Rack: First Principles

Posted by Ilija Eftimov on June 16, 2015

About three years ago, when I started working with Ruby and Rails, I noticed that the term “Rack” always came up in my Google searches. Overwhelmed with all of the stuff I needed to learn combined with the awesomeness of Rails, which shields the new Rails devs from it’s internals, I never really understood Rack or writing Rack apps. Although I used to see people mentioning “middleware” or “Rack middleware” I never really wrote (or tried to write) any middleware. But, you know what’s really funny?

As a Rails developer you use Rack every single day, it’s just that you’re not aware of it. So, for anyone that is wondering what Rack does and how you can leverage it’s awesomeness, lets take a look at Rack from first principles.

What is Rack?

Wait, isn’t it Rake, not Rack? Nope, Rake is “Make for Ruby” and it’s basically a task runner. On the other hand Rack is something else.

Taken from Rack’s documentation:

Rack provides a minimal, modular, and adaptable interface for developing web applications in Ruby. By wrapping HTTP requests and responses in the simplest way possible, it unifies and distills the API for web servers, web frameworks, and software in between (the so-called middleware) into a single method call.

What does this tell us? It wraps HTTP requests and responses. Okay, so you can actually use the request (as an object) and write/create the response which the Rack app will return. Cool? Next, has an API for servers, frameworks and middleware into a single method call.

A single method call?

Yes, a Rack application is an object that responds to a method called call.

Taken from Rack’s specification:

A Rack application is a Ruby object (not a class) that responds to call. It takes exactly one argument, the environment and returns an Array of exactly three values: The status, the headers, and the body.

What’s the environment? Well basically, it’s the environment of the server, because Rack is an interface (or an API) for web servers. More specifically, it’s a Hash that has a lot of headers (key-value pairs). Also, it contains some Rack specific variables. You can explore the environment here.

Why does it return exactly three values? That’s what always the web clients are interested in when receiving the response from the server:

  • the status - did the request fail, or was it ok
  • the headers - HTTP headers. Rarely mentioned, but they are great!
  • the body - the useful stuff that the server returns (usually a HTTP page or JSON payload)

Makes sense, right?

Rack application

Let’s see how we can create a simple Rack application.

First, we need to install Rack. Believe it or not (sarcasm sign!) Rack is a gem and we can install it by running this command in our command line:

gem install rack

After you have Rack installed, you can use Rack in two ways. What’s interesting is that a Proc can be a Rack application. But why Proc? Well, if you remember, a Rack application is an object that responds to the call method. ;-)

# rack_app.rb
require 'rack'

rack_app = Proc.new do |env|
  [200, {'Content-Type' => 'application/json'}, ["{'response':'OK'}"]]

Rack::Handler::WEBrick.run rack_app

If we run this app in our command line:

ruby rack_app.rb

we can see that a webserver booted and it ran the Rack application. This is the output of my webserver (the output of yours will vary):

[2015-06-16 21:26:38] INFO  WEBrick 1.3.1
[2015-06-16 21:26:38] INFO  ruby 2.0.0 (2014-09-19) [x86_64-darwin13.4.0]
[2015-06-16 21:26:38] INFO  WEBrick::HTTPServer#start: pid=12649 port=8080

As you can see, Rack started a WEBrick webserver that is running the Rack application we wrote. This happens because WEBrick is built in into Rack, as a Handler, therefore the last line in the Ruby file is:

*** snip ***
Rack::Handler::WEBrick.run rack_app

Another way to run this applciation is using the rackup command. To use this command we’ll need to modify the application a bit:

# rack_app.ru
app = Proc.new do |env|
  [200, {'Content-Type' => 'application/json'}, ["{'response':'OK'}"]]

run app

Much simpler, but what happened? Well, the Proc object still stands in, but the rest of the stuff is gone. As you can see also the filename changed. Why?

First and foremost, when using the rackup command we need a “rackup” file, whose extension is .ru. Yes, just like Russia’s TLD! Also, we use the run method which basically tells rackup to call call on the object that run takes as a parameter (or, the app).

Booting the server and running the app is trivial, just like before:

rackup rack_app.ru

The output, again, is very similar:

[2015-06-16 21:37:09] INFO  WEBrick 1.3.1
[2015-06-16 21:37:09] INFO  ruby 2.0.0 (2014-09-19) [x86_64-darwin13.4.0]
[2015-06-16 21:37:09] INFO  WEBrick::HTTPServer#start: pid=14930 port=9292

Oh, and did I mention you can cURL the app or view the app running in the browser? WEBrick is a server, which means your Rack application is running on a webserver. In fact, notice the last line of the log above - it says that the server is running on port 9292. To see your Rack application in action, you can visit http://localhost:9292 in the browser. Or, you can curl!

This is what I got when i cURLed to localhost:9292:


That’s it. The server returned the response that the Rack application created. There you go - this is a very tiny functioning Rack application.


I hope this post helped you understand what Rack is and what it’s used for. Next time we will take a look at writing Rack middleware and integrating it into a Rails application. Until than, feel free to share your thoughts on Rack with me and the ways you use it. Also, I am looking for suggestions for the middleware example for the next post - what kind of middleware would you like to me to cover?

Update 1 (2015-06-19): Thanks to BlazeBoy for the Rack vs Rake suggestion. Thanks to André for noticing a typo.

comments powered by Disqus