A peer-to-peer, nearly-server-less, rich-client web app library using WebRTC.
Want to see pretty graphs? Log in now!
npm install nodetron
|4||downloads in the last week|
|14||downloads in the last month|
|Last Published By|
|Version||0.0.12 last updated 9 months ago|
|Keywords||webrtc, framework, library, p2p, peer, dht, worker, indexeddb|
|Dependencies||mongoose, restify, socket.io, generator-karma, generator-mocha|
This project is under heavy development. Its API is subject to major, breaking changes.
A peer-to-peer, rich-client web app library that uses HTML5 WebRTC and Web Workers to reduce reliance on central servers and enable greater decentralization by routing requests directly to peers.
Nodetron helps you easily create peer-to-peer applications that do more than just send chats or transfer files. Nodetron consists of a routing and discovery system with which developers can build rich client-side apps that control all user data locally. With Nodetron, all data validation and access permissions are delegated to and handled by the client. Developers are encouraged to then take advantage of IndexedDB, AppCache, and other HTML 5 APIs. In the future, Nodetron will feature tighter integration with those kinds of rich-client APIs.
In more detail:
- Clients store all of their own data.
- Clients make certain pieces of data public available so that other clients can discover them. This data can be stored by other clients and central servers.
- Clients discover other clients by querying a local cache, other clients, or a central server.
- the current implementation only allows for querying a central server. This will change in the future.
- Clients access another client's non-public data by directly sending requests to that other client, with associated metadata.
- Clients handle those requests and determine their response. Clients are responsible for authentication of requests. Nodetron will include authentication mechanisms in the future.
Technologies used: WebRTC, socket.io, MongoDB, node-restify.
Note: only recent versions of Chrome and Firefox are currently supported
- On the client side: Include the
nodetron.min.js) scripts manually or use
bower install nodetron.
- On the server side:
npm install nodetron.
npm install && bower installin the project root.
git submodule init && git submodule update --remote
--remoterequires git version >=1.8
- Since this project is under continuous development, you may need to re-run the above the commands after pulling the lastest upstream changes.
- To build the project (the client-side portion), run
- To run the server, install MongoDB. If you're on a Mac, use homebrew!
- For debugging, install global dependencies:
npm -g install node-inspector
- When working inside of the submodules of this project, make sure to checkout the correct branch.
If you intend to run the
demoapp, be sure to follow the submodule README instructions.
grunt allto run the server and the
democlient-side app, with livereload.
:<number>to specify how many demo client-side instances you want to run.
:debugto run node-inspector as well.
grunt:4:debugruns 4 demo client instances at ports 9000-9004 and runs node-inspector.
grunt clientto serve the demo client-side app, with livereload.
:unitto run unit tests.
grunt serverto run the server, with reloading.
:unitto run unit tests.
:debugto run node-inspector.
grunt e2eto run client e2e tests on Chrome (not implemented yet).
grunt crossto run client e2e tests on Firefox as well as Chrome (not implemented yet).
If you get a
Error: Cannot find module './build/Debug/DTraceProviderBindings' error, remove
restify from your local
node_modules folder and re-run
npm install in the project root.
- Workflow instructions at https://github.com/bchu/nodetron/wiki/Workflow
- Web Workers: IE10+
- IndexedDB: IE10+, should use moz/webkit prefixes, FF10+, CH23+, no Safari
- WebSQL shim for Safari and mobile
- WebRTC: no IE, FF20+, CH26+, no Safari.
Current Prototype Implementation
Clients send 'discovery queries' to a central server/database. The server responds with potential matches. Clients then contact the matches directly over a WebRTC connection that is brokered by the server. Clients then freely exchange information. Clients specify what information they make publicly available for discovery on the central server. Clients are also responsible for granting or denying access over WebRTC.
- Privacy - outside parties cannot inspect private info
- Data access redundancy - if central server goes down you can still exchange data with others.
- User discovery redundancy - if central server goes down you can still discover other clients.
- Client must be active (browser window open) to send/receive requests.
- Greater decentralization exposes the peer network (as a whole) to malicious attacks.
The client-side WebRTC code includes a forked version of PeerJS (https://github.com/peers/peerjs).
The Nodetron server code is also a heavily modified fork of PeerJS's PeerServer (https://github.com/peers/peerjs-server). Thanks to their great work!