Regression Test Selection Approaches: Research Trends in Regression Test Selection of Software Systems

Abstract

Regression testing ensures that changes in a software system do not intro- duce defects in previously tested functionality, but it is often a resource-intensive process. Regression Test Selection (RTS) techniques aim to reduce this cost by identifying a sub- set of test cases that must be executed after changes. This chapter explores the evolution and effectiveness of RTS methods, focusing on their evaluation criteria, including safety and precision. The chapter categorizes RTS approaches into model-based, code-based, information retrieval-based, fuzzy logic-based, machine learning-based, and security-aware methods, detailing their methodologies, advantages, and limitations. The chapter also includes an experimental evaluation comparing static RTS (STARTS) with dynamic RTS (Ekstazi) in terms of safety, precision, and test suite reduction. Finally, the chapter discusses future directions. This comprehensive overview underscores the critical role of RTS in improving the efficiency and effectiveness of regression testing for modern software systems.