PACGBI: A Pipeline for Automated Code Generation from Backlog Items
While there exist several tools to leverage Large Language Models (LLMs) for code generation, their capabilities are limited to the source code editor and are disconnected from the overall software development process. These tools typically generate standalone code snippets that still require manual integration into the codebase. There is still a lack of integrated solutions that seamlessly automate the entire development cycle, from backlog items to code generation and merge requests. We present the Pipeline for Automated Code Generation from Backlog Items (PACGBI), an LLM-assisted pipeline integrated into GitLab CI. PACGBI reads backlog items in the code repository, automatically generates the corresponding code, and creates merge requests for the generated changes. Our case study demonstrates the potential of PACGBI in automating agile software development processes, allowing parallelization of development and reduction of development costs. PACGBI can be utilized by software developers and enables non-technical stakeholders and designers by providing a holistic solution for using LLMs in software development. A screencast of this tool is available at https://youtu.be/TI53m-fIoyc, its source code at https://github.com/Masa-99/pacgbi.
Wed 30 OctDisplayed time zone: Pacific Time (US & Canada) change
10:30 - 12:00 | Code generation 2Research Papers / Tool Demonstrations at Gardenia Chair(s): Yangruibo Ding Columbia University | ||
10:30 15mTalk | Preference-Guided Refactored Tuning for Retrieval Augmented Code Generation Research Papers Xinyu Gao , Yun Xiong Fudan University, Deze Wang National University of Defense Technology, Zhenhan Guan Fudan University, Zejian Shi Fudan University, Haofen Wang Tongji University, Shanshan Li National University of Defense Technology Pre-print | ||
10:45 15mTalk | Sifting through the Chaff: On Utilizing Execution Feedback for Ranking the Generated Code Candidates Research Papers Zhihong Sun Shandong Normal University, Yao Wan Huazhong University of Science and Technology, Jia Li , Hongyu Zhang Chongqing University, Zhi Jin Peking University, Ge Li Peking University, Chen Lyu Shandong Normal University | ||
11:00 15mTalk | Promise and Peril of Collaborative Code Generation Models: Balancing Effectiveness and Memorization Research Papers Pre-print | ||
11:15 15mTalk | JavaBench: A Benchmark of Object-Oriented Code Generation for Evaluating Large Language Models Research Papers Jialun Cao Hong Kong University of Science and Technology, Zhiyong Chen Nanjing University, Jiarong Wu The Hong Kong University of Science and Technology, Shing-Chi Cheung Hong Kong University of Science and Technology, Chang Xu Nanjing University | ||
11:30 15mTalk | PACGBI: A Pipeline for Automated Code Generation from Backlog Items Tool Demonstrations Mahja Sarschar Hochschule für Technik und Wirtschaft Berlin, Gefei Zhang HTW Berlin, Annika Nowak Capgemini | ||
11:45 15mTalk | Contextualized Data-Wrangling Code Generation in Computational Notebooks Research Papers Junjie Huang The Chinese University of Hong Kong, Daya Guo Sun-yat Sen University, Chenglong Wang Microsoft Research, Jiazhen Gu Chinese University of Hong Kong, Shuai Lu Microsoft Research, Jeevana Priya Inala Microsoft Research, Cong Yan Microsoft Research, Jianfeng Gao Microsoft Research, Nan Duan Microsoft Research, Michael Lyu The Chinese University of Hong Kong | ||