How Cost Reduction in Recovery Improves Performance in Program Design Tasks

Changing source code often leads to undesired implications, raising the need for recovery actions. Programmers need to manually keep recovery costs low by working in a structured and disciplined manner and regularly performing practices such as testing and versioning. While additional tool support can alleviate this constant need, the question is whether it affects programming performance? In a controlled lab study, 22 participants improved the design of two different applications. Using a repeated measurement setup, we compared the effect of two sets of tools on programming performance: a traditional setting and a setting with our recovery tool called CoExist. CoExist makes it possible to easily revert to previous development states even, if they are not committed explicitly. It also allows forgoing test runs, while still being able to understand the impact of each change later. The results suggest that additional recovery support such as provided with CoExist positively affects programming performance in explorative programming tasks.