On the semantics of rules in database programming languages

Abstract

The tension between the notion of ‘type’ as found in programming languages and ‘class’ as found in semantic and object-oriented database models is examined. A class includes structural and behavioral information as does a type, and also an extent, i.e., set of members, associated with the class by the current database state. The membership of extents can change over time, but they are nevertheless used in a manner analogous to programming language types. As a result, verifying “type correctness” in this realm is centered around integrity constraints rather than traditional typing paradigms. The notion of dytype (‘dynamic type’) is introduced to capture this phenomenon of using mutable sets as types.This paper reports on preliminary research focused on the design and investigation of database programming language (DBPL) features to facilitate programming with dytypes. The incorporation of rules into the update semantics of a DBPL is proposed as an appropriate mechanism for partially enforcing dytype conditions. The preliminary design of a DBPL, RUSE, which can serve as the basis for investigating formal properties of alternative syntax and semantics for rules in DBPLs is introduced. Some representative semantics based on “overwriting” and “accumulation” are discussed. Semantics based on overwriting have characteristics of imperative programming while semantics based on accumulation have characteristics of declarative programming.