Rust Glancer LSP Server Reduces RAM Usage by Two Orders of Magnitude

Introduction to Rust Glancer

Rust Glancer is a newly introduced functional LSP server designed for the Rust programming language. Its primary distinction is its ability to operate with significantly less RAM, reportedly two orders of magnitude less, compared to other Rust LSP servers.

Addressing Memory Consumption in Rust Development

The high memory usage of existing Rust LSP servers, particularly when dealing with extensive dependency trees, has been a point of concern for developers. Rust Glancer aims to alleviate this by focusing on shallow analysis of dependencies, which constitute the majority of code in many projects but are rarely directly edited. This approach contrasts with more pedantic analysis models that process all code deeply.

Design Philosophy and Technical Considerations

The design of Rust Glancer prioritizes efficiency for the common use case of analyzing large codebases. While some existing tools excel at incremental parsing and DOM-mutation style refactorings, these are considered niche. Rust Glancer suggests that for the 99% use case of analyzing dependencies, a simpler AST structure, such as a list of arrays, is more effective. The author also notes that proc macros contribute significantly to memory bloat due to their code generation and execution during analysis.

Potential Optimizations and Future Directions

Further optimizations for Rust Glancer could include lazy processing of function bodies, indexing all items but only deeply analyzing the currently open file. This hybrid approach could combine the benefits of both shallow and deep analysis. The author also mentions a 'Sorbet trick' where metaprogramming is not run, but rather a plugin interface explains its effects, potentially reducing the need to execute complex code like `serde` during analysis.