Improving Evolutionary Testing By Flag Removal

Abstract

Abstract This paper argues that Evolutionary testing can be improved by transforming programs with flags into flag free programs. The approach is evaluated by comparing results from the application of the DaimlerChrysler Evolutionary Testing System to programs with flags and their transformed flagfree counterparts. The results of this empirical study are very encouraging. Programs which could not be fully covered become fully coverable and the number of generations required to achieve full coverage is greatly reduced.