Implement or design a promise pool that runs at most N async tasks at a time.

Short Interview Answer

Keep a queue of task factories, start up to N, and start the next when one finishes; return results in input order if required.

Detailed Explanation

APIs often need bounded concurrency for rate limits and memory. Pattern: map items to async functions, maintain `active` count, dequeue when slots free. Preserve order with result index array. Decide error policy: fail fast (cancel remaining) vs collect errors. Use AbortSignal for cancellation. Libraries exist (p-limit, p-map), but interviews may ask for a sketch. Discuss fairness and starvation under mixed task durations.

Example

async function pool(items, worker, n = 4) {
  const ret = new Array(items.length);
  let i = 0;
  async function run() {
    while (i < items.length) {
      const idx = i++;
      ret[idx] = await worker(items[idx], idx);
    }
  }
  await Promise.all(Array.from({ length: Math.min(n, items.length) }, run));
  return ret;
}

Interview Tip

Clarify ordered results vs completion order up front.

Common Mistake

Starting all promises then only awaiting slices — does not limit start concurrency.