typed-scheduler

A Scheduler written in TypeScript for concurrency-limiting and rate-limiting

Docs | GitHub | npm | Yarn

Table of Contents

• Installing
  • npm
  • Yarn
• Importing
  • ES Module
  • CommonJS
  • Browser Global
  • AMD
• Usage
  • Example
  • priority
  • ready() and idle()
  • concurrency and rate
• API
• License

Installing

npm

$ npm i --save typed-scheduler

Yarn

$ yarn add typed-scheduler

Importing

typed-scheduler is pre-bundled using the UMD pattern to support the following module formats.

ES Module

import Scheduler from 'typed-scheduler'

CommonJS

const Scheduler = require('typed-scheduler')

Browser Global

<script src="https://unpkg.com/typed-scheduler"></script>

AMD

define(['typed-scheduler'], function (Scheduler) {

})

Usage

Example

// rate limit to 2 messages every second
// defaults to 3 priorities
const scheduler = new Scheduler({ concurrency: 2, rate: 1000 })

// queue 120 messages synchronously
// scheduler will throttle to 2 messages per second
for (let i = 1; i <= 120; i++) {
  // schedule with normal priority
  scheduler.scheduleNormal(console.log, `message ${i}`)
}

priority

The second argument of schedule() is used to set the priority class of the scheduled function. The scheduler will handle scheduled functions in FIFO order within each priority class, and higher priorities will always be handled before lower priorities. A lower value means a higher priority.

ready() and idle()

The output of this program demonstrates when each of these methods resolves.

import Scheduler from 'typed-scheduler'

function print (...args) {
  console.log(`${(Date.now() / 1000).toFixed(3)}s`, ...args)
}

const scheduler = new Scheduler({ concurrency: 1, rate: 1000, priorities: 3 })

for (let i = 0; i < 3; i++) {
  for (let j = 0; j < 3; j++) {
    // interleave priority classes
    scheduler.schedule(
      () => print('task', i * 3 + j),
      j
    )
  }

  scheduler.ready(i).then(
    () => print(`priority ${i} ready`)
  )
}

scheduler.idle().then(
  () => print('scheduler idle')
)

Output

0.079s task 0
1.103s task 3
2.109s task 6
2.109s priority 0 ready
3.113s task 1
4.119s task 4
5.119s task 7
5.119s priority 1 ready
6.123s task 2
7.123s task 5
8.129s task 8
8.129s priority 2 ready
9.135s scheduler idle

concurrency and rate

Note that the following schedulers do not behave identically.

// 2 tasks every second
new Scheduler({ concurrency: 2, rate: 1000 })
// 1 task every half second
new Scheduler({ concurrency: 1, rate: 500 })

Using the options { concurrency: 2, rate: 1000 }, the scheduler will execute two tasks without delay, then wait a full second after either completes before executing another task.

Using { concurrency: 1, rate: 500 } will only execute one task at a time, then wait for half a second after it completes before executing another task.

API

The complete reference API is available on GitHub Pages.

License

Copyright © 2019 Patrick Roberts

MIT License