A Method Of Fault Forewarning Based On Program Contracts And Dataflow

Computer software has been widely applied in all fields of national economy and the existence of faults in software is inevitable. Improving system reliability becomes increasingly important. As an important technology to improve reliability, Software fault dynamic forewarning is becoming a more and more growing concern. Fault Dynamic forewarning method can give a warning during runtime which can reduce losses and provide maintenance convenience when software fault occurs.This thesis proposes a software fault dynamic forewarning method based upon previous works and the analysis and comparison of several current popular methods. The method merge the program contract based fault forewarning method and dataflow based fault forewarning method together. The method parses source code statically with respect to the rules for program contracts forewarning and program dataflow forewarning to extract the set of static information, which can be used and stores as the source model described with XML, and then processed using source code instrumentation technology. In the contract-based fault forewarning method, program contracts are instruments as annotations in the source code to automatically generate the code which can be employed for dynamic run-time checks of contracts violation. In the dataflow-based fault forewarning method, variables usage probes are instrumented in the source code and then when program are running there are tracing information to detect dataflow anomaly dynamically. After preprocess the code with the function of dynamically detecting faults during runtime. This thesis adopts the way of combining two methods to improve fault dynamic detection rate and make a more comprehensive fault forewarning method.With practical examples, the thesis proves that the fault forewarning method has a good performance, shows detection results to users in graphical form and presents the availability of the method.