Automatic Performance Testing for Image Displaying in Android Apps

Abstract

Image displaying in Android apps is resource-intensive. Improperly displayed images result in performance degradation or even more severe consequences like app crashes. Existing static performance anti-pattern checkers are conservative and limited to a small set of bugs. This paper presents ImMut, the first test augmentation approach to performance testing for image displaying in Android apps to complement these static checkers. Given a functional test case, ImMut mutates it towards a performance test case by either (1) injecting external-source images with large ones or (2) copy-pasting a repeatable fragment and slightly mutating the copies to display many (potentially distinct) images. Evaluation on our prototype implementation showed promising results that ImMut revealed 14 previously unknown performance bugs that are beyond the capability of state-of-the-art static checkers.