GitHub has introduced the GitHub Copilot project, which develops an intelligent assistant capable of generating standard structures while writing code. The system was created in collaboration with the OpenAI project and utilizes the OpenAI Codex machine learning platform, trained on a vast array of source texts hosted in public GitHub repositories.
Unlike traditional code completion systems, GitHub Copilot has the ability to generate quite complex blocks of code, including complete functions, synthesized with consideration of the current context. GitHub Copilot adapts to the developer's coding style and takes into account the APIs and frameworks used in the program. For example, if there is a JSON structure example in the comment when starting to write a function to parse this structure, GitHub Copilot will suggest ready-made code, while for writing routine enumerations of repetitive descriptions, it will generate the remaining positions.
Currently, GitHub Copilot is available as a plugin for the Visual Studio Code editor. It supports code generation in programming languages such as Python, JavaScript, TypeScript, Ruby, and Go using various frameworks. In the future, there are plans to expand the number of supported languages and development systems. The operation of the plugin is performed by making requests to an external service running on GitHub, which also receives the content of the edited code file.
Source: opennet.ru
