The document discusses various methods of evolutionary testing for crash reproduction, focusing on the analysis and replication of bugs in Apache Commons Collections. It compares different approaches, such as record-and-play, core dump-based methods, and stack trace-based techniques, highlighting their advantages and disadvantages. A pilot study evaluates the effectiveness of different algorithms on real bugs, suggesting that tailored fitness functions can improve crash reproduction outcomes.
Related topics: