or may not be comparable seems like a mug's game.

I believe the inputs the OP gave originally were comparable, therefore joinable.

p Received on Thu Jun 15 2006 - 20:41:04 CEST

Original text of this message