What's a domain? What's a type? The original version of the relational model (defined by Codd in 1969) had domains and not types. The original version of the SQL standard (defined by committee in 1986, but based on work done by Boyce and Chamberlin et al. in the 1970s) had types and not domains. Later, Codd added types and SQL added domains. But no two of these constructs-domains per Codd, types per Codd, domains per SQL, types per SQL-are the same thing! Other researchers, designers, and implementers then got into the act with further definitions of their own, and the picture became, and remains, more confused than ever. This book is an attempt to clear the air. To be specific, it argues that:
- Domains per Codd aren't really domains in the mathematical sense but are much closer to types in the computing sense, and types per Codd don't belong in the model at all. Types per SQL are really several different concepts all mixed together in confusing ways, and domains per SQL are just a red herring. Properly understood, domains and types are the same thing after all.
The book justifies these claims by describing and analyzing the pertinent aspects of both SQL and the relational model in detail.