Robustness constitutes an important characteristic of software components and systems. To evaluate this feature, the testing of negative scenarios is essential. Triggering error handling, however, might terminate the system under test before testing all values and value combinations of a negative test case. As a result, faults might remain undetected and propagate to system crashes. To counteract this effect, combinatorial robustness testing extends combinatorial testing and separates the generation of positive and negative test cases. For this procedure, the testing framework requires additional information in form of error-constraints describing invalid values and invalid value combinations. Nonetheless, manually constructing those constraints might lead to over- or under-constrained models. This thesis presents a concept that equips a combinatorial robustness testing framework with the capability to automatically generate error-constraints using fault characterization algorithms in order to replace the manual modeling. To evaluate this concept, we provide a prototypical implementation in Java and conduct several experiments on multiple benchmark test models. For those experiments, we implement several fault characterization algorithms possessing different characteristics and representing various approaches.

Info

• • • Language: German, English
      • Supervisor: Konrad Fögen (E-Mail)
      • coffee4j on Github
      • ROBUSTA – Combinatorial Robustness Testing