Characterizing Multi-Hunk Patches: Divergence, Proximity, and LLM Repair Challenges

2025-06-04

AI Summary

Prior automated program repair (APR) research predominantly focuses on single-hunk patches, overlooking the coordination challenges arising from semantically interdependent changes across multiple code hunks. Method: This paper systematically characterizes the structural properties of multi-hunk patches, introducing two key dimensions: *inter-hunk divergence*—quantifying semantic, syntactic, and file-level differences—and *program-level spatial proximity*, a novel classification framework. Using the real-world HUNK4J bug dataset, we conduct lexical, syntactic, and file-level analyses and empirically evaluate six state-of-the-art LLMs. Contribution/Results: We find that LLM-based repair success rates decline significantly with increasing inter-hunk divergence and spatial dispersion; all models fail entirely on highly dispersed “Fragment”-class patches. Inter-hunk divergence emerges as the primary limiting factor for LLM repair capability. Consequently, we propose a divergence-aware APR paradigm—a new foundation for scalable, semantics-guided automated repair.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Multi-hunk bugs, where fixes span disjoint regions of code, are common in practice, yet remain underrepresented in automated repair. Existing techniques and benchmarks pre-dominantly target single-hunk scenarios, overlooking the added complexity of coordinating semantically related changes across the codebase. In this work, we characterize HUNK4J, a dataset of multi-hunk patches derived from 372 real-world defects. We propose hunk divergence, a metric that quantifies the variation among edits in a patch by capturing lexical, structural, and file-level differences, while incorporating the number of hunks involved. We further define spatial proximity, a classification that models how hunks are spatially distributed across the program hierarchy. Our empirical study spanning six LLMs reveals that model success rates decline with increased divergence and spatial dispersion. Notably, when using the LLM alone, no model succeeds in the most dispersed Fragment class. These findings highlight a critical gap in LLM capabilities and motivate divergence-aware repair strategies.

Problem

• Characterizing multi-hunk bug patches in real-world defects
• Measuring hunk divergence and spatial proximity in patches
• Evaluating LLM repair challenges with divergent and dispersed patches

Innovation

• Introduces HUNK4J dataset for multi-hunk patches
• Proposes hunk divergence metric for patch variation
• Defines spatial proximity classification for hunk distribution