March 2026 Summaries
2 posts from Code Input
Filter
Month:
Year:
Post Summaries
Back to Blog
The author explores the challenge of sharing consistent code logic across multiple editors and platforms while developing a CodeOwners platform, focusing on integrating CODEOWNERS rules into various editors and potentially LLM agents. Initially considering WebAssembly for consistency, the author encounters performance limitations and instead explores using the Language Server Protocol (LSP) to create a server that can communicate with any editor implementing LSP. Despite initial doubts about the complexity of building LSP servers, the author finds it manageable, especially with tools like tower-lsp-server in Rust, which streamlines the process. The text illustrates how LSP servers can handle tasks like autocomplete and automatic text edits, and even suggests the potential for integrating AI chatbots within editors. The author concludes with a reflection on why LSP servers are not more broadly used beyond programming languages, suggesting that while they have specific limitations, they offer a standardized approach that could be beneficial amid the growing influence of AI.
Mar 19, 2026
1,651 words in the original blog post.
The author reflects on their transition from Rust to TypeScript while working on a front-end project, noting the challenges and differences, particularly in error handling and type systems. They express a preference for Rust's Result type for its structured error management and minimal chance of unpredictable failures, which TypeScript lacks out of the box. To address this gap, they explore using the neverthrow library, which approximates Rust's error handling ergonomics by allowing custom error types and structured error management, although not as seamlessly as Rust's native functionalities. The author acknowledges that while TypeScript's type system is complex and lacks certain Rust features, using neverthrow and other strategies like generator functions can improve error handling. They briefly mention an alternative library, Effect, which offers advanced features but comes with a steep learning curve and potential trade-offs, ultimately choosing to stick with neverthrow for its balance of functionality and simplicity.
Mar 09, 2026
870 words in the original blog post.