Combined Constraint-Based Analysis for Efficient Software Regression Detection in Evolving Programs

Abstract

Software regression is a bug that makes software stop functioning normally after a certain event. In this paper, we investigate detecting regression when software evolves to a new version. In this context, regression bugs occur in parts of software that already passed testing process in the old version. Hence, such kind of bugs is difficult to be discovered if normal strategy like white-box testing is applied. Moreover, since both old and new versions must be taken into account during the testing process, the computational cost is usually high in this con-text. Concolic testing in an emerging extension of white-box testing that can reduce significantly the number of execution paths needed to be analyzed. However, the typical concolic testing is not really efficient when dealing with software regression. Thus, we propose a new approach based on combined constraint to solve this problem, known as CTGE (Efficient Constraint-based Test-cases Generation) approach. The soundness of our theoretical contribution is formally proved and supported by some initial experiments conducted in education environment. © Springer-Verlag Berlin Heidelberg 2013.