Regression Testing in Software Product Lines

Abstract

Software product lines is a strategic choice for structuring many large-scale software today to support systematic reuse while reducing development costs.
One key challenge for software product lines is to efficiently manage variability throughout their life-cycle while avoiding any regression on untouched code. In this thesis, we address the problem of regression testing in software product lines during their evolution (e.g., addition of functionality, modification or dele- tion of code). Hence, we propose an approach to the problem and illustrate it using the classical example of the vending machine. Indeed, we implemented the example in a feature model on FeatureIDE and made it to evolve. Then, using a regression testing tool, EvoSuiteR, running generated tests on both the evolved software product line and the original software product line, we were able to successfully generate valid regression tests for the original software product line. (In each case, we have tested the evolved feature models for regression and were able to find a sample test suite for the purpose).
This paper therefore demonstrates a valid practical approach for regression test generation in software product lines. Finally, we give preliminary results in the assessment of our regression test generation method.

Date of Award	30 Aug 2021
Original language	English
Awarding Institution	University of Namur
Supervisor	Pierre Yves Schobbens (Supervisor)