I find I have the need to do a series of concurrent tasks, with a max concurrency, but also with a max total run time. After this time, whatever hasn't finished or started can be ignored.

I propose to add a new API

sema.cancel()

Which rejects all pending .acquire() calls. This way I can do the following

const timeout = setTimeout(() => sema.cancel(), 5000);
await Promise.all(jobs.map(async job => {
  try {
    sema.acquire();
  } catch (err) {
    if (err.code === 'CANCELLED') {
      return null
    }
    throw err
  }

  try {
    await doWork(job);
  } finally {
    sema.release();
  }
}));
clearTimeout(timeout);

sorry to interrupt. what does nr stands for, I thought it was num of resource, but then what is the capacity variable?

Whenever I import Sema or RateLimit.

i.e import { RateLimit, Sema } from 'async-sema';

I'm getting the following error:

Uncaught TypeError: Class extends value #<Object> is not a constructor or null

Is there any workaround for it?

Thanks.

🚨 You need to enable Continuous Integration on all branches of this repository. 🚨

To enable Greenkeeper, you need to make sure that a commit status is reported on all branches. This is required by Greenkeeper because it uses your CI build statuses to figure out when to notify you about breaking changes.

Since we didn’t receive a CI status on the greenkeeper/initial branch, it’s possible that you don’t have CI set up yet. We recommend using Travis CI, but Greenkeeper will work with every other CI service as well.

If you have already set up a CI for this repository, you might need to check how it’s configured. Make sure it is set to run on all new branches. If you don’t want it to run on absolutely every branch, you can whitelist branches starting with greenkeeper/.

Once you have installed and configured CI on this repository correctly, you’ll need to re-trigger Greenkeeper’s initial pull request. To do this, please delete the greenkeeper/initial branch in this repository, and then remove and re-add this repository to the Greenkeeper App’s white list on Github. You'll find this list on your repo or organization’s settings page, under Installed GitHub Apps.

Thanks for this code!

const lim = RateLimit(2, {
  timeUnit: (5000),
});

await Promise.all(
  images.map(async (image, index) => {

    await lim();

    const parsed = await _prepImage.call(this, image);

    await util.downloadImage(parsed.url, parsed.target);

  })
);

•  Cleanup
• Proper examples
• Readme

I'm trying to limit uploads to 10 simultaneously, and I'm doing it like this:

	const uploadSema = new Sema(10)
	for (const upload of uploads) {
		await uploadSema.acquire()
		uploadOne(upload).finally(() => uploadSema.release())
	}
	await uploadSema.drain()

export const queuedWork = async (items, fn, workers = 10) => {
	const sema = new Sema(workers)
	let threw = null
	for (const item of items) {
		if (threw) break
		await sema.acquire()
		// eslint-disable-next-line promise/catch-or-return
		Promise.resolve(fn(item))
			.catch(err => {
				threw = err
			})
			.finally(() => sema.release())
	}
	await sema.drain()
	if (threw) throw threw
}

test('queuedWork async', async () => {
	const out = []
	await queuedWork([1, 5, 7, 89, 2], async n => out.push(n), 2)
	expect(out).toEqual([1, 5, 7, 89, 2])
})

test('queuedWork sync', async () => {
	const out = []
	await queuedWork([1, 5, 7, 89, 2], n => out.push(n), 2)
	expect(out).toEqual([1, 5, 7, 89, 2])
})

test('queuedWork throws', async () => {
	const out = []
	await expect(
		queuedWork(
			[1, 2, 3, 4, 5],
			async n => {
				if (n === 2) throw new Error('meep')
				else out.push(n)
			},
			2
		)
	).rejects.toThrow('meep')
	expect(out).toEqual([1, 3])
})


test('queuedWork throws sync', async () => {
	const out = []
	await expect(
		queuedWork(
			[1, 2, 3, 4, 5],
			n => {
				if (n === 2) throw new Error('meep')
				else out.push(n)
			},
			2
		)
	).rejects.toThrow('meep')
	expect(out).toEqual([1])
})

I think adding the following utility function might simplify working with semaphores a lot

export class Sema {
    public runTask<T>(task: (token?: any) => Promise<T>): Promise<T> {
        return this.acquire().then(token =>
            Promise.resolve(token)
                .then(task)
                .finally(() => sem.release(token))
        );
    }
}

const sem = new Sema(3);

// Run many task in parallel limited by sem
for (let i = 0; i < 100; i++) {
    sem.runTask(async () => {
        // do some stuff
    });
}

// Alternatively use Promise.all and map
await Promise.all(items.map(item => sem.runTask(() => processItem(item))))

How should one best figure out how to meaningfully determine this value? The examples seem to not follow any strict pattern.

Can the readme be made to include a rough guideline or set of heuristics to determine this for a given type of application?

Hi all, currently typescript definition seems broken, typescript compiler complains about malformed type definition file (.d.ts) syntax.

maybe it's off topic, I love this package. so i've migrated async-sema package to typescript. i'd like to discuss about publishing typescript version of async-sema. should i publish typescript version of async-sema package as individual package? or just send PR to this repo? Please kindly give me your advice. Thanks :)

All examples are currently in an examples directory, but we should perhaps include the basic on on the README itself

I am using the simple implementation RateLimiter and I am using the lowest value of 1 (1 request / sec) but I still get 500 errors from the server which I am calling. so I wonder if its possible to further reduce the number of calls say 1 every 2 seconds?

When adding this to a project, my app no longer loads in IE 11. I can't find any polyfills which alleviate the problem.

Has anyone had success running in IE?

The nr value given to Sema() does not stop .release() from being called as many times as a user wants.

Example below shows an nr value of 1, but the user is able to obtain 2 concurrent uses.

const Sema = require("async-sema").Sema;

let i = 0;
const a = new Sema(1);
a.release();
a.acquire().then(() => {
	i++;
	console.log(`i = ${i}`);
	return a.acquire();
}).then(() => {
	i++;
	console.log(`i = ${i}`);
	return a.acquire();
});

Now the constructor requires an on empty object for default args:

new Sema(10, {})

while

new Sema(10)

would be optimal.

Looking at the source code and the examples, it seems to me that the library is using p() to signal/free up a resource, whereas v() is used to wait/"block" on the semaphore.

However, according to the definition of a semaphore, the uses of p() and v() are the other way around. Some resources to verify are:

• http://www.cs.ucsb.edu/~rich/class/cs170/notes/Semaphores/index.html
• https://www.student.cs.uwaterloo.ca/~cs343/documents/notes.pdf (counting semaphore, page 104)

Please let me know if I'm misunderstanding the use case here. :)