Research On Software Architecture Evolution Based-on Messages And Component Operations

Software development based-on components are beneficial to reuse components, to improve development efficiency and software quality. Interaction by passing messages among components makes software more reliable, components more independent. Software evolvement is an important part of software life cycle, and how to develop evolvable software has become a hotspot in IT community. Software architecture(SA) offers a base for component combination and software evolution. It has significance in theory and applications to research software architecture evolution based-on messages and component combination operations in this paper.According to the message types which a component can receive and send, a component combination model based-on messages is employed, component executing paths are defined, and combination operations are depicted formally. Graphs based-on component combination operations are given, an algorithm on conversion between the graph and algebraic express of SA is designed.The design pattern is used to bewrite the decomposite of non-function requires,the relation between non-function requires and SA is set up with structure workflows.The non-function attributes of software architectures are evaluated synthetically based on structure workflows and component combinational operations. According to the evaluation result, a model on software architecture evolution based-on non-function requirements is given.SA intuitionistic complexity metric method is given, because software complexity is prone to introduce errors, to make software understand difficultly, and to cause software insecure. A simplifying arithmetic of system models is proposed,and this lays a foundation for research on software architecture evolution automatically.An analysis approach of ripple-effect of component operations is described based-on component combination operations and SA,a new metric of SA evolution is developed, and an metric algorithm is designed. Traditional metric deficiency of too much data needed, excessive dependence on manpower participating in and individual experience, and only adaptation to small size systems, is overcome.In order to keep application business continuous to avoid unacceptable delay, expense and risk by closing and restarting systems, the dynamic characteristic of components and connectors based-on messages is depicted; a dynamic evolution framework of software architecture centering connectors is given by using reflective mechanism.In order to associate requirement variations with software evolution(SE),the message dependency matrix is defined, and arithmetic is designed to get feature traces from message dependency matrixes. According to the relation between requirement variations and feature traces, SE is defined, and classifed based-on the types of requirement variations, its influence is analyzed quantificationally based-on feature traces.SE found on requires and feature traces is propitious to evolution location, evolution management and evolution verification.