Szz

SZZ is a method used in software engineering to identify bug-introducing commits in a project's version-control history. The core idea is to start from a bug-fix commit, examine the code changes it introduced to determine which lines were defective, and then trace those lines backward through the project's history to locate the original commits that introduced them. The approach typically relies on blame or annotate data to map each line of code to the commit that last touched it.

In practice, the algorithm analyzes the diffs of bug-fix commits and, for every changed line, uses history traversal to find the earliest commit that introduced that line or caused the defect. This often involves handling code renames and moves, as well as merges, through heuristic rules. When applied to all changes in a bug-fix commit, the process yields a set of candidate bug-introducing commits for the associated bug report.

Several variants and refinements of SZZ have been proposed to address limitations such as noise from refactoring, false positives due to non-bug-related edits, or complex merge histories. Enhancements commonly seek to improve accuracy by incorporating additional signals, such as test outcomes, bug trackers, or more sophisticated handling of code movement.

Applications of SZZ include studies of defect origins and defective-change rates, identification of high-risk modules or files, and evaluation of defect-prediction models. It is widely cited in mining software repositories research and used as a baseline technique for benchmarking bug-localization and fault-diagnostic methods.

Limitations include reliance on high-quality historical data and meaningful bug-fix commits; not all bugs are tied to single-line changes; refactoring, incomplete test coverage, and merges can complicate tracing and introduce errors.