Συλλογές
Τίτλος Effective methods to tackle the equivalent mutant problem when testing software with mutation
Δημιουργός Kintis, Marinos
Συντελεστής Malevris, Nikos
Athens University of Economics and Business, Department of Informatics
Τύπος Text
Φυσική περιγραφή 195p.
Γλώσσα en
Περίληψη Mutation Testing is undoubtedly one of the most effective software testing techniques that has been applied to different software artefacts at different testing levels. Apart from mutation’s versatility, its most important characteristic is its ability to detect real faults. Unfortunately, mutation’s adoption in practice is inhibited, primarily due to the manual effort involved in its application. This effort is attributed to the Equivalent Mutant Problem. The Equivalent Mutant Problem is a well-known impediment to mutation’s practical adoption that affects all phases of its application. To exacerbate the situation, the Equivalent Mutant Problem has been shown to be undecidable in its general form. Thus, no complete, automated solution exists. Although previous research has attempted to address this problem, its circumvention remains largely an open issue. This thesis argues that effective techniques that considerably ameliorate the problem’s adverse effects can be devised. To this end, the thesis introduces and empirically evaluates several such approaches that are based on Mutant Classification, Static Analysis and Code Similarity. First, the thesis proposes a novel mutant classification technique, named Isolating Equivalent Mutants (I-EQM) classifier, whose salient feature is the utilization of second order mutants to automatically isolate first order equivalent ones. The empirical evaluation of the approach, based on real-world test subjects, suggests that I-EQM outperforms the existing techniques and results in a more effective testing process. Second, the thesis formally defines nine data flow patterns that can automatically detect equivalent and partially equivalent mutants. Their empirical evaluation corroborates this statement, providing evidence of their existence in real-world software and their equivalent mutant detection capabilities. Third, MEDIC (Mutants’ Equivalence Discovery), an automated framework that implements the aforementioned patterns and manages to detect equivalent and partially equivalent mutants in different programming languages, is introduced. The experimental investigation of the tool, based on a large set of manually analyzed mutants, reveals that MEDIC can detect efficiently more than half of the considered equivalent mutants and provides evidence of automated stubborn mutant detection. Finally, the thesis proposes the concept of mirrored mutants, that is mutants affecting similar code fragments and, more precisely, analogous code locations within these fragments. It is postulated that mirrored mutants exhibit analogous behavior of this concept supports this statement and suggests that the number of the equivalent mirrored mutants that have to be manually analyzed can be reduced approximately by half
Λέξη κλειδί Equivalent Mutant Problem
Isolating Equivalent Mutants (I-EQM)
Mutation Testing
Mutants’ Equivalence Discovery (MEDIC)
Ημερομηνία 30-06-2016
Άδεια χρήσης https://creativecommons.org/licenses/by/4.0/