discovery-swarm-web

Abstracts away discovery-swarm interaction with WebRTC and a websocket gateway.

npm install --save discovery-swarm-web

Example:

const DiscoverySwarmWeb = require('discovery-swarm-web')
const swarm = new DiscoverySwarmWeb({
  stream: () => archive.replicate(),
})

swarm.join(archive.discoveryKey)

How it works

Under the hood it uses @geut/discovery-swarm-webrtc to find WebRTC peers for channels, and uses discovery-swarm-stream over a websocket to bridge to the rest of the Dat network by using discovery-swarm along with the dat-swarm-defaults configuration.

A websocket gets created for signaling WebRTC peers, and another one gets created to proxy connections from the P2P network.

API

new DiscoverySwarmWeb(opts)

opts include:

• stream: The only required field. Return a stream to handle an incoming connection
• id: The ID you want to use to show up in the swarm
• signalhub: Either a URL for a signalhubws server used for WebRTC signaling, or an object that has the same interface as signalhub. Uses signalhubws.mauve.moe by default. Note that each signalhub server creates a new WebRTC swarm, so you probably shouldn't change this.
• discovery: A discovery-swarm-web server URl to connect to. By default it uses discoveryswarm.mauve.moe, please supply your own if you're deploying to production. All discovery servers reach out to the same P2P network.

swarm.join(key)

Starts looking for peers for the given key. This should be either a string or a Buffer. For Dat, you should use the discoveryKey of your hypercore or hyperdrive.

swarm.leave(key)

Stops looking for peers for the given key.

swarm.close(), swarm.destroy()

Close the swarm: closes all connections to peers, to the discovery server, and to the signalhub.

CLI

You can start your own gateway for added performance and privacy using the CLI, or if you want to host one for an application you're building. By default it will listen on TCP 3472 (DISC), but that can be changed with the --port flag.

npm install -g discovery-swarm-web

discovery-swarm-web --port 3472

Roadmap:

• Attempt to connect to a local discovery instance before the internet
• Generate ID if one is not provided
• Auto-reconnect logic to discover-swarm-stream websocket
• Emit the connection event
• Have limits for the number of connections made for a given channel (auto-close new connections after reaching it)
  • Resolve bugs with close event not firing
• Don't join discovery for a key unless you get no WebRTC peers (then join after a delay)
• Disconnect peers with known ID
• Make stream optional with a default handshake function