# Re: Limits

Date: Fri, 25 Jul 2008 17:58:00 -0700 (PDT)

Message-ID: <493fb875-219c-4ce3-a52d-7bb6132887f3@e53g2000hsa.googlegroups.com>

On Jul 26, 1:17 am, Tegiri Nenashi <TegiriNena..._at_gmail.com> wrote:

*> On Jul 25, 2:58 pm, JOG <j..._at_cs.nott.ac.uk> wrote:*

*>*

*> > For instance if we have relation:*

*> > R := {w, x, y}*

*>*

*> > where the tuples are:*

*> > w := (a:1, b:1)*

*> > x := (a:2, b:2)*

*> > y := (a:3, b:2)*

*>*

*> > then "ordering by a" yields a total ordering over R:*

*> > {(w, x), (w,y), (x,y)}*

*>*

*> > but "ordering by b" gives the partial ordering:*

*> > {(w, x), (w,y)}*

*>*

*> OK. But who the recepient of the ORDER BY output? I assume it is some*

*> sort of SQL client. Can it understand partial order?*

No, of course not, because clients tend to allocate tuples into an an array, and certainly don't store them as a set even logically. However, this is somewhat of a digression, because I was concerned with the theory - whether such concepts can be integrated into RA, and if doing so offers any utility. Ordering seems important. Received on Fri Jul 25 2008 - 19:58:00 CDT