async-patterns-guide

Native async fn in traits is supported in Rust 1.75+. Use `async fn` directly in trait definitions instead of the async-trait macro. This approach avoids dynamic dispatch overhead and is the modern standard for trait-based async code.

Migrate to native async fn when your Rust version supports it (1.75+) and you need static dispatch performance. Migration involves removing the #[async_trait] macro and rewriting trait methods as native async fn. This eliminates Box allocation and improves runtime efficiency.

Async closures are closures that return futures, enabling async behavior in inline functions. They simplify passing async logic to higher-order functions without defining separate trait implementations or trait objects.

Static dispatch uses concrete types and native async fn for monomorphization at compile time with zero runtime overhead. Dynamic dispatch uses trait objects and Box, adding indirection. Choose static dispatch for performance-critical code and dynamic dispatch when type flexibility is required.

Yes, trait objects work with async via the async-trait macro or by boxing futures. However, trait objects incur dynamic dispatch costs. For better performance, use native async fn in traits when possible to leverage static dispatch.

Benchmark your code before and after migration. Remove #[async_trait] macro calls, convert trait methods to native async fn, and eliminate Box allocation. Profiling often shows measurable latency reduction from eliminating dynamic dispatch overhead.