Synchronisation, concurrent object-oriented programming and the inheritance anomaly

Abstract

There have been a number of models proposed for integrating concurrency and object-oriented programming. Unfortunately, there have been few criteria proposed for evaluating them. In focusing on issues of inheritance, designers of new concurrent object-oriented programming languages appear to have forgotten the experiences learned from over two decades of concurrent programming. In this paper we review Bloom's criteria for evaluating the expressive power of synchronisation primitives, and apply it in an object-oriented framework. We show that most of the available concurrent object-oriented programming languages lack expressive power, and that this contributes to the so-called inheritance anomaly. We then propose a new model which does address all of Bloom's criteria.