Research On Dynamic Evolution Method Of Aspect-Oriented Software Architecture

An increasing complexity and evolvablity of software is enhanced according to the users' requirements and computing environments.Software evolution refers to the process that software changes itself to achieve the anticipant form,and it mainly divided into two categories,static evolution and dynamic evolution.Owing to the merit of continuously providing services,dynamic evolution has become a hotspot in software engineering presently.Software architecture is a key factor of effecting software evolution,but existing software architecture usually not pays enough attention on the non-functional(crosscutting) attributes,resulting in the difficulty of separation of concerns and degradation or even abandon of software architecture,it makes dynamic evolution of software system more difficult and costly.To address the problems above,this dissertation introduces the idea of Aspect-Oriented into software architecture.With a sketch on basic concepts,main principle and advantage analysis of AOP,its concepts are introduced into software architecture and a core model of Aspect-Oriented Software Architecture(AOSA) is established.We present a formal specification of all AOSA elements,which lays the foundation for analysis,deduction and verification of dynamic evolution behaviors,based on the temporal logic.Considering the nonvisibility and uncontrollability of software architecture at runtime, in this dissertation,a reflective architecuture RAOSA of supporting dynamic evolution, which based on the principle of separation of concerns,is proposed;key elements of AOSA and a group of evolution rules and evolution strategies that based on ECA pattern is formalized.Based on all aforementioned work,with a simple application of Internet auction system,this dissertation establishes the formal specification of its RAOSA framework;and we use the evolution rules and evolution strategies defined above to show the feasibility and validity of the proposed dynamic evolution method. From the above,thanks to the penetrability of separation concerns from requirement to implementation,the core model of AOSA in this dissertation is helpful to improve the flexibility and reusability of system;the reflective architecture RAOSA and the method of all AOSA elements based on Temproal Logic is also helpful to the implementation of dynamic evolution and analysis,deduction and verification of their behaviors.