Re: A Simple Notation
Date: Tue, 10 Jul 2007 15:13:31 -0000
Message-ID: <1184080411.566994.228070_at_w3g2000hsg.googlegroups.com>
On 10 jul, 16:38, "David Cressey" <cresse..._at_verizon.net> wrote:
> "Brian Selzer" <b..._at_selzer-software.com> wrote in message
>
> news:hGLki.45816$5j1.18058_at_newssvr21.news.prodigy.net...
>
>
>
> > By the definition of a natural join, there can only be one relationship.
> In
> > order for a tuple to appear in the result, there must be value agreement
> for
> > *all* common attributes in the tuples from the operands.
>
> Agreed. But the fact that common attributes are discovered via common names
> is surely a setback (IMO).
Indeed, it is well known and provable that the class of queries expressible with {select, project, join, union, difference} is a proper subset of those expressible in the algebra that also includes the renaming operation, even if your query is a yes/no query, i.e. returns a relation with no columns.
- Jan Hidders