# faye-rails

faye-rails is a Ruby gem which handles embedding Faye's rack-based server into the rails stack and providing it with access to controllers and views based on bindings and observers.

[![Build Status](https://travis-ci.org/jamesotron/faye-rails.png?branch=master)](https://travis-ci.org/jamesotron/faye-rails)
[![Dependency Status](https://gemnasium.com/jamesotron/faye-rails.png)](https://gemnasium.com/jamesotron/faye-rails)
[![Code Climate](https://codeclimate.com/github/jamesotron/faye-rails.png)](https://codeclimate.com/github/jamesotron/faye-rails)

A *very* small demonstration app is available for your perusal [on Heroku](http://faye-rails-demo.herokuapp.com/). The source is [here on Github](https://github.com/jamesotron/faye-rails-demo).

# Heroku Add-on

If you're planning on running Faye on Heroku you're probably going to have a bad time.  Take a look at [MessageRocket](https://messagerocket.co/) as an alternative, and help support the author to maintain more great open source projects.

## Caveat

faye-rails is used in production by a lot of folks, none of whom are me.  As
I don't use faye-rails in my daily life I will not be responding to issues
unless they have a corresponding PR.  If you'd like to take over maintaining
this project then get in contact.

# Embedded server

Due to the limitations of most Rack-based web servers available Faye can only be run on Thin, however if you are using thin, then you can add as many Faye servers as you want to the Rails middleware stack like so:

```ruby
# application.rb
config.middleware.use FayeRails::Middleware, mount: '/faye', :timeout => 25
```

You can also pass a block to `faye_server` which will be executed in the context of the Faye server, thus you can call any methods on `Faye::RackAdapter` from within the block:

```ruby
config.middleware.use FayeRails::Middleware, mount: '/faye', :timeout => 25 do
  # can be defined anywhere, like app/faye_extensions/mock_extension.rb 
  class MockExtension
    def incoming(message, callback)
      callback.call(message)
    end
  end

  add_extension(MockExtension.new)
end
```

You can also do some rudimentary routing using the map method:

```ruby
config.middleware.use FayeRails::Middleware, mount: '/faye', :timeout => 25 do
  map '/widgets/**' => WidgetsController  
  map :default => :block
end
```

You can find more details on the #map method in the [rdoc](http://rubydoc.info/github/jamesotron/faye-rails/master/FayeRails/RackAdapter)

# Controller

faye-rails includes a controller for handling the binding between model events and channels with it's own DSL for managing channel-based events.

```ruby
class WidgetController < FayeRails::Controller
end
```

## Model observers

You can subscribe to changes in models using the controller's observer DSL:

```ruby
class WidgetController < FayeRails::Controller
  observe Widget, :after_create do |new_widget|
    WidgetController.publish('/widgets', new_widget.attributes)
  end
end
```

The available callbacks are derived from the ActiveRecord callback stack. See [ActiveRecord::Callbacks](http://api.rubyonrails.org/classes/ActiveRecord/Callbacks.html) for more information regarding the callback queue.

See the [rdoc](http://rubydoc.info/github/jamesotron/faye-rails/master/FayeRails/Controller.observe) for more information.

## Channel DSL

The controller DSL elegantly wraps channel-based aspects of the Faye API so that you can easily group code based on specific channels.

### Monitoring

You can make use of Faye's [monitoring API](http://faye.jcoglan.com/ruby/monitoring.html) by adding calls to `monitor` within the channel block. You are able to monitor `:subscribe`, `:unsubscribe` and `:publish` events. Blocks are executed within the context of a `FayeRails::Controller::Monitor` instance which will give you access to `#client_id`, `#channel` and `#data` (`#data` only having a value on `:publish` events).

```ruby
class WidgetController < FayeRails::Controller
  channel '/widgets' do
    monitor :subscribe do
      puts "Client #{client_id} subscribed to #{channel}."
    end
    monitor :unsubscribe do
      puts "Client #{client_id} unsubscribed from #{channel}."
    end
    monitor :publish do
      puts "Client #{client_id} published #{data.inspect} to #{channel}."
    end
  end
end
```

### Filtering

You can quickly and easily filter incoming and outgoing messages for your specific channel using the controller's filter API, which wraps Faye's [extensions API](http://faye.jcoglan.com/ruby/extensions.html) in a concise and channel-specific way.

```ruby
class WidgetController < FayeRails::Controller
  channel '/widgets' do
    filter :in do
      puts "Inbound message #{message}."
      pass
    end
  end
end
```

You can add filters for `:in`, `:out` and `:any`, which will allow you to filter messages entering the server, exiting the server or both. The block passed to the `filter` is executed in the context of a `FayeRails::Filter::DSL` instance, which gives you access to the `#message` method, which contains the entire message payload from the client (including meta information you wouldn't see other ways). You also have access to the `#pass`, `#modify`, `#block` and `#drop` methods which are sugar around Faye's callback system - which is accessible via the `#callback` method if you want to do it that way. Check out the [FayeRails::Filter::DSL rdoc](http://rubydoc.info/github/jamesotron/faye-rails/master/FayeRails/Filter/DSL) for more information.  Please note that all filters must call `callback.call` either via the sugar methods or directly to ensure that requests are not lost (not to mention potential memory leaks).

### Subscribing

You can easily subscribe to a channel using the 'subscribe' method inside your channel block, like so:

```ruby
class WidgetController < FayeRails::Controller
  channel '/widgets' do
    subscribe do
      puts "Received on channel #{channel}: #{message.inspect}"
    end
  end
end
```

# Non-server environments

Often you'll find yourself running the Rails environment without the server running - eg when doing background job processing, or running the console.  If you have any actions which use Faye then you'll need to make sure that you have the EventMachine reactor running.  The easiest solution to this is to create an initialiser in `config/initializers` which calls `Faye.ensure_reactor_running!`. For workers in production you probably also want to make sure that you are using the Redis engine for Faye to ensure that multiple server instances see the same data.

```ruby
config.middleware.use FayeRails::Middleware, mount: '/faye', engine: {type: Faye::Redis, host: 'localhost'}, :timeout => 25 do
  map '/announce/**' => SomeController  
end
```

See more details in [this issue on GitHub](https://github.com/jamesotron/faye-rails/issues/26).

# Running on Phusion Passenger

If you want to run faye-rails on passenger, make sure you are using passenger 4.0 standalone or passenger 4.0 on nginx 1.4+ for nginx with websocket support. Passenger on apache is not supported. Because passenger uses a multi-process model, you must use the faye redis backend. Add `gem 'faye-redis'` to your Gemfile and configure your routes like this:

```ruby
config.middleware.use FayeRails::Middleware, mount: '/faye', :timeout => 25, server: 'passenger', engine: {type: Faye::Redis, host: 'localhost'}
```

# Thanks.

Thanks to James Coglan for the excellent Faye Bayeux implementetation and great support for Faye users.