From behaviour preservation to behaviour modification: constraint-based mutant generation
Proceedings of the 32nd ACM/IEEE International Conference on Software Engineering - Volume 1
On generating mutants for AspectJ programs
Information and Software Technology
| Hi-index | 0.00 |
This paper presents some connections between dependence analysis and mutation testing. Specifically, dependence analysis can be applied to two problems in mutation testing, captured by the questions: 1. How do we avoid the creation of equivalent mutants? 2. How do we generate test data that kills non-equivalent mutants? The theoretical connections described here suggest ways in which a dependence analysis tool might be used, in combination with existing tools for mutation testing, for test-data generation and equivalent-mutant decision. In this paper the variable orientated, fine grained dependence framework of Jackson and Rollins is used to achieve these two goals. This framework of dependence analysis appears to be better suited to mutation testing than the more traditional, Program Dependence Graph (PDG) approach, used in slicing and other forms of program analysis. The relationship between dependence analysis and mutation testing is used to define an augmented mutation testing process, with starts and ends with dependence analysis phases. The pre-analysis removes a class of equivalent mutants from further analysis, while the post-analysis phase is used to simplify the human effort required to study the few mutants that evade the automated phases of the process.