Re: relations aren't types?

From: Bob Badour <bbadour_at_golden.net>
Date: Sat, 27 Dec 2003 01:28:38 -0500
Message-ID: <xv2dnfQqmKszu3CiRVn-hw@golden.net>

"Marshall Spight" <mspight_at_dnai.com> wrote in message

news:Ni7Hb.470838$Dw6.1384358_at_attbi_s02...

> "Joe "Nuke Me Xemu" Foster" <joe_at_bftsi0.UUCP> wrote in message
news:1072496648.293318_at_news-1.nethere.net...
> >
> > Interoperability with programs written in static-typed languages
> > would be a concern, as would the problem of declaring constraints.
>
> Hmmm. I'm not sure I see it. Right now, the type system of
> the application language and the type system of the database
> are only marginally related, and one usually goes through
> an API such as ODBC to communicate. That API doesn't
> give you much in the way of static typing.
>
> As far as constraints, isn't saying "column A has type int"
> a constraint?

What would type would the result of a join have? And a union? And an intersection? Would the degree operation mean anything? Yes, attribute data types constrain relvars. The relational model requires certain miminal constraints for all relvars: a relvar name, named/typed attributes and one or more candidate keys.

The points you raise regarding APIs and ODBC only convince me we need better application programming languages. Received on Sat Dec 27 2003 - 00:28:38 CST