
Rust 1.90 Upgrade Crash Report: My Codebase Didn't Make It
Performance tanked after upgrading to Rust 1.90? Here’s why and how to claw it back.
16 posts

Performance tanked after upgrading to Rust 1.90? Here’s why and how to claw it back.
Why does Rust async work on Windows? Unveiling Microsoft’s undocumented AFD driver with real-world analogies to understand Windows async programming under the hood

A panoramic Rust Macros 2.0 guide: unify macro namespaces, master the TokenStream pipeline and macro hygiene, and migrate from macro 1.0 to macro 2.0 metaprogramming with practical examples.

If Rust build times have ever ruined your coding flow, this release fixes what matters most

When traffic surges like a flood, equip your Rust microservices with four gates—rate limiting, backpressure, batching, and middleware—using Tokio + Tower. Keep P99 steady and make speed happen in order.
Build a production-ready async microservice with Rust and Axum: job queue, background workers, exponential backoff retries, and graceful shutdown. Includes full project structure and Docker example.

Master graceful shutdown for Rust Axum services: listen for SIGINT/SIGTERM, refuse new connections with with_graceful_shutdown, stop background tasks via CancellationToken, set timeouts, and integrate with Docker/Kubernetes termination — protect data integrity and user experience.

A step‑by‑step Axum guide to build a modern high‑performance Web service: routing, path and query parameters, JSON, shared state, and middleware (tracing, CORS) with complete code samples and best practices.
Don’t treat unsafe like a get-out-of-jail-free card. With practical examples, this article dissects 6 common unsafe pitfalls: silencing the compiler with duct tape, dereferencing random raw pointers, violating borrow rules inside unsafe, reinventing wheels, abusing unsafe impl, and skipping Safety comments — helping you build the right risk mindset and usage patterns.

tokio::spawn is not a thread but a lightweight task. Under Tokio’s scheduler you build highly concurrent services with sleep, timeout and select — 1000 tasks in 2 seconds with negligible overhead.