Software Instability Analysis

University Of California-Santa Cruz, Santa Cruz CA

Investigators

Abstract

CCR-0234603 Whitehead, Emmet James Univ. of Calif. - Santa Cruz Software today is tightly woven into the daily fabric of business, government, and homes. This software is maintained for many years to accommodate changing operational environments and to add new capabilities. An increasing body of evidence shows that the structure of software decays over time, leading to a cluster of related problems: changes take longer to implement since they touch more files, modular boundaries soften, and large recent changes are a good predictor of software faults. This project aims to improve the effectiveness of perfective software maintenance techniques by developing a novel tool called IVA (Instability Visualization and Analysis) that allows the rapid identification and ranking of unstable code regions, so they can be redesigned first. The concept of an instability region is introduced as a set of dependent software fragments that have been frequently modified together, and the project uses analysis and visualization techniques to show the locations and relative severity of these instabilities utilizing Instability Identification, Analysis, and Visualization. It is expected that this approach will dramatically reduce the time required to identify and prioritize instability regions. The project takes full advantage of the NASA testbed to perform activities related to validating the correctness, usability, scalability, and performance of the IVA system. Furthermore, the project will validate that IVA would have provided much earlier identification of code regions that were ultimately refactored. In addition, IVA will be incorporated in undergraduate and graduate teaching and research.