As I moved the code to github this week, I wanted to give a quick overview of the project and its technical underlyings, just in case someone might be interested :)

As the title of this post might already suggest, its about selecting and layering loops from audiofiles hosted on soundcloud.com and thereby creating new music.

Those collections of layered loops are called a session. Sessions can be created anonymously by everyone and are auto-saved on every action. By distributing the URL of a session other users can join and every action is synced between all users in realtime.
(complete mayhem, indeed :)

The application is built with HTML5 and JS w/ jQuery using the soundcloud API on the client side and runs on node.js & mongoDB with express, socket.io and mongoose on the server side.

I basically started the project to find out if you could play multiple loops asynchronous and synced with low latency using the Web Audio API while collaboratively working together with multiple users in realtime. 

As that turned out to be not only possible but performed real nicely, the experiment then developed in somewhat of an application and finally got a name and a face.

There’s a whole lot of things i’d like to implement / refactor and do when i get some time on my hands, as

– adding a “record” feature to record, save and export created sessions as audiofiles

– rewriting the client side with backbone.js (as things got a bit messy …)

– adding some audio effects to the tracks

– write some tests

I’d appreciate any form of contribution as much as any questions or feedback on the project :)

Feel free to check out the live version at  http://mashcloud.net/ or the sources at https://github.com/gherkins/mashcloud

Starting with Apache and mod_proxy, i ended up with a plain node solution, which i really like.

Let’s take a quick look on some different approaches out there:

—1—

Using apache on port 80 as a proxy

ProxyPass /nodeurls/ http://localhost:9000/
ProxyPassReverse /nodeurls/ http://localhost:9000/

via stackoverflow

— no websockets
++ probably the easiest way to integrate with your running AMPP-stack

—2—

Using a node.js app on port 80 as a Wrapper for other node apps.

express.createServer()
  .use(express.vhost('hostname1.com', require('/path/to/hostname1').app)
  .use(express.vhost('hostname2.com', require('/path/to/hostname2').app)
.listen(80)

via stackoverflow

++ you can use websockets on port 80
— apps crash/restart/stop globally
–what about your apache or the like?

—3—

Using node.js with node-http-proxy on port 80

var http = require('http')
, httpProxy = require('http-proxy');

httpProxy.createServer({
  hostnameOnly: true,
  router: {
    //web-development.cc
    'www.my-domain.com': '127.0.0.1:3001',
    'www.my-other-domain.de' : '127.0.0.1:3002'
  }
}).listen(80);

++ proxy websockets to any port
— you might need to move your old web server to another port

The really cool thing about using node-http-proxy is its capability of proxying websockets.
So you can have your apps running independtly on different ports while serving everything to the user over port 80 and use stuff like socket.io.

Since i’m new to node.js and miles away from beeing a admin, any feedback is highly appreciated :)

Given you have a string which contains invalid unicode like:

This will trouble the browser and the socket connection.

If you prepare your json with PHP and  json_encode the Unicode will be escaped to some strings like these:

\ud83d\ude31\ud83d\ude31\ud83d\ude04\ud83d\ude04\ud83d\udc9c\ud83d\udc9c\ud83d\udc4a

But on clientside it will still result in invalid Unicode.

So after a lot of recherche i found this and used the decodeUnicodeString function from Zend to convert the escaped Unicode characters again to their unescaped representation. (dont forget the replacement as described in the post at stackoverflow if you extract it from Zend)
If the character is invalid it will be replaced by a question mark ‘?’.
This is at least what i asume.
The ? went through fine and the socket connection stays alive.

Ok the downside is you loose weird signs for an ‘?’, but i can live with that.
But im not really convinced with that solution, so if anybody knows another way to detect invalid Unicode sequences, let me know!