Re: Relation Schemata vs. Relation Variables

From: Jan Hidders <hidders_at_gmail.com>
Date: 7 Sep 2006 02:13:14 -0700
Message-ID: <1157620394.798164.8890_at_m79g2000cwm.googlegroups.com>


vc wrote:
> Jan Hidders wrote:
> > vc wrote:
> > > Jan Hidders wrote:
> > > [...]
> > > >. Note btw. that they are a strict
> > > > subclass of the restrictions that might be expressed by some kind of
> > > > temporal logic.
> > >
> > > What temporal logic do you have in mind ?
> >
> > First order logic extended with the 'since', 'until', 'next' and
> > 'previous' operators.
> >
> > > It's known that first-order
> > > temporal logic is strictly less expressive than FOL for a relational
> > > database augmented in a natural way with a timestamp column, [...]
> >
> > Indeed, even if you extend it with temporal operators for all regular
> > expressions, but that is not what I was comparing it to.

>

> So, if we are in agreement as to the FOTL expressive power, what
> exactly did you mean when you said that "they" (presumably constraints
> expressed as FOL predicates) are a " subclass of the restrictions that
> might be expressed by some kind of temporal logic" ?

Apologies for being a bit terse and grumpy. I was thinking of the approach where the transition constraints are expressed by a FOL formula over a duplicated signature (once for the "old" database, and once for the "new" database). So if your database schema / signature is R_1, .., R_n then the formula is over R_1, ..., R_n, R'_1, ..., R'_n where R_1, .., R_n contain the old state and R'_1, ..., R'_n contain the new state.

> An example would be nice, too.

You can probably think of them yourself now, but something like "once an employee is fired he or she is never hired again" or "the wage of an employee never decreases, even if he or she was not employed by us during some period".

  • Jan Hidders
Received on Thu Sep 07 2006 - 11:13:14 CEST

Original text of this message