Advances in Software Development and Maintenance

The field of software development and maintenance is moving towards more efficient and effective methods of integrating changes, debugging, and logging. Researchers are exploring new approaches to patch integration, taint analysis, and performance optimization. One notable direction is the use of machine learning and natural language processing to improve log level prediction, log parsing, and template generation. Another area of focus is the development of tools and techniques to support developer sensemaking and debugging, such as question-driven debugging interfaces and interactive summary tools. Noteworthy papers in this area include Refactoring-Aware Patch Integration Across Structurally Divergent Java Forks, which presents a system for integrating bug-fix patches across divergent Java forks, and TraceLens, which proposes a question-answer style debugging interface for taint analysis. Additionally, papers like OmniLLP and LLMLog demonstrate the potential of large language models in log level prediction and template generation.

Sources

Refactoring-Aware Patch Integration Across Structurally Divergent Java Forks

TraceLens: Question-Driven Debugging for Taint Flow Understanding

An Empirical Study on Method-Level Performance Evolution in Open-Source Java Projects

From Noise to Knowledge: Interactive Summaries for Developer Alerts

Improving Merge Pipeline Throughput in Continuous Integration via Pull Request Prioritization

OmniLLP: Enhancing LLM-based Log Level Prediction with Context-Aware Retrieval

Empirical Analysis of Temporal and Spatial Fault Characteristics in Multi-Fault Bug Repositories

Plug it and Play on Logs: A Configuration-Free Statistic-Based Log Parser

LLMLog: Advanced Log Template Generation via LLM-driven Multi-Round Annotation

Built with on top of