Orbital was built as an in-house demo for DISQUS to show a realtime representation of comments as they happen on the network. It was heavily inspired by Mozilla's Glow project, but was mostly rewritten in order to achieve true real-time. It was originally released to the public at PyCon 2012.
The code itself does not contain the firehose-like mechanism DISQUS uses to obtain the data from their API. It does, however, contain a script which allows you to send mock data to demonstrate how to create a publisher.
You can view the current version of orbital on the web at http://map.labs.disqus.com
You can install most requirements via Homebrew and PIP
brew install geoip libevent zeromq pip install -r requirements.txt
Install Maxmind's GeoIP city data files to /usr/share/GeoIP/
cp GeoIPCity.dat /usr/share/GeoIP/GeoIPCity.dat
Run the server
Configure haproxy (or something) to forward websockets on port 80 to port 7000, and setup something to serve site/.
Two example feeders are included. One explicitly takes your accounts DISQUS data and feeds it in through the production DISQUS API. For example, if you run three forums, it will continually stream data for all comments made on those forums through the server. The other feeder will simply send fake data, and can be used as example code to create your own custom data feeder.
To stream data from DISQUS you'll first need to configure the application. To do this visit the following URL, and register a new application:
Once you've created the application, make sure the Default Access (under "Settings") is set to "Read, Write, and Manage Forums". You'll need the Manage Forums bit explicitly, as otherwise you wont gain admin access to the forums you moderate, which means DISQUS won't send any private user data (such as IPs).
First you'll want to copy the default configuration file:
cp app.cfg.example app.cfg
Now open this file in your favorite editor and adjust the settings to match your API application. You'll specifically need the values for "Consumer Secret" and "Access Token".
Finally, run the feeder:
You'll see a lot of data when you first run it, as it gets the "last 100 posts" to start the stream.
The second stream which is included sends sample data which is useful for modeling and testing your frontend application. No configuration is required for this stream, and you can simply run it:
Map design and original concept from Mozilla's Glow project.