QADroid: regression event selection for Android applications

Abstract

Popular Android applications undergo frequent releases. Ensuring functional testing of the new features, as well as regression testing of the previous functionality, are time-consuming and error-prone. Thus, there is a need for a tool that eases the testing efforts as well as saves the overall time of the product release cycle. In this work, we present QADroid, the first activity- and event-aware regression selection tool for Android apps. Salient features of QADroid are: (i) a richer change-set analyzer that covers code as well as non-code components for regression, (ii) it presents a pictorial representation of the app’s functioning, and (iii) it displays the regression points in the app as a mapping between activities to user-elements to events. Features (ii) and (iii) help the testers in understanding the technical findings better. We evaluated QADroid on 1105 releases of 50 open source Android projects. The results show that QADroid reduced the activity selection by 58% and event selection by 74% compared to the traditional way of exhaustive testing of all activities and events, thereby significantly reducing the manual testing efforts.