Vielzeug

Appearance

queue ​

Problem ​

You have multiple async jobs that must not all run at once — for example uploads or API calls that must be serialized or bounded to avoid overload.

Solution ​

Use queue({ concurrency }) to create a shared job queue. .add(fn) enqueues a task and returns its promise. .onIdle() resolves when all enqueued tasks complete.

ts
import { queue } from '@vielzeug/arsenal';

const q = queue({ concurrency: 2 });

const a = q.add(() => fetch('/api/a').then((r) => r.json()));
const b = q.add(() => fetch('/api/b').then((r) => r.json()));
const c = q.add(() => fetch('/api/c').then((r) => r.json()));

await q.onIdle(); // wait for all three to finish

const [dataA, dataB, dataC] = await Promise.all([a, b, c]);

Inspecting queue state ​

ts
import { queue } from '@vielzeug/arsenal';

const q = queue({ concurrency: 1 });

q.add(() => sleep(100));
q.add(() => sleep(100));

q.active; // tasks currently running
q.pending; // tasks waiting to start
q.size; // active + pending

q.clear(); // discard pending tasks (running tasks continue)

Pitfalls ​

  • concurrency defaults to 1 (serial). Pass a value greater than 1 for bounded parallelism.
  • .clear() discards pending tasks but does not cancel already-running ones.
  • .onIdle() only resolves once — it does not re-arm if new tasks are added after the queue drains.