Document Skill original
Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.
Bun Workers
Bun supports Web Workers and Node.js worker_threads for parallel execution.
Web Workers
Basic Usage
typescript
// main.ts
const worker = new Worker(new URL("./worker.ts", import.meta.url));
worker.postMessage({ type: "start", data: [1, 2, 3, 4, 5] });
worker.onmessage = (event) => {
console.log("Result:", event.data);
};
worker.onerror = (error) => {
console.error("Worker error:", error.message);
};
// worker.ts
self.onmessage = (event) => {
const { type, data } = event.data;
if (type === "start") {
const result = data.map((x) => x * 2);
self.postMessage(result);
}
};Worker with URL
typescript
// Import from file path
const worker = new Worker(new URL("./worker.ts", import.meta.url));
// Or with blob URL
const code = `
self.onmessage = (e) => {
self.postMessage(e.data * 2);
};
`;
const blob = new Blob([code], { type: "application/javascript" });
const worker = new Worker(URL.createObjectURL(blob));Transferable Objects
typescript
// main.ts
const buffer = new ArrayBuffer(1024 * 1024); // 1MB
const view = new Uint8Array(buffer);
view.fill(42);
// Transfer ownership (zero-copy)
worker.postMessage({ buffer }, [buffer]);
// buffer is now detached (empty)
// worker.ts
self.onmessage = (event) => {
const { buffer } = event.data;
const view = new Uint8Array(buffer);
// Process buffer...
// Transfer back
self.postMessage({ buffer }, [buffer]);
};Shared Memory
typescript
// main.ts
const shared = new SharedArrayBuffer(1024);
const view = new Int32Array(shared);
worker.postMessage({ shared });
// Both main and worker can access
Atomics.add(view, 0, 1);
// worker.ts
self.onmessage = (event) => {
const { shared } = event.data;
const view = new Int32Array(shared);
// Atomic operations for thread safety
Atomics.add(view, 0, 1);
Atomics.notify(view, 0);
};Node.js worker_threads
typescript
// main.ts
import { Worker, isMainThread, parentPort, workerData } from "worker_threads";
if (isMainThread) {
const worker = new Worker(import.meta.filename, {
workerData: { numbers: [1, 2, 3, 4, 5] },
});
worker.on("message", (result) => {
console.log("Result:", result);
});
worker.on("error", (err) => {
console.error("Error:", err);
});
worker.on("exit", (code) => {
console.log("Worker exited with code:", code);
});
} else {
// Worker code
const { numbers } = workerData;
const sum = numbers.reduce((a, b) => a + b, 0);
parentPort?.postMessage(sum);
}Worker Pool
typescript
// worker-pool.ts
import { Worker } from "worker_threads";
class WorkerPool {
private workers: Worker[] = [];
private queue: Array<{
task: any;
resolve: (value: any) => void;
reject: (err: Error) => void;
}> = [];
private activeWorkers = new Set<Worker>();
constructor(
private workerPath: string,
private poolSize: number
) {
for (let i = 0; i < poolSize; i++) {
this.addWorker();
}
}
private addWorker() {
const worker = new Worker(this.workerPath);
worker.on("message", (result) => {
this.activeWorkers.delete(worker);
this.processQueue();
});
worker.on("error", (err) => {
this.activeWorkers.delete(worker);
console.error("Worker error:", err);
});
this.workers.push(worker);
}
async execute(task: any): Promise<any> {
return new Promise((resolve, reject) => {
this.queue.push({ task, resolve, reject });
this.processQueue();
});
}
private processQueue() {
for (const worker of this.workers) {
if (!this.activeWorkers.has(worker) && this.queue.length > 0) {
const { task, resolve, reject } = this.queue.shift()!;
this.activeWorkers.add(worker);
worker.once("message", resolve);
worker.once("error", reject);
worker.postMessage(task);
}
}
}
terminate() {
this.workers.forEach((w) => w.terminate());
}
}
// Usage
const pool = new WorkerPool("./worker.ts", 4);
const results = await Promise.all([
pool.execute({ task: 1 }),
pool.execute({ task: 2 }),
pool.execute({ task: 3 }),
]);
pool.terminate();Patterns
CPU-Intensive Tasks
typescript
// main.ts
const worker = new Worker(new URL("./cpu-worker.ts", import.meta.url));
// Process large dataset
const data = Array.from({ length: 1000000 }, () => Math.random());
worker.postMessage({ type: "process", data });
worker.onmessage = (event) => {
if (event.data.type === "progress") {
console.log(`Progress: ${event.data.percent}%`);
} else if (event.data.type === "result") {
console.log("Done:", event.data.result);
}
};
// cpu-worker.ts
self.onmessage = (event) => {
const { type, data } = event.data;
if (type === "process") {
const chunkSize = 10000;
let result = 0;
for (let i = 0; i < data.length; i++) {
result += Math.sqrt(data[i]);
// Report progress
if (i % chunkSize === 0) {
self.postMessage({
type: "progress",
percent: Math.round((i / data.length) * 100),
});
}
}
self.postMessage({ type: "result", result });
}
};Parallel Map
typescript
async function parallelMap<T, R>(
items: T[],
fn: string, // Function name in worker
workerUrl: URL,
concurrency = 4
): Promise<R[]> {
const results: R[] = new Array(items.length);
const workers: Worker[] = [];
// Create workers
for (let i = 0; i < concurrency; i++) {
workers.push(new Worker(workerUrl));
}
// Process items
let nextIndex = 0;
const processNext = (worker: Worker): Promise<void> => {
return new Promise((resolve) => {
if (nextIndex >= items.length) {
resolve();
return;
}
const index = nextIndex++;
worker.postMessage({ fn, item: items[index], index });
worker.onmessage = (event) => {
results[event.data.index] = event.data.result;
processNext(worker).then(resolve);
};
});
};
await Promise.all(workers.map(processNext));
workers.forEach((w) => w.terminate());
return results;
}Message Channel
typescript
// Create channel for worker-to-worker communication
const channel = new MessageChannel();
const worker1 = new Worker(new URL("./worker1.ts", import.meta.url));
const worker2 = new Worker(new URL("./worker2.ts", import.meta.url));
// Give each worker a port
worker1.postMessage({ port: channel.port1 }, [channel.port1]);
worker2.postMessage({ port: channel.port2 }, [channel.port2]);
// worker1.ts
let port: MessagePort;
self.onmessage = (event) => {
if (event.data.port) {
port = event.data.port;
port.onmessage = (e) => console.log("From worker2:", e.data);
port.postMessage("Hello from worker1!");
}
};Error Handling
typescript
const worker = new Worker(new URL("./worker.ts", import.meta.url));
worker.onerror = (error) => {
console.error("Uncaught error in worker:", error.message);
error.preventDefault(); // Prevent bubbling
};
worker.onmessageerror = (event) => {
console.error("Message deserialization failed");
};
// In worker
self.onerror = (error) => {
self.postMessage({ type: "error", message: error.message });
};Termination
typescript
const worker = new Worker(new URL("./worker.ts", import.meta.url));
// Request graceful shutdown
worker.postMessage({ type: "shutdown" });
// Force terminate after timeout
setTimeout(() => {
worker.terminate();
}, 5000);
// In worker
self.onmessage = (event) => {
if (event.data.type === "shutdown") {
// Cleanup
self.close();
}
};Common Errors
| Error | Cause | Fix |
|---|---|---|
Worker not found | Wrong URL | Check worker file path |
Cannot serialize | Non-transferable data | Use transferable objects |
DataCloneError | Functions/DOM in message | Send only serializable data |
Worker terminated | Premature terminate | Check termination logic |
When to Load References
Load references/optimization.md when:
- Worker pool tuning
- Memory management
- Performance profiling
Load references/patterns.md when:
- Complex coordination
- Backpressure handling
- Error recovery
