Statistical Debugging and Automated Program Failure Triage

Recent years have seen great advances in software engineering and programming languages, but unfortunately, software is still far from bug-free. Static analysis is an effective approach to eliminating numerous bugs, but its conservative nature of analysis unavoidably constrains its capacity. Dynamic analysis, on the other hand, utilizes program runtime execution data, and automatically infers about program bugs. The two approaches essentially complement each other, and this book focuses on dynamic techniques, and demonstrates how to leverage program runtime data to improve software quality. The first part of this book introduces statistical debugging algorithms, which aim at automated localization of program bugs in the source code based on statistical analysis of the runtime data. The second part then dives into the discusion of automated program failure triage, exploring effective ways to prioritize software development. For both parts, comprehensive reviews of related studies are provided so that readers can easily grasp the state of the art. This book is designed for both software engineering researchers and practitioners, and can also supplement course instruction.