Re: header part of the value?

From: Jan Hidders <hidders_at_gmail.com>
Date: Thu, 28 Feb 2008 14:16:29 -0800 (PST)
Message-ID: <a9868cc7-1107-44cc-b733-8ee7cd3ca75b_at_41g2000hsc.googlegroups.com>


On 28 feb, 22:24, Tegiri Nenashi <TegiriNena..._at_gmail.com> wrote:
> On Feb 28, 1:02 pm, Jan Hidders <hidd..._at_gmail.com> wrote:
>
> > Column names are just there to identify the role that the objects in
> > the column play in the predicate, it is the name of the relation that
> > should indicate the predicate. Why would we want to change that?
>
> Well, it is debatable whether the chatty part of the relation
>
> "The person name is ... and the person salary is greater than ..."
>
> belongs to the relation name, or more naturally can be split into the
> attribute names. IMO the first alternative can't really exploit
> anything out of this additional information in the header.

I doubt that the second alternative can. Choosing different labels doesn't change the meaning of the tuples. Unless you are going to assume these names have structure that carries meaning, but that would depart from the standard relational model, and then all bets are off anyway. Until you are going to show an example that actually shows how this is useful, I'm going to remain very skeptical.

  • Jan Hidders
Received on Thu Feb 28 2008 - 23:16:29 CET

Original text of this message