Articles

The article details Rslint, a new frontend code linting tool developed by ByteDance's Web Infra Team, designed to address the performance bottlenecks and complex configuration issues of ESLint in large-scale projects. Rslint is optimized from the discontinued tsgolint project, developed in Go, and deeply integrated with TypeScript semantics, achieving a 20-40x speed improvement. Its core acceleration points are Go's concurrent performance, type-aware inspection mechanism, and native support for parallelization and Monorepo. The article also highlights Rslint's advantages, including exceptional performance, out-of-the-box TypeScript support, an enhanced developer experience, and low migration costs. Finally, it provides a simple installation, configuration, and operation guide, and looks forward to Rslint's potential as a high-performance, type-driven Linter tool in the future frontend ecosystem.

The article details the core updates of React Native version 0.81. First, the new version fully supports Android 16 (API 36), enabling Edge-to-Edge immersive interface and predictive back gesture by default, which requires developers to re-evaluate UI compatibility. Second, JavaScriptCore (JSC) is no longer provided with the core package and is now maintained by the community, which means that the officially recommended Hermes Engine is poised to become the dominant engine. The most notable is the significant improvement in iOS build speed. Through the experimental precompiled build feature introduced in cooperation with Meta and Expo, the compilation speed of large iOS projects can be increased by up to 10 times. In addition, the new version also improves environmental requirements (Node.js ≥ 20.19.4, Xcode ≥ 16.1), optimizes the stability of Metro Bundler, and improves the debugging experience, making error information more complete and troubleshooting more intuitive. These updates are designed to improve React Native's adaptation capabilities on multiple platforms and the overall experience for developers.

This article provides a detailed introduction to the open-source project shadcn-ui-mcp-server, which aims to address the lack of contextual understanding of specific component libraries when AI code assistants generate UI code. It uses the core mechanism of Model Context Protocol (MCP) to structure the documentation, source code, examples, and other information of shadcn/ui components into a format understandable by AI models. This transforms AI assistants from mere code generators into comprehensive front-end resources, enabling them to accurately locate components, generate complete layouts, and complete code structures. The article elaborates on the functions of the MCP Server, including viewing component source code, supporting full-page layouts, and integration methods with mainstream editors such as VS Code (via the Continue plugin) and Cursor. Finally, the article provides a detailed guide on how to run and configure the project, emphasizing that this solution can save developers who use shadcn/ui a lot of time in consulting documentation and significantly improve development efficiency.