Re: boolean datatype ... wtf?

On 1 okt, 18:48, Tony Andrews <tony.andrew..._at_gmail.com> wrote:
> On Oct 1, 5:37 pm, Tegiri Nenashi <tegirinena..._at_gmail.com> wrote:
>
> > Two unary relations? CarsWithImmobiliser and CarsWithoutImmobiliser?
> > It sounds awkward, but think about it would you ever have any trouble
> > when querying such a system (as compared to 3VL with NULL)?
>
> I though that was frowned upon due to POOD?

The strong form of POOD as I understand it ("No two distinct relvars can be allowed to have the same heading") is not tenable. For it implies that there can only exist one single niladic relvar, meaning that the database can contain at most one niladic predicate, sort of meaning that "there can only exist at most one truism".

In (appendix E of) the manifesto book, Darwen suggests that Date knows and understands that and no longer goes by that strong form of POOD. Darwen also suggests that Date has even abandoned the weaker forms of POOD (though I cannot precisely explain what that might mean). In "Database Explorations", Date seems to suggest that the latter is not really the case (that is, he still does believe there is value in POOD). The situation is extremely foggy from where I stand, and it is probably flat out wrong to make any firm statement on the matter. Received on Fri Oct 01 2010 - 20:07:25 CEST