Re: Network Example: Sibling of Opposite Gender

Marshall wrote:
> On Dec 30, 10:51 am, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
> >
> > Check out some of the articles at:
> >
> > http://www.thethirdmanifesto.com/
> >
> > In particular, try:http://www.dcs.warwick.ac.uk/~hugh/TTM/Missing-info-without-nulls.pdf
>
> Well, as much as I respect Mr. Darwen, (Dr. Darwen?) I don't find this
> paper
> particularly compelling. While it certainly addresses the semantic
> issues well,
> it doesn't show us much about how we're going to do queries, and it
> doesn't
> particularly justify what appears to be added complexity in updates.
>
> And as near as I can tell, his solution goes to a great deal of trouble
> simply to
> emulated tagged unions in a relational language without them. It would
> be
> ever so much simpler just to say, let our relational language support
> tagged
> unions.
>
> http://en.wikipedia.org/wiki/Tagged_union

With tagged unions, would the solution resemble this?

CREATE TABLE PERS_INFO
(ID INTEGER,
 NAME VARCHAR2(30),
 JOB {VARCHAR2(30), UNKNOWN_JOB, UNEMPLOYED},  SALARY {INTEGER, UNKNOWN_SAL, UNPAID}); If so, how might one use these tags?

SELECT NAME FROM PERS_INFO
WHERE JOB.TYPE = UNEMPLOYED SELECT SUM(SALARY)
FROM PERS_INFO; -- generates compile time error

SELECT SUM(SALARY)
FROM PERS_INFO
WHERE SALARY.TYPE = INTEGER; -- works correctly

Or am I completely missing the boat here?
>
> (Amusingly, I expect that a particularly nice implementation of tagged
> unions
> in a relational language would bear a distinct resemblance to the
> logical
> representation in that paper. However that doesn't mean I want that
> to be the logical interface that I have to work with.)
>
> I would be interested in a discussion about how this all should
> work in regards to aggregate functions. I would particularly
> be interested in a discussion about how all these same issues
> interact with nested relations, and aggregates on same.
>
> It's all very complicated. I tried writing up my current understanding
> of these issues and comparing various solutions to them, but
> it got rather bogged down. I must resurrect that effort at some
> point.
>
>
> Marshall
Received on Wed Jan 03 2007 - 00:25:27 CET