Adaptive Fault Localization in Regression Testing Using Multi-Fault Coverage-Driven Test Minimization
Appari Pavan Kalyan, Harsh Pratap Singh, B. Kavitha Rani · Turkish Journal of Computer and Mathematics Education (TURCOMAT) · 2021
In software testing, Spectrum-based Fault Localization (SBFL) relies on the coverage of test cases and their outcomes (pass/fail) to assess the "suspiciousness" of program components, such as lines of code. SBFL is widely recognized for its simplicity and scalability, making it a prevalent technique in fault localization. However, traditional SBFL heuristics often struggle in scenarios where multiple faulty components coexist within a program, leading to suboptimal fault identification. To address this limitation, we propose a novel algorithm inspired by Multi-Fault Coverage-Driven Test Minimization (MFCDTM). Our approach enhances existing SBFL heuristics by re-ranking faulty components that are initially ranked low by base SBFL metrics, thereby improving their prioritization. We have implemented this algorithm, which aims to "bubble up" faulty components that might otherwise be overlooked by conventional SBFL strategies. Comparative evaluations demonstrate that our technique significantly reduces the developer effort required for fault localization, particularly in multi-fault scenarios, with statistically significant improvements over traditional SBFL methods.