trystero

📯 Trystero

Serverless WebRTC matchmaking for painless P2P: make any site multiplayer in a few lines

👉 TRY THE DEMO 👈

Trystero manages a clandestine courier network that lets your application's users talk directly with one another, encrypted and without a server middleman.

Peers can connect via BitTorrent, Firebase, or IPFS – all using the same API.

---

• How it works
• Get started
• Listen for events
• Broadcast events
• Advanced
• API
• Strategy comparison
• Firebase setup

---

How it works

👉 If you just want to try out Trystero, you can skip this explainer and jump into using it.

To establish a direct peer-to-peer connection with WebRTC, a signalling channel is needed to exchange peer information (SDP). Typically this involves running your own matchmaking server but Trystero abstracts this away for you and offers multiple "serverless" strategies for connecting peers (currently BitTorrent, Firebase, and IPFS).

The important point to remember is this:

🔒

Beyond peer discovery, your app's data never touches the strategy medium and is sent directly peer-to-peer and end-to-end encrypted between users.

👆

You can compare strategies here.

Get started

You can install with npm (npm i trystero) and import like so:

import {joinRoom} from 'trystero'

Or maybe you prefer a simple script tag?

<script type="module">
  import {joinRoom} from 'https://cdn.skypack.dev/trystero'
</script>

By default, the BitTorrent strategy is used. To use a different one just deep import like so (your bundler should handle including only relevant code):

import {joinRoom} from 'trystero/firebase'
// or
import {joinRoom} from 'trystero/ipfs'

Next, join the user to a room with a namespace:

const config = {appId: 'san_narciso_3d'}
const room = joinRoom(config, 'yoyodyne')

The first argument is a configuration object that requires an appId. This should be a completely unique identifier for your app (for the BitTorrent and IPFS strategies) or your Firebase database ID if you're using Firebase. The second argument is the room name.

Why rooms? Browsers can only handle a limited amount of WebRTC connections at a time so it's recommended to design your app such that users are divided into groups (or rooms, or namespaces, or channels... whatever you'd like to call them).

Listen for events

Listen for peers joining the room:

room.onPeerJoin(id => console.log(`${id} joined`))

Listen for peers leaving the room:

room.onPeerLeave(id => console.log(`${id} left`))

Listen for peers sending their audio/video streams:

room.onPeerStream((stream, id) => (peerElements[id].video.srcObject = stream))

To unsubscribe from events, leave the room:

room.leave()

Broadcast events

Send peers your video stream:

room.addStream(
  await navigator.mediaDevices.getUserMedia({audio: true, video: true})
)

Send and subscribe to custom P2P actions:

const [sendDrink, getDrink] = room.makeAction('drink')

// buy drink for a friend
sendDrink({drink: 'negroni', withIce: true}, friendId)

// buy round for the house (second argument omitted)
sendDrink({drink: 'mezcal', withIce: false})

// listen for drinks sent to you
getDrink((data, id) =>
  console.log(
    `got a ${data.drink} with${data.withIce ? '' : 'out'} ice from ${id}`
  )
)

You can also use actions to send binary data, like images:

const [sendPic, getPic] = room.makeAction('pic')

// blobs are automatically handled, as are any form of TypedArray
canvas.toBlob(blob => sendPic(blob))

// binary data is received as raw ArrayBuffers so your handling code should
// interpret it in a way that makes sense
getPic((data, id) => (imgs[id].src = URL.createObjectURL(new Blob([data]))))

Let's say we want users to be able to name themselves:

const idsToNames = {}
const [sendName, getName] = room.makeAction('name')

// tell other peers currently in the room our name
sendName('Oedipa')

// tell newcomers
room.onPeerJoin(id => sendName('Oedipa', id))

// listen for peers naming themselves
getName((name, id) => (idsToNames[id] = name))

room.onPeerLeave(id =>
  console.log(`${idsToNames[id] || 'a weird stranger'} left`)
)

Actions are smart and handle serialization and chunking for you behind the scenes. This means you can send very large files and whatever data you send will be received on the other side as the same type (a number as a number, a string as a string, an object as an object, binary as binary, etc.).

Advanced

Binary metadata

Let's say your app supports sending various types of files and you want to annotate the raw bytes being sent with metadata about how they should be interpreted. Instead of manually adding metadata bytes to the buffer you can simply pass a metadata argument in the sender action for your binary payload:

const [sendFile, getFile] = makeAction('file')

getFile((data, id, meta) =>
  console.log(
    `got a file (${meta.name}) from ${id} with type ${meta.type}`,
    data
  )
)

// to send metadata, pass a third argument
// to broadcast to the whole room, set the second peer ID argument to null
sendFile(buffer, null, {name: 'The Courierʼs Tragedy', type: 'application/pdf'})

Action promises

Action sender functions return a promise that resolves when they're done sending. You can optionally use this to indicate to the user when a large transfer is done.

await sendFile(amplePayload)
console.log('done sending')

API

joinRoom(config, namespace)

Adds local user to room whereby other peers in the same namespace will open communication channels and send events.

• config - Configuration object containing the following keys:

  • appId - (required) A unique string identifying your app. If using Firebase this should be the database ID (also see firebaseApp below for an alternative way of configuring the Firebase strategy).

  • rtcConfig - (optional) Specifies a custom RTCConfiguration for all peer connections.

  • trackerUrls - (optional, 🌊 BitTorrent only) Custom list of torrent tracker URLs to use. They must support WebSocket connections.

  • trackerRedundancy - (optional, 🌊 BitTorrent only) Integer specifying how many torrent trackers to connect to simultaneously in case some fail. Defaults to 2, maximum of 4. Passing a trackerUrls option will cause this option to be ignored as the entire list will be used.

  • firebaseApp - (optional, 🔥 Firebase only) You can pass an already initialized Firebase app instance instead of an appId. Normally Trystero will initialize a Firebase app based on the appId but this will fail if youʼve already initialized it for use elsewhere.

  • rootPath - (optional, 🔥 Firebase only) String specifying path where Trystero writes its matchmaking data in your database ('__trystero__' by default). Changing this is useful if you want to run multiple apps using the same database and don't want to worry about namespace collisions.

  • swarmAddresses - (optional, 🪐 IPFS only) List of IPFS multiaddrs to be passed to config.Addresses.Swarm.

• namespace - A string to namespace peers and events within a room.

Returns an object with the following methods:

• leave()

  Remove local user from room and unsubscribe from room events.

• getPeers()

  Returns a list of peer IDs present in room (not including the local user).

• addStream(stream, [peerId], [metadata])

  Broadcasts media stream to other peers.

  • stream - A MediaStream with audio and/or video to send to peers in the room.

  • peerId - (optional) If specified, the stream is sent only to the target peer ID (string) or list of peer IDs (array).

  • metadata - (optional) Additional metadata (any serializable type) to be sent with the stream. This is useful when sending multiple streams so recipients know which is which (e.g. a webcam versus a screen capture). If you want to broadcast a stream to all peers in the room with a metadata argument, pass null as the second argument.

• removeStream(stream, [peerId])

  Stops sending previously sent media stream to other peers.

  • stream - A previously sent MediaStream to stop sending.

  • peerId - (optional) If specified, the stream is removed only from the target peer ID (string) or list of peer IDs (array).

• onPeerJoin(callback)

  Registers a callback function that will be called when a peer joins the room. If called more than once, only the latest callback registered is ever called.

  • callback(peerId) - Function to run whenever a peer joins, called with the peer's ID.

  Example:

  onPeerJoin(id => console.log(`${id} joined`))
  

• onPeerLeave(callback)

  Registers a callback function that will be called when a peer leaves the room. If called more than once, only the latest callback registered is ever called.

  • callback(peerId) - Function to run whenever a peer leaves, called with the peer's ID.

  Example:

  onPeerLeave(id => console.log(`${id} left`))
  

• onPeerStream(callback)

  Registers a callback function that will be called when a peer sends a media stream. If called more than once, only the latest callback registered is ever called.

  • callback(stream, peerId, metadata) - Function to run whenever a peer sends a media stream, called with the the peer's stream, ID, and optional metadata (see addStream() above for details).

  Example:

  onPeerStream((stream, id) => console.log(`got stream from ${id}`, stream))
  

• makeAction(namespace)

  Listen for and send custom data actions.

  • namespace - A string to register this action consistently among all peers.

  Returns a pair containing a function to send the action to peers and a function to register a listener. The sender function takes any JSON-serializable value (primitive or object) or binary data as its first argument and takes an optional second argument of a peer ID or a list of peer IDs to send to. By default it will broadcast the value to all peers in the room. If the sender function is called with binary data (Blob, TypedArray), it will be received on the other end as an ArrayBuffer of agnostic bytes. The sender function returns a promise that resolves when all target peers are finished receiving data.

  Example:

  const [sendCursor, getCursor] = room.makeAction('cursormove')
  
  window.addEventListener('mousemove', e => sendCursor([e.clientX, e.clientY]))
  
  getCursor(([x, y], id) => {
    const peerCursor = cursorMap[id]
    peerCursor.style.left = x + 'px'
    peerCursor.style.top = y + 'px'
  })
  

• ping(peerId)

  Takes a peer ID and returns a promise that resolves to the milliseconds the round-trip to that peer took. Use this for measuring latency.

  • peerId - Peer ID string of the target peer.

  Example:

  // log round-trip time every 2 seconds
  room.onPeerJoin(id =>
    setInterval(async () => console.log(`took ${await room.ping(id)}ms`), 2000)
  )
  

selfId

A unique ID string other peers will know the local user as globally across rooms.

getOccupants(config, namespace)

(🔥 Firebase only) Returns a promise that resolves to a list of user IDs present in the given namespace. This is useful for checking how many users are in a room without joining it.

• config - A configuration object
• namespace - A namespace string that you'd pass to joinRoom().

Example:

console.log((await trystero.getOccupants(config, 'the_scope')).length)
// => 3

Strategy comparison

Loose, (overly) simple advice for choosing a strategy: Use the BitTorrent or IPFS strategy for experiments or when your heart yearns for fuller decentralization, use Firebase for "production" apps where you need full control and reliability. IPFS is itself in alpha so the Trystero IPFS strategy should be considered experimental.

Trystero makes it trivial to switch between strategies – just change a single import line:

import {joinRoom} from 'trystero/[torrent|firebase|ipfs]'

	setup¹	reliability²	time to connect³	bundle size⁴	occupancy polling⁵
🌊 BitTorrent	none ✅	variable	better	~24K ✅	none
🔥 Firebase	~5 mins	reliable ✅	best ✅	~173K	yes ✅
🪐 IPFS	none ✅	variable	good	~1.65M 👀	none

¹ Firebase requires an account and project which take a few minutes to set up.

² Firebase has a 99.95% SLA. The BitTorrent strategy uses public trackers which may go down/misbehave at their own whim. Trystero has a built-in redundancy approach that connects to multiple trackers simultaneously to avoid issues. IPFS relies on public gateways which are also prone to downtime.

³ Relative speed of peers connecting to each other when joining a room. Firebase is near-instantaneous while the other strategies are a bit slower.

⁴ Calculated via Rollup bundling + Terser compression.

⁵ The Firebase strategy supports calling getOccupants() on a room to see which/how many users are currently present without joining the room.

Firebase setup

If you want to use the Firebase strategy and don't have an existing project:

1. Create a Firebase project
2. Create a new Realtime Database
3. Copy the database ID and use it as the appId in your Trystero config
4. [Optional] Configure the database with security rules to limit activity:

{
  "rules": {
    ".read": false,
    ".write": false,
    "__trystero__": {
      ".read": false,
      ".write": false,
      "$room_id": {
        ".read": true,
        ".write": true
      }
    }
  }
}

These rules ensure room peer presence is only readable if the room namespace is known ahead of time.

---

Trystero by Dan Motzenbecker