Automatically Identifying Dead Fields in Test Code by Resolving Method Call and Field Dependency

Dead fields are the unused setup fields in the test code which reduce the comprehensibility and maintainability of a software system. A test class contains dead fields when developers initialize setup fields without analyzing the usage of fields properly. Manually identifying dead fields to remove from the code is a time consuming and error-prone task. In this paper, a technique named Dead Field Detector (DFD) has been proposed to detect dead fields automatically. The technique constructs Call Graph (CG) and Data Dependence Graph (DDG) from test code to find method invocation and field dependency relationships, respectively. It identifies the fields initialized in the setup method and its invoked methods from CG. It finds setup fields by collecting the initialized fields and their dependent fields from DG. To determine the usage of the setup fields, it checks the bodies of the test methods and their invoked methods obtained from CG. All the unused setup fields are separated from the used fields and considered as dead fields. In order to evaluate DFD, an open source project named eGit was used. The result analysis shows that DFD has identified 14.03% more setup fields and 60.98% more dead fields than an existing technique named TestHound for eGit.