A new approach to event- and model-based feature-driven software testing and comparison with similar approaches

Abstract

A software can be thought as a composition of features. Feature-oriented software development (FOSD) builds the development process on features. Part of the FOSD process is testing, and accordingly, it should be feature-driven. In model-based testing, test cases are systematically generated using the model. This research concentrates on event-based graphical models and utilizes event sequence graphs (ESGs). We develop a new test sequence generation algorithm for ESGs and named it short and frequent test sequences (SFT). Then we compare it with the existing test sequence generation algorithm called TSD. Moreover, we introduce two model-building approaches, namely daisy and swim lane, for ESGs and analyze their effects on feature-driven testing. For the evaluation, we use five different feature-driven software models. The evaluation results shows that both modeling approaches are advantageous in certain test objectives. For testing the software product as a whole, test sequence(s) should be generated by TSD from daisy modeled ESG. If a certain feature within the software product or its interaction with another feature is to be tested, then test sequence(s) should be generated by SFT from swim lane modeled ESG.