As mobile devices become more capable and mobile adaption grows acrossthe world, the complexity of mobile applications increases. For example, mo-bile applications let consumers view their banking details and legitimize them-selves. With mobile applications serving these crucial functions, the need forquality and robustness increases. Unit tests are one way to measure the qualityof an application. However, there is a need to evaluate the unit tests themselvesin order to increase their effectiveness. Mutation testing is one approach to testthe application source code by deliberately mutating it in an effort to see if theunit tests of the application detect the change.
This study investigates mutation testing on an Android application. Morespecifically, it evaluates a form of extreme mutation testing using pseudo-tested methods. With pseudo-tested methods, entire method bodies are re-placed with a single return statement. This study introduces a framework fordetecting pseudo-tested methods within the continuous integration environ-ment as well as the capability of measuring the usefulness of generated re-ports that highlight these methods. Moreover, the study conducts interviewswith multiple developers discussing the pseudo-tested methods in detail.
We implement this framework within the continuous integration environ-ment using multiple components. Our results show that developers are pos-itive towards having a tool for detecting pseudo-tested within the continuousintegration environment. Moreover, the results indicate that methods writtenby developers are more useful to test than auto-generated methods. We drawthe conclusion that detection of pseudo-tested methods can be beneficial inincreasing the quality of mobile applications. However, we note that furtherwork is needed to eliminate the prevalence of false-positives found in our study.