Study On The Extraction Method Of Concurrent Regression Defects Based On Historical Evolutionary Information

Regression defect means that during the development process of the program,the program function of the normal working program cannot operate normally due to the wrong modification by the developer or maintenance personnel.Researchers have proposed a variety of regression defect locating techniques,but few research work is used to locate concurrent regression defects in a multithreaded environment.One of the main challenges of concurrent regression defect research is the lack of concurrent regression data sets for experiments in the community.In order to promote research in the field of concurrent regression defects and provide an effective research evaluation benchmark,this article mainly completes the following tasks:(1)Benchmark project research.The excellent results in the field of concurrency and regression defects were investigated,and the frequency of occurrence of open source projects that were used as evaluation benchmarks among them,the degree of attention of integrated projects in the open source community,and the analysis of the effectiveness of information in the defect integrated system were analyzed.A filtering method for concurrent regression defect reports in a widely used defect tracking system.(2)Source code association and code evolution analysis.Analyzed 35 JAVA projects and more than 600 defect reports,and carried out related source code correlation and code evolution analysis for these reports,which verified the effectiveness of the SZZ algorithm in locating regression defects and introducing changes.(3)Benchmark tool construction.A data set of concurrent regression defects,Con Reg4 J,was constructed,which contained 20 real defects,60 executable programs and required environments,and historical information about the evolution between versions of programs.Program defect introduction version and defect repair version,with a maximum interval of about 700 days and 10 million lines of code modification.(4)Multi-threaded trajectory modeling.Improved a regression defect location method based on program execution trajectory,realized Mcirobat,and added multi-threaded trajectory recording and storage features to it,and proposed a dynamic multi-threaded trajectory modeling method.In summary,this article has made the following contributions: 1)We introduce a new concurrent regression defect set Con Reg4 J,which can be used as an effective evaluation benchmark to promote research in related fields.2)We added a new feature of concurrent program trajectory modeling to Microbat to facilitate the research of concurrent regression defect location based on trajectory.