Approaches to Semantic Mutation of Behavioral State Machines in Model-Driven Software Development

Abstract

Behavior of UML state machines can be a source of interpretation problems in model to code transformation. Different solutions to the semantic variants could be defined as a special kind of mutations, similarly as in the mutation testing. State machines together with class models can be a source of an Model-Driven Software Development process aimed at building an executable application. We have compared several approaches to creating applications based on models in which semantic mutation operators of state machine behavior are used. The most promising approach has been utilized to extend the Framework for eXecutable UML (FXU) with semantic mutation facilities. The framework supports code generation from UML classes and their state machines as well as developing C# applications according to selected mutations of state machine behavior. The tool has been used in evaluation of a case study.