Search Results

This report documents the program and the outcomes of Dagstuhl Seminar "Code Search" (24172). The seminar brought together researchers and practitioners working on techniques that enable software developers to find code and artifacts related to code. The participants discussed the state of the art in code search, identified open problems, and discussed future directions for research and practice. The seminar was structured with keynote talks, short talks, and breakout groups. Breakout groups identified how researchers can situate their code search research in terms of the targeted user groups, the access point for the developer, and the stage of software development that is most relevant to the code search tasks. Synergies between generative AI and Code Search were discussed, concluding that for some users and some tasks, generative AI can work with Code Search to enhance the developer experience and effectiveness. For other tasks, code search without generative AI would be more effective because of concerns regarding data provenance, update frequency, privacy, and the need for correctness.