Skip to content
This repository has been archived by the owner on Oct 1, 2024. It is now read-only.

Latest commit

 

History

History

semaphore

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 

@shopify/semaphore

Caution

@shopify/semaphore is deprecated.

Shopifolk, see Shopify/quilt-internal for information on the latest packages available for use internally.

Build Status Build Status License: MIT npm version npm bundle size (minified + gzip)

The Semaphore class implements a counting semaphore. It can be useful to control concurrent access to a pool of resources such as:

  1. Maintaining a pool of web workers to run background scripts
  2. Limiting the number of concurrent requests that can be made to an API endpoint

In more concrete terms, if we take a semaphore with count 3 as an example, the first 3 calls to acquire a permit will resolve immediately and the 4th call will only be resolved when one of the earlier permits is released.

A real-life anology is parking spots at a parking lot. If the parking lot has a capacity for 10 cars, the first 10 cars to arrive will immediately park, but an 11th car will have to wait for one of the cars to leave so that a parking spot is available.

Installation

yarn add @shopify/semaphore

Usage

Instantiation

Create a semaphore instance by calling the Semaphore constructor with a count argument:

const semaphore = new Semaphore(3);

If you need a lock/mutex, a semaphore with a count of 1 will effectively act as one:

const mutex = new Semaphore(1);

Acquiring permits

Call .acquire() on a Semaphore instance to acquire a permit. The result is a promise that gets resolved with a Permit instance when a permit is available:

const permit = semaphore.acquire();

Releasing permits

Call the .release() method on a Permit instance to release it:

permit.release();

The .release)() method returns a promise that gets resolved when the permit is released and an earlier permit request that had been pending is resolved. Waiting on the resolution of the .release() is optional and could be useful in situations where you're having timing issues (e.g. in unit tests that utilize a Semaphore instance):

await permit.release();

Example

const MAX_SIMULTANEOUS_FETCHES = 2;

const fetchSemaphore = new Semaphore(MAX_SIMULTANEOUS_FETCHES);

async function callApi(path) {
  const permit = await fetchSemaphore.acquire();

  return fetch(path).finally(() => permit.release());
}

callApi(apples).then(renderApples);
callApi(oranges).then(renderOranges);
// The next acquire call won't resolve until one of the earlier permits is released
callApi(bananas).then(renderBananas);