Enabling test case selection of automatically generated unit tests through traceability: An empirical study

Abstract

Anticipating the effects of changes/modifications to source code, is a difficult if not an impossible process, unless the right tools or methods are applied. One way of handling change impact analysis is through test case selection which can cut down the testing time, as it only selects and runs the tests that have been affected by a change. In order to realize this, the method of traceability is applied on source code and automatically generated unit tests. This approach aims at facilitating software maintenance by cutting down the time and the effort required to re-validate changes. This paper investigates the impact of traceability with the intention of evaluating the effects on debugging time and mutation kill score. By conducting an experiment with 8 subjects, the results showed that no major statistical differences were found, which is likely to change with a larger sample size. Nonetheless, to generalize the impact of traceability between code and automated unit tests, further research is required, however the paper provides insights and deeper understanding of the problem as well a guideline for future studies.