"Eric Kaun" <ekaun_at_yahoo.com> wrote in message
news:xuO%b.18054$P_4.12318_at_newssvr31.news.prodigy.com...
> "Neo" <neo55592_at_hotmail.com> wrote in message
> news:4b45d3ad.0402261307.1315965d_at_posting.google.com...
> > > Also now the domain/type EYE_COLOR has tuples "Not Applicable" or
> > > "Unknown". Doesn't this violate basic set theory?
> >
> > As Marshall pointed out, Color = {Red, green, blue, "Not Applicable",
> > "Unknown"} is a valid set. I stated incorrectly, however the effect
> > remains similar: weakens realtionship between set and its elements and
> > hinders closure.
>
> Closure doesn't have to do with the set, of course, but with the type -
> you're talking about operations over relations. I don't think the set
above
> has a problem with closure, though. Joins, projections, etc. which
reference
> an EYE_COLOR attribute can do so meaningfully and without the ambiguities
of
> nulls. Given proper type definition facilities, you can do whatever you
like
> with the NOT_APPLICABLE value.
One can define any number of operations on any given type. That subset of
those operations defined solely in terms of the type define a closed
algebra.
Received on Fri Feb 27 2004 - 15:41:26 CST
