Software History under the Lens: A Study on Why and How Developers Examine It
Reviewed by Rohan Verma / 2021-11-16
Keywords: Software Evolution, Tools, Version Control
If you have ever wondered why and how developers analyze software history, or if you are a tool builder who is unsure of which features to include in your software history tool, then Codoban2015 may have some answers for you. They interviewed 14 experienced developers to examine developers' motivation, needs, and strategy for software history analysis, as well as the challenges they face during analysis. The authors then surveyed 217 people to extend these interviews' findings.
Codoban2015 found that reasons like debugging, reverse engineering, understanding the ongoing work, keeping up with changes, or analyzing the impact of a change can motivate the developers to analyze the software history. Moreover, for analysis, developers follow strategies like narrowing down the search space for commit by looking at the task, followed by identifying interesting commits, and finally analyzing the intent of the commit.
The paper also talk about the challenges faced by developers in analyzing the software history. For example, developers always suffer simultaneously from both a lack of informative commit messages and information overload. Limited tool functionality, understanding the reasons for changes, and tracing older versions are also problems; when asked, they want to be able to filter changes, group commits, get selective change notifications and trace previous versions in a better way.
But the authors did not just look at where we are: they also introduced what they call a 3-lens history model that suggests measures that tool builders can use to ensure their tool does everything developers want. This model subsumes the isolated measures developed by previous researchers, and introduces 3 lenses; Immediate lens, Awareness lens and Archaeology lens. The Immediate lens suggests to track uncommitted changes,the Awareness lens focuses on understanding the latest changes and the Archaeology lens helps in implementing features to recover knowledge from software history.
Codoban2015 Mihai Codoban, Sruti Srinivasa Ragavan, Danny Dig, and Brian Bailey: "Software history under the lens: A study on why and how developers examine it". Proc. International Conference on Software Maintenance (ICSM), 2015, 10.1109/icsm.2015.7332446.
Despite software history being indispensable for developers, there is little empirical knowledge about how they examine software history. Without such knowledge, researchers and tool builders are in danger of making wrong assumptions and building inadequate tools. In this paper we present an in-depth empirical study about the motivations developers have for examining software history, the strategies they use, and the challenges they encounter. To learn these, we interviewed 14 experienced developers from industry, and then extended our findings by surveying 217 developers. We found that history does not begin with the latest commit but with uncommitted changes. Moreover, we found that developers had different motivations for examining recent and old history. Based on these findings we propose 3-LENS HISTORY, a novel unified model for reasoning about software history.