Towards Tool-Support for Test Case Selection in Manual Regression Testing

Manual regression testing can be a time-intensive and costly activity. Required efforts can be reduced by selecting only the tests for re-testing that verify actually modified system parts. However, if testers are not familiar with the system implementation the selection of relevant test cases may become difficult. In this paper we report on our experiences with the development of a tool-based approach supporting the selection of manual regression tests. The presented approach is developed together with the software testing department of an international company. Test cases are selected by analyzing different kinds of information resources, i.e., code coverage information and data provided by versioning systems. Experience shows that code coverage information can assist in selecting candidate test cases for regression testing. However, we also encountered some principal challenges in implementing test case selection in practice: Relying solely on code coverage often leads to a large set of test cases, available versioning systems lack the necessary details to map code changes to relevant structural elements, and collecting and keeping coverage data for manual regressing testing up-to-date involves additional costs and effort.