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

"Alfredo Novoa" <alfredo_at_ncs.es> ...

<snip - various that might be commented later ...>

> >What is the distinction between "User class" and "unidimensional
collection
> >of User objects" that makes the first a blunder and the second not?

> To mix types with values or variables.

This is a circular argument (or less: just a refrasing, definition - not a logical conclusion):

"Why is it a great blunder to equate class with relvar"
"Because it is a great blunder to mix type with value or variable". And
"type = class"

Try again: _Why_ is it a great blunder to "mix types with values"? The answer "that's obvious" is of no (logical) value - that's the hole point in the paper this thread discusses.
As you know, a type is (amongst other things) a set (or a list, if you like) of values. Likewise, a class is (amongst other things) a set (or a list, if you like) of objects.
You allow a set of objects to map (as a non-BG) to a relation. Why and how can an atrribute of a relation have another relation as its domain without thus mixing type with values? (type = domain) Is the clue that "class <> relvar", but "class = relation value"?

If you just are repeating D&D's words from TTM, you add nothing to the discussion. Received on Mon Nov 22 2004 - 15:33:16 CET