InfCode-C++: Intent-Guided Semantic Retrieval and AST-Structured Search for C++ Issue Resolution
Qingao Dong, Mengfei Wang, Hengzhi Zhang, Zhichao Li, Yuan Yuan, Mu Li, Xiang Gao, Hailong Sun, Chunming Hu, Weifeng Lv · 2025
Read InfCode-C++: Intent-Guided Semantic Retrieval and AST-Structured Search for C++ Issue Resolution by Qingao Dong, Mengfei Wang, Hengzhi Zhang, Zhichao Li, Yuan Yuan, Mu Li, Xiang Gao, Hailong Sun, Chunming Hu, Weifeng Lv (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-16005.txt.
Description
Qingao Dong, Mengfei Wang, and eight co-authors build InfCode-C++, an autonomous system that fixes issues in C++ software repositories. Language model agents resolve repository issues well in Python, but C++ resists those systems because overloaded names, nested namespaces, templates, and deep control flow defeat word-based search and shallow navigation. Performance of Python-oriented agents drops drastically on C++ benchmarks.
Two retrieval mechanisms supply precise context: semantic retrieval guided by code intent, and deterministic queries over the abstract syntax tree, the parsed structure of the code. Together they localize faults and support robust patch synthesis in large typed codebases. On a multilingual benchmark C++ subset, the system beat the strongest prior agent by a wide margin.
Catalog record
| CREATOR | Qingao Dong, Mengfei Wang, Hengzhi Zhang, Zhichao Li, Yuan Yuan, Mu Li, Xiang Gao, Hailong Sun, Chunming Hu, Weifeng Lv |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-16005