@seclink: Classic pitfall of sccache and incremental being mutually exclusive: enabling both results in 20GB of incremental cache yet dependencies being recompiled every time. After turning off incremental: large third-party libraries (tokio/serde/axum/reqwest, etc.) compile once, across…
Summary
Explains the mutual exclusivity issue between sccache and incremental compilation: enabling both leads to 20GB incremental cache consumption and dependencies being recompiled each time. After disabling incremental, large third-party libraries compile once and are reused across projects, achieving a cache hit rate of 53–93% after the first compilation.
View Cached Full Text
Cached at: 07/30/26, 07:59 PM
Classic pitfall: sccache and incremental are mutually exclusive
When both are enabled, the result is a 20GB incremental cache, yet dependencies get recompiled every time. After turning off incremental:
- Large third-party libraries (tokio, serde, axum, reqwest, etc.) are compiled once and reused across projects and across
cargo build/cargo check/cargo testcommands. - After the first compilation, the hit rate reaches 53–93% (during verification we observed 562 Rust cache hits).
- Your other Rust projects also share the same dependency cache.
Similar Articles
Typst: Designing for Incrementality
Typst uses constrained memoization (comemo) and pure function design to make the language and compiler work together, achieving efficient incremental compilation and real-time preview. The article details the design ideas of layout caching, module evaluation memoization, function purity, and the introspection system.
@billtheinvestor: Give Claude Code and Codex infinite memory, programming efficiency improved by 92%! The Agentmemory tool has quickly gained 4000+ stars on GitHub and is completely free. It saves all information from your coding sessions through smart compression, and automatically extracts relevant context in future sessions, avoiding re...
Agentmemory is an open-source tool that provides infinite memory for Claude Code and Codex, reducing token usage through intelligent compression, improving programming efficiency, and has gained 4000+ stars on GitHub.
@vintcessun: Found a new approach: coding agent can be this lightweight. 16MB memory, 0% idle CPU, 26MB binary — these numbers are insane in the agent space. It's built with Rust, packing all standard tools, permission system, session management, and MCP into 17k LoC...
Zerostack is a minimal coding agent written in Rust, consuming only 16MB of memory and a 26MB binary. It achieves lightweight design through feature gates, supporting multiple providers and tools.
@AriXZone: The Linux kernel, a very large project with 28 million lines of code and 75,000 files, takes only about 3 minutes to index, with sub-millisecond query responses. codebase-memory-mcp is an open-source MCP server from DeusData that provides AI coding assistants (Claude Cod…
DeusData has open-sourced an MCP server called codebase-memory-mcp, which pre-parses code repositories into persistent knowledge graphs, providing AI coding assistants with sub-millisecond code structure query capabilities. It supports 158 languages, reduces token consumption by approximately 99% compared to traditional grep, and has extremely fast indexing speed.
RaReCache: Bridging the Gap in Cross-Model KV Cache Reuse via Rank disagreement-based Selective Recomputation
RaReCache 提出一种基于秩分歧（rank disagreement）的选择性重计算框架，使大型模型能够直接复用小型模型预填充的 KV 缓存，在大幅缩小参数差距的同时保留 95-99% 的目标精度，并实现最高 3.04 倍的预填充加速。