Re: 1GB Tables as Classes, or Tables as Types, and all that refuted

On Sun, 14 Nov 2004 20:45:12 -0800, Costin Cozianu <c_cozianu_at_hotmail.com> wrote:

>In The Third Manifesto, D&D affirm that 'tis a great blunder to equate
>"classes" with relations or with relation variables.

This is rather evident that it is a blunder to confuse types with variables or values.

> Some people still
>believe that, although the riguorous proof of the drastic consequences
>supposed to follow is completely lacking.

It is hard to prove the practical consequences of a mistake that almost nobody makes.

I don't know any developer that confuses types with variables or values.

>Most of the client programmers, be it in Java, C++, whatever, create a
>User class with respective attributes.

This is probably a little mistake, but in almost all cases it does not have relationship with Date's 1stGB.

Date's 1GB is related to a vanished fad of the 90's called OODBMS. It has very little relevance today and it was never very relevant IMO.

>if that was a blunder, we'd all be suffering terrible consequences, but
>we don't, and even if that wasn't a proof enough in itself, let's
>consider how we should do it:

Most OO database projects suffer terrible consequences due to other blunders, but the developers are not aware of it and they think that the terrible consequences are a part of the job.

>So the pragmatic knowledge that everybody and their grandma programming
>systems these days do have a Users table and do have a User class, and
>nothing bad happens because of this designed is directly confirmed by
>simple logical reasoning.

This is not true. There are many systems that don't have an User class and they have a DataSet or RecordSet instance that represents an Users table.

IMO an User class would be superfluous in an application, but it would not cause big problems necessarily.

Regards Received on Mon Nov 15 2004 - 11:50:18 CET