Android testing tools generate sequences of input events to exercise the state space of the app-under-test. Existing search-based techniques systematically evolve a population of event sequences so as to reach certain objectives such as maximal code coverage. The hope is that the mutation of fit event sequences leads to the generation of even fitter event sequences. However, such random mutations may truncate the path of the generated sequence through the app’s state space at the point of the first mutated event. States that contributed considerably to the original sequence’s fitness may not be reached by the sequence’s offspring.
In this paper, we propose instead to evolve a population of states which can be captured upon discovery and resumed when needed. The hope is that generating events on a fit program state leads to the transition to even fitter states. For instance, we can quickly deprioritize testing the main screen state which is visited by most event sequences, and instead focus our limited resources on testing more interesting states that are otherwise difficult to reach.
We call our approach time-travel testing because of this ability to travel back to any state that has been observed in the past. We implemented time-travel testing into TimeMachine, a time-travel-enabled version of the successful, automated Android testing tool Monkey. In our experiments on a large number of open- and closed-source Android apps, TimeMachine outperforms the state-of-the-art search-based Android testing tools Sapienz and Stoat, both in terms of coverage achieved and crashes found.
Wed 27 May
|16:10 - 16:25|
|16:25 - 16:35|
|16:35 - 16:50|
|16:50 - 17:10|
Jiwei YanInstitute of Software, Chinese Academy of Sciences, Hao LiuBeijing University of Technology, Linjie PanInstitute of Software, Chinese Academy of Sciences, Jun YanInstitute of Software, Chinese Academy of Sciences, Jian ZhangInstitute of Software, Chinese Academy of Sciences, Bin LiangRenmin University of China, China
|17:10 - 17:30|
|17:30 - 17:50|