Feasibility of using Source Code Changes on the Selection of Text-based Regression Test Cases

• Joelson Araújo UFPE
• Jaedson Araújo UFPE
• Cláudio Magalhães UFPE
• João Andrade UFPE
• Alexandre Mota UFPE

Abstract

This paper investigates the relationship between recently modified source code and the selection of text-based regression test cases. The main reason is that our industrial partner uses release notes documents to perform such a selection, but these documents are not so well-written as the text-based test cases. Therefore, we intend to extract useful text from source code to see whether they can serve as a source of keywords in the selection process. We present an experiment that shows promising results about this hypothesis.

Keywords: Test case selection, Source code, Regression campaign, Information retrieval