When building a Tauri app, the main thread handles the UI while Tokio manages async tasks. Blocking the main thread freezes the interface, and long-running commands can time out the frontend. To keep your app snappy—even on older hardware like an 8-year-old MacBook Air—you need smart async patterns. Below are five essential techniques, each explained in a Q&A format.
1. What is the golden rule for keeping the UI responsive in Tauri commands?
Never perform blocking work in a #[tauri::command] without async. If you call a heavy function directly, it blocks the thread pool and freezes the UI until it finishes. Instead, wrap CPU-intensive tasks inside tokio::task::spawn_blocking(), which moves them to a dedicated thread pool. This frees the async executor to handle other tasks, like UI events or other commands.
For example, a compress_pdf command that takes 3 seconds should be written as async fn compress_pdf(...) and use spawn_blocking to run the compression work. The frontend then stays responsive because Tauri's async runtime isn't tied up.
2. How can I show progress for long-running operations?
Use Tauri's event system to push progress updates from the backend to the frontend. In your command, emit an event like "batch-progress" with JSON payload containing current, total, and percent. On the frontend, listen for that event with await listen('batch-progress', ...) and update a progress bar or indicator.
In the Rust code, iterate over your items and after each one, call window.emit("batch-progress", payload). This lets users see real-time status like "3 of 10 processed (30%)". It's simple yet powerful—no polling required.
3. How do I let users cancel a long operation?
Implement a shared cancellation flag using Arc<AtomicBool>. Create a CancelToken struct that wraps the atomic boolean. Expose it as Tauri state so both the command and a cancel button's handler can access it.
In your command, check cancel_token.is_cancelled() inside the loop. If true, return an error (e.g., "cancelled"). The frontend can invoke a separate cancel_batch command that calls token.cancel(). This pattern is lightweight, thread-safe, and gives users control over long batch processes.
4. How can I process multiple files concurrently without overwhelming the system?
Use a tokio::sync::Semaphore to limit the number of concurrent tasks. Wrap it in an Arc and acquire a permit before starting each file. This prevents spawning hundreds of tasks that might exhaust memory or CPU.
For example, if you have 20 files but only want 4 running at once, set the semaphore's initial count to 4. Each spawned task awaits a permit, processes a file, and then releases it. The overall throughput improves without overloading older hardware.
5. What's the difference between spawn_blocking and regular async work?
Regular async functions (e.g., network requests) yield control automatically when waiting. But CPU-heavy calculations don't yield—they hog the runtime thread. spawn_blocking moves such work to a separate thread pool dedicated to blocking tasks, keeping the main async executor free for lightweight I/O and UI updates.
Use it for tasks like image processing, PDF compression, or file encryption. For I/O-bound work (database queries, HTTP calls), standard async is fine. The rule of thumb: if it takes more than a few milliseconds and doesn't do I/O, offload it with spawn_blocking.