Re: NATURAL JOIN without join columbs
From: paul c <toledobythesea_at_oohay.ac>
Date: Thu, 01 Sep 2005 23:17:32 GMT
Message-ID: <g_LRe.349013$5V4.156421_at_pd7tw3no>
> Since the natural join can be defined as a restricted catesian product,
> where the restriction is the equality condition on the join attributes
> and where the second set of join attributes is "projected away" after
> restriction, you obviously have a cartesian product when there are no
> common columns. Besides, "errors" will never be a result of a relational
> operation, because every relational operator returns again a relation.
> Greetings,
> Andreas
Date: Thu, 01 Sep 2005 23:17:32 GMT
Message-ID: <g_LRe.349013$5V4.156421_at_pd7tw3no>
Andreas Walter wrote:
> Heikki Simperi schrieb:
>
>> What should happend in theory when two tables are joined by natural >> and they do not have columns with same name? >> >> >> A (a,b,c) >> B (d,e) >> >> A |><| B >> >> >> Cartesian or empty or error? >> >> >> >> Heikki Simperi >>
> Since the natural join can be defined as a restricted catesian product,
> where the restriction is the equality condition on the join attributes
> and where the second set of join attributes is "projected away" after
> restriction, you obviously have a cartesian product when there are no
> common columns. Besides, "errors" will never be a result of a relational
> operation, because every relational operator returns again a relation.
> Greetings,
> Andreas
while there may not be any common 'columns', some people say that there is always at least one common attribute, namely the empty set. this agrees with Andreas' answer with a slight twist on the reason.
p
p Received on Fri Sep 02 2005 - 01:17:32 CEST