Dynamic Evolution Of Software Evolution Credible Running Instance

As more and more software requirements are submitted by the people, the software become more and more complex; the difficulty of developing software is becoming bigger and bigger, how to make software development everything in good order and well arranged to become a hot research direction. From the perspective of the software process to improve the quality of software has become an effective method. Software process requires constant evolution to adapt to the changing needs of users, and can be dynamically added trustworthy attribute in the evolution process to make the whole process more credible, and can better meet the needs of users.The software evolution process can be divided into static and dynamic evolution, dynamic evolution is more practical. This paper research on the dynamic evolution and efficient process inheritance technique is used to test instance migration. Although process inheritance technology is very suitable to test the ability of migration of the trustworthy process, but its limitations larger, this paper introduces the Petri nets decomposition method based on basic structure that using polychromatic sets theory and proved that this decomposition method in accordance with some basic and important properties of Petri nets. If multiple trustworthy process cannot be migrated in one step according to the original plan, using this method that above mentioned made multiple process break down to migrate in multi-step and as far as possible to ensure its parallelism. For the process that does not use process inheritance technology, this paper presents the evolution of the structural adjustment, and write the relevant evolutionary algorithm framework. Through the above method can increase the success rate of trustworthy evolution, and make the software more reliable. In trustworthy evolution process often have many abnormal conditions make the whole software evolution process is interrupted, so that destroyed the automatic evolution, reduces the efficiency of evolution. In order to solve this problems we must be handle the error that appeared in the trustworthy evolution process. First of all is to classify exceptions and to know what kind of exception is we can solve in the modeling phase. Then proposed a Petri nets-based modeling tool to support for predictable exception handling and designed a framework for exception handling. Finally, using the cosine similarity algorithm to find the best match for exception handling.In short, this thesis research on the dynamic trustworthy evolution of the running instance of the software evolution process from modeling to instance migration and then exception handling. This trustworthy evolution method for enhancing the credibility of the entire software has very important practical significance.