Re: Oids

From: vc <boston103_at_hotmail.com>
Date: 17 Aug 2005 13:49:36 -0700
Message-ID: <1124311776.273878.155420_at_g14g2000cwa.googlegroups.com>


Vadim Tropashko wrote:
> vc wrote:
> > Vadim Tropashko wrote:
> > > Mikito Harakiri wrote:
> > > > vc wrote:
> > > > > To sum up, the three '1' '0' combinations are :
> > > > >
> > > > > 00 union A = 01
> > > >
> > > > No rowids, then. Sigh.
> > >
> > > To wake up Marshall: rowids are object identifiers.
> >
> > What rowids do you keep talking about ?
> >
> > If I understand your "generalized union" definition correctly, '00
> > union A' will return at best a relation with one empty tuple. Where's
> > the rowid ?

>

> According to your corrected definition
>

> rowids(A) = Ox = 00 union A
>

> Well, in your model rowids(A) is either 00 or 01, and your model might
> even turn out to be consistent. Note, however, nothing prevents us from
> defining
>

> rowids(A) = Ox = 00 union A
>

> where x is the set corresponding to the natural number - cardinality of
> A

According to your own definition, the generalized union is (in terms of the standard RA):

'gu' = union(project( a&b, A), project( a&b, B),

where A and B are relations, a and b are their respective sets of attributes, & is the set intersection Applying this definition to

'00' 'gu' A, we obtain either 00 or 01. Received on Wed Aug 17 2005 - 22:49:36 CEST

Original text of this message