Re: Dreaming About Redesigning SQL

"Anthony W. Youngman" <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
 
> I notice you don't include me ... why not?

Because it was not my intention to make an exhaustive list. 

> Oh - and by the way - why are IBM (with DB2), MS (with SQL-Server) and
> Oracle all busily pinching our (Dawn and my) favourite database's ideas?

If you mean attributes with collection types there is not anything
wrong in that.

The modern interpretation of The Relational Model prescribes relation
and tuple typed attributes (without having any problem with 1NF).

> But any body who believes it is a good model just
> because it's mathematically perfect quite obviously doesn't live in the
> real world

It is not the reason. There are many reasons. You can learn about it
reading the references sugested thousands of times in the group.

> ... Occams razor is an incredibly simple proof as to why it
> MUST be imperfect - stuffing a four- (or ten-, or eleven-) dimensional
> world into a two-dimensional model can't be right :-)

If you had readen the references you would know that relations have n
dimensions where n is the number of attributes. The Relational model
is n dimensional.

For instance this is a 3D relation which could represent the
coordinates of a cube in the 3D space.

relation { 
  tuple { x 0, y 0, z 0 },
  tuple { x 0, y 0, z 1 },
  tuple { x 0, y 1, z 0 },
  tuple { x 0, y 1, z 1 },
  tuple { x 1, y 0, z 0 },
  tuple { x 1, y 0, z 1 },
  tuple { x 1, y 1, z 0 },
  tuple { x 1, y 1, z 1 }
};

You might represent a four (or ten, or eleven) dimension hypercube in
the same way.


Regards
  Alfredo