EVOLVE: Adaptive Specification Techniques for Object-Oriented Software Evolution

The increased complexity of object-oriented models necessitates the enhancement of adaptiveness and robustness of an object-oriented design towards changing requirements. The understanding of what properties are critical for construction of an adaptive schema design becomes increasingly important in software evolution. In this paper we present two groups of techniques for enhancing the adaptiveness and the robustness of an object-oriented design in anticipation of future requirement changes. The rst group of techniques consists of a selection of adaptive schema style rules for achieving validity, minimality, extensibility and normality of a schema design. We encourage to use this set of rules as a means for validating quality of a schema, and for transforming an object-oriented schema into a better style, in terms of adaptiveness and robustness of a schema design, rather than as a user-oriented method solely for designing the schema. The second group of techniques includes the use of propagation patterns and propagation pattern reenement. Propagation patterns are employed as an interesting speciication technique for modeling the behavioral requirements. They encourage the reuse of operational speciications against the structural modiication of an object-oriented schema. Propagation pattern reenement is suited for the speciication of reusable operational modules. It promotes the reusability of propagation patterns towards the operational requirement changes. The main innovations are in raising the level of abstraction for behavioral schema design, and for making possible the derivation of operational semantics from structural speciications. We argue that, by using these adaptive speciication techniques, the workload required for reorganization and reprogramming of the existing investment (object base and programs), after parts of the system have been changed, can largely be avoided or minimized.