On behavioral descriptions in object-oriented modeling

In the past decade, object-oriented programming has come to be considered a panacea for all computing headaches. We view object orientation as a modeling paradigm, first and foremost. We attempt to provide a formal characterization of objects' behaviors as a step toward the formal analysis and validation of object-oriented systems. In particular, we address the following questions: (1) what should be the scope of objects' behavior? (2) what are the desirable functional properties of behaviors? and (3) what implicit and explicit behavioral composition paradigms need to be supported by an object-oriented model? Unfortunately, but expectedly, the answers to these questions rely on some formal proofs, but also on common sense and general software engineering principles. In particular, we argue that all that an object-oriented model needs to support for behavioral specifications are (1) purely functional behaviors, with no side-effects on the objects implementing them; (2) purely side-effectal behaviors, whose effects are restricted to the implementer and its components; and (3) constraints (between objects). An object-oriented model based on these principles has proven valuable for predicting the behavioral properties of object-oriented systems. The model is briefly discussed.