Is code cloning in games really different?

Abstract

Since there are a tremendous number of similar functionalities related to images, 3D graphics, sounds, and script in games software, there is a common wisdom that there might be more cloned code in games compared to traditional software. Also, there might be more cloned code across games since many of these games share similar strategies and libraries. In this study, we attempt to investigate whether such statements are true by conducting a large empirical study using 32 games and 9 non-games software, written in three different programming languages C, Java, and C#, for the case of both exact and near-miss clones. Using a hybrid clone detection tool NiCad and a visualization tool VisCad, we examine and compare the cloning status in them and compare it to the non-games, and examine the cloned methods across game engines. The results show that code reuse in open source games is much different from that of other software systems. Specifically, in contrast to the common wisdom, there are fewer function clones in game open source comparing to non-game open source software systems. Similar to non-games open source, we observed that cloning status changes between different programming languages of the games. In addition, there are very fewer clones across games and mostly no clones (no code reuse) across different game engines. But clones exist heavily across recreated (cloned) games.