Issues in the design of object-oriented database programming languages

Abstract

We see a trend toward extending object-oriented languages in the direction of databases, and, at the same time, toward extending database systems with object-oriented ideas. On the surface, these two activities seem to be moving in a consistent direction. However, at a deeper level, we see difficulties that may inhibit their ending up at the same point. We feel that many of these difficufties are a result of the underlying assumptions that are inherent in the fields of programming language and database systems research. Many of these assumptions are historical and contribute to a set of cultural biases that often prevent the two communities from interacting as effectively as possible.

An Overview of the Object-Oriented Database Programming Language DBPQL

“…Fundamental issues that are dealt with differently in OOPLs and in DB languages have been discussed in various papers (Bloom and Zdonik, 1987;Atkinson et al, 1989;Kim, 1993).…”

Reusable distributed “data environments”

“…The objective is the definition of a single data model for both persistent and non persistent data [10,26]. The main problem in achieving this integration is represented by the difference between the declarative description of data assumed in database management systems and the behavioral, encapsulated format of data used in programming languages [16].…”

A data model and query language for EXODUS

“…This makes it possible for a database to include more than one col lection of instances of a given type, which can be quite useful in scientific and engineering applications [Lohm83,Kemp87]. While this separation is common in programming languages, it is less common in the database world [Bloo87], As an example, the com mands in Figure 1 define a new schema type called Person, which is a tuple type. Two sets for storing Person instances are then created, the Students set and the Employees set.…”

“…© 1988 ACM 0-89791-268-3/88/0006/0413 S1.50 models that would have been called "semantic data models" two years ago. In fact, the former kind of object-oriented DBMS almost seems like a step back to the days of navigational data manipulation languages, as it is not obvious how one will support ad-hoc queries (or optimize accesses effectively) for such systems [Bloo87,Ullm87).…”