One year of Dioxus fullstack in production: the good, the painful

After a year running Dioxus fullstack in production, a solo SaaS developer finds that Rust’s main benefit is not raw coding speed but a calmer, more coherent workflow. A single Rust workspace handles server rendering, backend functions, and interactive WebAssembly clients; shared types connect database results to server responses and UI components, making changes easier to check and refactor. The same component model could also support desktop or mobile apps. In production, the system has been dependable, and compile-time checks prevent many errors from reaching users. The experience spans SeggWat, an in-app feedback service with a sizable Dioxus dashboard, and infra.page.

The serious cost is compilation. Building both server and WebAssembly targets, across a dozen Dioxus-dependent crates and a macro-heavy interface, makes the browser build take about 22 minutes. A release image takes roughly 90 minutes on GitHub’s two-core runners, while a self-hosted runner cuts that to around 40; linting also exceeded hosted memory limits. Caching, combined test-and-build workflows, and a faster linker helped only partially, and releases now build less often. “All Rust” also has exceptions: charts use Chart.js and JavaScript glue, while Tailwind requires Bun. The developer would still choose Dioxus for app-like products needing complex client state, offline use, or mobile support, but considers its WebAssembly cost excessive for ordinary dashboards. The next project will test experimental Topcoat, which aims to preserve typed, shared Rust code while avoiding a WebAssembly bundle; whether it improves build times remains unmeasured.