DALO-APR: LLM-based automatic program repair with data augmentation and loss function optimization

The Journal of Supercomputing, Journal Year: 2025, Volume and Issue: 81(5)

Published: March 21, 2025

Language: Английский

---

CREF: An LLM-Based Conversational Software Repair Framework for Programming Tutors

Published: Sept. 11, 2024

Language: Английский

---

Poracle: Testing Patches under Preservation Conditions to Combat the Overfitting Problem of Program Repair

ACM Transactions on Software Engineering and Methodology, Journal Year: 2023, Volume and Issue: 33(2), P. 1 - 39

Published: Sept. 26, 2023

To date, the users of test-driven program repair tools suffer from overfitting problem; a generated patch may pass all available tests without being correct. In existing work, are treated as merely passive consumers tests. However, what if they willing to modify test better assess patches obtained tool? this we propose novel semi-automatic patch-classification methodology named Poracle . Our key contributions three-fold. First, design lightweight specification method that reuses test. Specifically, extend failing with preservation condition —the under which patched and pre-patched versions should produce same output. Second, develop fuzzer performs differential fuzzing containing condition. Once find an input satisfies specified but produces different outputs between versions, classify incorrect high confidence. We show our approach is more effective than four state-of-the-art classification approaches. Last, through user study assessment preferable manual assessment.

Language: Английский

---

Rust-lancet: Automated Ownership-Rule-Violation Fixing with Behavior Preservation

Published: April 12, 2024

As a relatively new programming language, Rust is designed to provide both memory safety and runtime performance. To achieve this goal, conducts rigorous static checks against its rules during compilation, effectively eliminating issues that plague C/C++ programs. Although useful, the pose challenges programmers, since programmers can easily violate when coding in Rust, leading their code be rejected by compiler, fact underscored recent user study. There exists desire automate process of fixing safety-rule violations enhance Rust's programmability.

Language: Английский

---

Enhancing the Efficiency of Automated Program Repair via Greybox Analysis

Published: Oct. 18, 2024

Language: Английский

---