Consistency preserving for evolving megamodels through axiomatic semantics

Software architectures are often made to evolve due to changing user needs and/or to execution environment. The paper addresses the problem of ensuring consistency in dynamically changing software systems. The proposed approach uses a runtime megamodel which represents the architectural model of the running system and is used as a basis for model management and for changes representation. For this purpose, it should then be considered as an execution environment or more simply as a (mega-) program and thus should have to exhibit some self-∗ properties such as self-management and self-healing. And for that, we use inference rules for reasoning about, specifying, and representing change operations. Thereby preserving the megamodel's structural consistency through a logic-based approach, by defining a safe execution and an execution semantic for each operation likely to modify the megamodel contents. An initial implementation of a DSL suite for validating the approach, called MOME, is briefly presented.