# Re: Does Codd's view of a relational database differ from that ofDate&Darwin?[M.Gittens]

Date: 11 Jul 2005 11:17:44 -0700

Message-ID: <1121105864.320633.257540_at_g44g2000cwa.googlegroups.com>

Jon Heggland wrote:

> In article <1120752683.210330.79280_at_g44g2000cwa.googlegroups.com>,

*> mikharakiri_nospaum_at_yahoo.com says...
**> > User defined types
**>
**> Whether they are user defined or not does not really matter.
**>
**> > present rather archaic approach to units. Modern
**> > approach is symbolic manipulation
**> >
**> > http://tinyurl.com/68rbm
**>
**> I'm not able to make sense of this. What are the domains of your
**> relations? What does null mean? Is "<distance = 20 meters, time = 3
**> hours>" really a predicate? How does the system solve your 34477.985 =
**> x*lb equation? In what way is this simpler or better than possreps?
*

Null means atomic expression. Atomic is a relative concept, of course. At some point we have enough knowledge to break it into components. Nulls aren't necessary, of course. There could be two tables: Atomic and Composite units.

Why I believe this [admittedly half-baked] system is superior to type system? It is more formal. It is conceptually more simple. There is nothing other than algebraic expression manipulation.

More important, however, what is all this referring to mathematical examples (Circle-Ellipse, Polar-Cartesian) in order to make some retarded OOP point? The correspondence between cartesian and polar is *interesting* from mathematical perspective. Ditto Circle-Ellipse. It is treated as implementation detail from OOP perspective. If the type details are hidden behing some (procedural) code, it leaves no possibilities for expression rewriting, optimization, etc. Received on Mon Jul 11 2005 - 20:17:44 CEST