220 likes | 309 Views
How Significant Is the Effect of Faults Interaction on Coverage Based Fault Localizations?. Xiaozhen Xue Advanced Empirical S oftware T esting G roup Department of Computer Science Texas Tech University 10.1.2013 . Outline. Introduction Research Questions Experimental procedure
E N D
How Significant Is the Effect of Faults Interaction on Coverage Based Fault Localizations? XiaozhenXue Advanced Empirical Software Testing Group Department of Computer Science Texas Tech University 10.1.2013
Outline • Introduction • Research Questions • Experimental procedure • Data analysis • Discussion • Conclusion
Introduction • Software testing is an important process in software life cycle. • Fault localization is a major activity in software testing. • Fault localization is time consuming and expensive.
Introduction • Coverage based fault localization
Introduction • example
Introduction • Coverage based fault localization techniques is effect when single fault in presence. • It is a common belief that the coverage based fault localization techniques are less effective in the presence of multiple faults.
Introduction • Motivation: Empirically investigate the effectiveness of Coverage Based Faults Localization when multiple faults exist in the program.
Research Questions • RQ1: How prevalent is the fault interferences phenomenon in the object-oriented programming languages such as Java? • RQ2: How significant is the influence of fault interactions on debugging cost? • RQ3: Do the frequencies of fault interferences vary for different ranking metrics? • RQ4: How different is the frequency of observing fault interference instances for various number of faults? • RQ5: How different is the frequency of observing fault interference instances for various type of faults?
Experimental procedure • Subjects programs
Experimental procedure • Multiple faults
Experimental procedure • metrics
Experimental procedure • Metrics proposed in our previous work
Data Analysis • Data visualization
Data Analysis • The significance of faults interactions
Data Analysis • The significance of faults interactions
Data Analysis • The significance of faults interactions
Data Analysis • The significance of faults interactions
Discussion • The prevalence of faults interference
Discussion • The prevalence of faults interference
Data Analysis • The roles of ranking metrics
Conclusion • The coverage-based fault localization techniques are still effective in localizing faults in the presence of multiple faults. • The effect of fault interactions on debugging cost is negligible and statistically insignificant. • The frequency of observing fault interference instances differ for various ranking metrics.