Re: Plural or singular table names

"Alfredo Novoa" <alfredo_at_ncs.es> wrote in message news:e4330f45.0309120835.6c08b19b_at_posting.google.com...
> "Bob Badour" <bbadour_at_golden.net> wrote in message
news:<cPQ6b.668$VW5.62768164_at_mantis.golden.net>...
>
> Sorry, I skipped this.
>
> > > Agreed, and in TTM keys are not part of the relational types (at least
> > > in the examples), they are shorthands for database constraints which
> > > are part of the database type.
> >
> > How, then, do the relational operations on values infer constraints?
>
> Relational operations on vales does not infer constaints. Constraints
> are checked in assignment operations (database updates).

Thus you are saying that a constraint affects the declared type of a variable and that constraint inference only applies to operations on variables by which the dbms derives additional variables. Seems fair enough--I'll think about it a little longer.

> > > Tutorial D's relation type generator doesn't allow to declare
> > > candidate key constraints.
> >
> > That's just a matter of shifting a couple productions around in the
grammar.
>
> Of course, but the author's intention seems clear.

I agree the authors' intentions are clear; I am just not convinced I agree with them.

> > > I find the idea of values having constrains as an absurd. Values are
> > > constants. We can not constraint a constant.
> >
> > People do it all the time.
> >
> > const float pi = 3.14159;
> > const unsigned int three = 3;
>
> What I meant is that pi and 3 are different values.

And each has a unique MST.

> If you have a variable holding the value pi, and you change it by the
> value 3 then pi is still pi and 3 is still 3.

The declarations above do not declare variables, and I maintain that the value 3 has a unique most specific type.

> > Even in TTM literals are typed, and types are constraints.
>
> Here you are playing with ambiguous terms.
>
> constants have types
> types are constraints
> ---------------------
> constants have constraints
>
> Both premises, the derivation rule and the conclusion are sloppy.
>
> Using a better terminology (but not a lot better), the conclussion is
> different.
>
> a constant belongs to N types
> types are sets of constants (among other things)
> type constraints define which constants belongs to a type
> -----------------------------------------------------------
> the membership of a constant to a type is defined by the type's
> constraints
>
> It seems a little more reasonible to me.

Okay, I'll think about things a little longer. Received on Fri Sep 12 2003 - 22:32:53 CEST