Re: relations aren't types?

From: John Jacob <jingleheimerschmitt_at_hotmail.com>
Date: 5 Jan 2004 19:13:30 -0800
Message-ID: <72f08f6c.0401051913.5beb993_at_posting.google.com>


> The important concepts are generic types vs. specific types. The relation
> type generator defines a generic type and instantiates specific types as
> necessary. Likewise, the tuple type generator defines a generic type and
> instantiates specific types as necessary. Similarly for interval types.

So what about scalar types? How do I define a new scalar type? Where is the scalar type generator? Do we not need user-defined scalar types? Even if we don't (a position I *strongly* disagree with), aren't the built-in scalar types a different kind of animal than relation and tuple types? Clearly the different categories of types require different treatment syntactically. This difference is directly exposed to the user through the language, so how is this not an important logical distinction?

If, as you suggest, scalar is not an important type category, then isn't it an unnecessary complication to include it in the data model? So I'll ask you once again, What does a relational language that does not differentiate between scalar types and relation types look like? Received on Tue Jan 06 2004 - 04:13:30 CET

Original text of this message