ON THE HEADER/BODY VIEW OF RELATIONS

From: Joseph Bui

Date: 24 Feb 2006

Regarding following comment in ON POFN* AND POOD* - TWO COMPLEMENTARY DATABASE DESIGN PRINCIPLES:

It is quite possible that the difference between you and Date stems from the difference between his header/body view of a relation, which you (and recently me) do not espouse, and therefore differing definitions of a tuple.

I have not seen anything on dbdebunk about this. I assume that Date's "header/body view of a relation" is the same as given in THE THIRD MANIFESTO. What is the alternative? Is it published anywhere?

From: Fabian Pascal

A mathematical relation/set does not have a header; therefore, do we really want a database relation to have one?

We believe only the data (body of rows) in a R-table to correspond to a relation. What Date/Darwen call the header is part of the metadata (the schema). This way the database framework remains loyal to the mathematics of relational theory.

Not yet. It is part of a formal framework being developed by David McGoveran, but not yet ready for publication. You can see some aspects of it in my revised PRACTICAL DATABASE FOUNDATIONS papers. I am working on an update of paper #2 which will further refine those aspects.

Posted 4/14/06