Re: relations aren't types?

> > > Scalar is not a useful concept so I fail to understand your question.
> >
> > So you advocate an untyped language?
>
> That's an absurd question.

How so? From the discussion we have had so far, I am under the impression that you make no distinction between the types of different values. How is this not an untyped language? If scalar is not a useful concept then what category of type is Integer? If there is no difference between a scalar and relation type, then how do I declare a scalar type? How do I declare a relation type for that matter? What does your language look like that does not differentiate between scalar and relation types?

> > First, the fact that any scalar can be represented with a finite
> > number of bits (true of every value, scalar or otherwise) does not
> > imply that any finite number of bits is a scalar. Second, and perhaps
> > more importantly, the physical representation of the value in the
> > computer system has no bearing on the logical model.
>
> What then is your useful definition of scalar?
>
> Remember that the logical model is an abstract model one can implement with
> real hardware without regard to the specifics of any actual hardware.

From TTM: A scalar type is a type with no user-visible components. This is by definition a logical construct and so preserves physical data independence. Of course the logical model is an abstract model, that's the whole point.

> > The primary reason for types is to imbue our code with meaning that
> > the compiler can understand and verify. The primary reason for scalar
> > types
>
> You have yet to demonstrate that a useful definition of scalar exists.

Again, what is your overall point? What language are you suggesting we use? You have yet to address this adequately. My useful definition of scalar is based not only on the work of Date and Darwen, but of strongly typed languages in general. At this point, I am forced to conclude that you feel Date and Darwen got it wrong, TTM is fundametally flawed, and that the relational model is unneccessarily complicated by the introduction of different type categories. These are bold claims. Can you back them up? Received on Thu Jan 01 2004 - 22:58:05 CET