In a typical software development process, approximately 50% of the total cost is expended in testing the software being developed. Software Test Coverage Analyzers are very much helpful in reducing the cost associated with testing. There are three different techniques to collect the coverage information namely source code instrumentation, object code instrumentation and interpreter instrumentation. We compared all the three techniques and chosen the source code instrumentation technique since it is a good technique.
We have implemented the test coverage analyzer for Java using source code insertion technique. The tool gives statement coverage, branch coverage, method coverage, class coverage and time spent in testing. The user can specify the classes whose coverage information is required or not required. The tool provides the tester with coverage statistics of the current session as well as of the overall summary of the coverage information of the previous test cases. The tool allows to add triggers on execution of blocks, branches and methods and coverage reporting format can be changed. We compared the performance of our tool with commercially available tool Koalog. We conducted experiments and our results shows the developed tool is outperforming the commercial tool viz., Koalog.
Full Thesis (PS-gzipped: 171K)