Code Smells in iOS Apps: How Do They Compare to Android?

Abstract

With billions of app downloads, the Apple App Store and Google Play Store succeeded to conquer mobile devices. However, this success also challenges app developers to publish high-quality apps to keep attracting and satisfying end-users. In particular, taming the ever-growing complexity of mobile apps to cope with maintenance and evolution tasks under such a pressure may lead to bad development choices. While these bad choices, a.k.a. code smells, are widely studied in object-oriented software, their study in the context of mobile apps, and in particular iOS apps, remains in its infancy. Therefore, in this paper, we consider the presence of object-oriented and iOS-specific code smells by analyzing 279 open-source iOS apps. As part of this empirical study, we extended the Paprika toolkit, which was previously designed to analyze Android apps, in order to support the analysis of iOS apps developed in Objective-C or Swift. We report on the results of this analysis as well as a comparison between iOS and Android apps. We comment our findings related to the quality of apps in these two ecosystems. Interestingly, we observed that iOS apps tend to contain the same proportions of code smells regardless of the development language, but they seem to be less prone to code smells compared to Android apps.