Re: Nearest Common Ancestor Report (XDb1's $1000 Challenge)

From: Dawn M. Wolthuis <dwolt_at_tincat-group.com>
Date: Mon, 17 May 2004 15:20:26 -0500
Message-ID: <c8b6qq$ovd$1_at_news.netins.net>


"Topmind" <topmind_at_technologist.com> wrote in message news:4e705869.0405161936.5fd850a5_at_posting.google.com...
> > > Tony wrote: If your analogy holds any water at all (to give you the
> > > benefit of very large doubt), it suggests that relational theory will
do
> > > just fine for pretty much anything we ever want to do "in the real
world".
>
> Relational may not be Turing Complete, but it does not have to be.
> Nor is it meant to be a "total solution". It is a very helpful
> tool that complements application code (and may even help
> organize it).
>
> Relational may not do everything well, just an awful lot well.
> I do agree that it would be nice
> if relational implementations had more
> hierarchical operators, but in practice most classification
> systems are not really trees when they grow beyond the
> non-trivial. Trees have some nasty limitations, yet some
> people keep seeing them as a the Ultimate Structure.
> The real world is not tree-shaped for the most part.
> Philosophers have known this for a hundred+ years.
> Tree-like elements pop up, but there are usually enough
> exceptions to make a pure tree impracticle. It degenerates
> into a graph, and relational many-to-many tables are pretty
> good at that.

Just to get terminology clearer -- a tree is a graph.

I don't have a large background in graph theory, so others can chime in, but some graphs have cycles and some don't -- some have direction, some are strict trees. For every graph there are some corresponding trees that can be useful in navigating the graph. The web is an implementation of a directed graph, for example, where our road system is an implementation of a graph, but not a directed one (or you can think of there as being directions assigned to each street, most of those directions being both ways). Implementations of graphs are all around us. There is nothing "ultimate" about them, but they do provide a reasonably straightforward way to model propositions, for example.

<snip>
--dawn Received on Mon May 17 2004 - 22:20:26 CEST

Original text of this message