Re: relations aren't types?

From: John Jacob <jingleheimerschmitt_at_hotmail.com>
Date: 5 Jan 2004 09:54:28 -0800
Message-ID: <72f08f6c.0401050954.2aec066e_at_posting.google.com>


> From a language grammar perspective, they may have named a production or two
> using "scalar", but I do not see any logical distinction.

Try right around 20 in Tutorial D. An industrial D would have many more. Out of 26 prescriptions in TTM, 8 deal specifically with scalar types, values, variables and operators. That's almost one third of the specification. Scalar is treated at least as much, if not more, than relation and tuple types combined. There is an entire chapter devoted to explaining the difference between scalar types (domains) and relations. The entire type inheritance model is based on the concept of a scalar type. I'm sorry, but this is a vast deal more than a minor difference. It is a cornerstone of TTM and a *huge* logical distinction.

> What can one do
> with a scalar that one cannot do with a non-scalar?

How about model data! Do you honestly claim that integers, floats and the like are not necessary? How do you propose to model scalars without scalar types? Received on Mon Jan 05 2004 - 18:54:28 CET

Original text of this message