Web Workers#

Purpose#

Move expensive calculations off the main thread and keep input, scrolling, and animations responsive.

Procedure#

1. Confirm the bottleneck first: consider workers only if the task exceeds about 10ms, user input delay is obvious, FPS drops, or large file parsing/image processing/encryption calculation.
2. Choose a communication model: use postMessage for simple tasks, use Comlink for functional RPC, use Worker pool for a large number of parallel small tasks, and use SharedWorker for cross-Tab synchronization.
3. Design the message protocol: request/response types, error types, cancellation or timeout strategies must be clear.
4. Prioritize Transferable Objects when transferring large objects; don’t let structured cloning swallow up the benefits brought by Worker.
5. terminate() when a component or page is unloaded and verify main thread improvements through Profiler, Performance panel or FPS metrics.

Detailed reference#

• Load references/basic-worker.md when you need native Worker files, build tool imports, React lifecycle encapsulation and cleanup examples.
• Load references/advanced-workers.md when Transferable Objects, Comlink, Worker pool, SharedWorker, CSP and SharedArrayBuffer considerations are required.

Constraints#

• Cannot access DOM, window, document, alert() or synchronize localStorage within Worker.
• High frequency communication may be slower than main thread calculations; measure first then split.
• Worker files must be loaded from the same source and are subject to worker-src CSP restrictions.
• Large binary data is transferred first, and the original reference will become invalid after the transfer.
• Each Worker has an independent JS heap and must actively manage life cycle and memory.

Expected Output#

Computationally intensive operations are performed on a background thread, main thread input and scrolling resume smoothly, error and cancellation paths are clear, and workers are terminated on unloading without memory leaks.