Preventing and Repairing Build Breakage

Build systems play a crucial role in modern software engineering. Recent studies have shown that many builds fail, mostly due to neglected maintenance. This blocks teams from continuing the development and costs time and resources to fix. The target of the thesis is to reduce build breakage by investigating changes that lead to failing builds, identifying bad and best practices for build configuration, and providing an approach to automatically repair broken builds. As a first step, we conduct empirical studies to determine changes and change patterns that lead to build breakage and reveal the reasons for build breakage. Based on these findings, we develop an approach to automatically refactor build configurations that are likely to fail and an approach to repair broken builds. We plan to evaluate our approaches first, quantitatively by measuring the performance of our approaches in open source projects and second, qualitatively by asking developers to use our approaches and give feedback on their applicability and usefulness.

Preprint, Extended Abstract, Poster