Re: A research effort on a computing model...
Date: Fri, 8 Feb 2008 01:05:59 -0800 (PST)
Message-ID: <6333f0b5-936e-49b1-96eb-ffb386362626_at_s12g2000prg.googlegroups.com>
On Feb 8, 1:00 am, "Brian Selzer" <br..._at_selzer-software.com> wrote:
> Trivializing functional dependencies? Eliminating inclusion dependencies?
To give you an idea:
> "Cimode" <cim..._at_hotmail.com> wrote in message
> Sounds to me like you'll have a big job selling that.
Thanks for the comments.
Quite frankly, I do not recall mentionning anything about dependencies
which are logical concepts that can be supported otherwise than by
using keys. Instead of keys, a concept prior to RM, I prefered the
concept of unique identifier, type and un-ary type. Just keep in mind
that this is not a SQL system. Besides, I really am not trying to
sell anything (the db core is designed to be open source) but rather
as a way to explore new pathes and open up the ground for new ideas
and perspectives to emerge .
if you consider...
PART: NUMBER(integer), NAME(string) and CAR: MODEL(string), PART(part) , PART in CAR has type PART , therefore *foreign key* is implicit though use of user defined type. I mentionned that a relation *de facto* constitutes a type with limited number of values that some other relation can draw from. I somehow believe that the concept was to use by Codd as a marketing effort to convince his IBM audience
Hope this clarifies... Received on Fri Feb 08 2008 - 10:05:59 CET