Jasmine Headless WebKit

Run your Jasmine specs at sonic boom speed!

Colored Output Jasmine is great. I love it. But running Jasmine when you need to test code that will run in a browser environment can be problematic and slow:

But there's a solution for fast, accurate browser-based testing. with a focus on continuous testing, using one of the most popular browser cores, and that dovetails perfectly into the Jasmine gem's already established protocols.

Enter jasmine-headless-webkit

jasmine-headless-webkit uses the QtWebKit widget to run your specs without needing to render a pixel. It's nearly as fast as running in a JavaScript engine like Node.js, and, since it's a real browser environment, all the modules you would normally use, like jQuery and Backbone.js, work without any modifications. If you write your tests correctly, they'll even work when running in the Jasmine gem's server with no changes to your code.

jasmine-headless-webkit also streamlines your workflow in other ways:

Is this for me?

That depends on what you need:

'round here, we focus on unit testing and mocking external interfaces. No using your app's views or routes, no hitting the app server to get resources, just mocking and stubbing the JavaScript code all by itself.

How do I use this wonderful toy?

You can use it standalone:

gem install jasmine-headless-webkit

Or you can use it with Bundler:

gem 'jasmine-headless-webkit'

However you install it, you'll get a jasmine-headless-webkit executable. You'll also need to set up your project to use the Jasmine gem:

gem install jasmine
jasmine init

Once you're good enough, you can make the spec/javascripts/support/jasmine.yml file yourself and skip the Pivotal Jasmine gem entirely. It's what the cool kids do.

What do I need to get it working?

Installation requires Qt 4.7. jasmine-headless-webkit has been tested in the following environments:

If it works in yours, leave me a message on GitHub or fork this site and add your setup.

Qt 4.7.X

The gem is compiled using qt4-qmake and you will need Qt 4.7.x or greater. The version you have installed should be detected correctly, and the appropriate message for installing Qt should be given if it's wrong. If it's not, please file a new issue!

Manually checking the Qt version

Test that qt4-qmake it is installed and verify your version. qmake --version

If you have the Qt 4.7.x or greater, you are ready to install jasmine-headless-webkit. QMake version 2.01a Using Qt version 4.7.2 in /usr/lib

If you receive a different message, you can install qt4-qmake using one of the following commands as root:

Ubuntu 11.04

sudo apt-get install libqt4-dev
sudo apt-get install qt4-qmake
sudo update-alternatives --config qmake # and select Qt 4's qmake

Ubuntu 9.10

Running sudo apt-get install libqt4-dev and sudo apt-get install qt4-qmake will install qt4, but it installs version 4.5.2, which will not be able to compile jasmine-headless-webkit, as it requires Qt 4.7.X or greater.

You will need to compile qt4-qmake from source Qt version 4.7.0. There are excellent directions on how to compile the source code. You will need to ensure Qt is exported to your $PATH before using qmake, as the source code will install to /usr/local/Trolltech/.

Mac OS X 10.6 & 10.7


sudo port install qt4-mac


brew install qt

(you may need to use --build-from-source on Lion)

My OS isn't on here!

capybara-webkit has the best instructions for installing Qt on various other systems that may not be covered here.

How does it work?

jasmine-headless-webkit generates a static HTML file that includes the Jasmine JavaScript library from the Jasmine gem, your application and spec files, and any helpers you may need. The runner then creates a WebKit widget that loads the HTML file, runs the tests, and grabs the results of the test to show back to you. Awesome!

jasmine-headless-webkit uses the same jasmine.yml file that the Jasmine gem uses to define where particular files for the testing process are located:

  - "**/*"
  - helpers/**/*
  - "**/*_spec.*"
src_dir: app/assets/javascripts
spec_dir: spec/javascripts

It also brings in the same copy of the Jasmine library that the Jasmine gem includes, so if you're testing in both environments, you're guaranteed to get the same results in your tests.

*.coffee in my jasmine.yml file?!

Yes, jasmine-headless-webkit will support *.coffee files in jasmine.yml, which the normal Jasmine server currently does not support out of the box. Once there's official support, you'll be able to easily switch between jasmine-headless-webkit and the Jasmine test server when you're using CoffeeScript. CoffeeScript files are compiled and injected into the generated HTML files.

Never done Jasmine in CoffeeScript? It looks like this:

describe 'Component', ->
  describe 'StorylineNode', ->
    model = null

    beforeEach ->
      model = new ComponentStorylineNode({id: 1})

    it 'should not be new', ->

...and it turns into this...

describe('Component', function() {
  return describe('StorylineNode', function() {
    var model;
    model = null;
    beforeEach(function() {
      return model = new ComponentStorylineNode({
        id: 1
    return it('should not be new', function() {
      return expect(model.isNew()).toEqual(false);

Server interaction

Since there's no Jasmine server running, there's no way to grab test files from the filesystem via Ajax. If you need to test server interaction, do one of the following:

Sprockets support

Nearly all of Sprockets is accessible to your test suite when using jasmine-headless-webkit. It's easier to list the parts that aren't accessible:

If any gems have vendor/assets/javascripts in their list of files, such as jquery-rails, those are put in the asset path along with the paths you define in src_dir:

- app/assets/javascripts
- vendor/assets/javascripts

Technically, spec_dir is in your asset path, too, but Jasmine's typical behavior of including helpers before spec_dir should give you all the include power you need for defining specs.

If you want to keep src_dir as a string for backwards compatibility, you can add additional asset paths with, you guessed it, asset_paths:

src_dir: app/assets/javascripts

- vendor/assets/javascripts

asset_paths are added to the Sprockets asset paths after src_dir.

In order for Sprockets support to work as intended, you should define your src_files and spec_files as such:

  - "**/*.*"
  - "**/*[Ss]pec.*"

This will include everything that Sprockets understands in all your src_dir and spec_dir paths. At that point, use Sprockets require statements to define the include order of your files. Using the --list option on the command line to list the load order of files, combined with the --runner-out option to write HTML runner files to a place where the browser can easily get to them, is very helpful when moving to a Sprockets-managed project.

JavaScript Templates are supported too, including haml-sprockets. Use them as you would any other JavaScript file, and ensure the load order is right, and the necessary code in the JST namespace will be created.

Since any gem with vendor/assets/javascripts is usable, that means Jasmine-specific gems are possible now. jasmine-spec-extras is the first such gem, which provides jasmine-jquery, sinon, and any other useful Jasmine helpers, vendored into the gem so you can easily include them into your project without having to manually manage them yourself:

#= require sinon
#= require backbone

describe "Spy thing", ->
  it 'should fire a callback', ->
    collection = new Backbone.Collection()
    spy = sinon.spy()
    collection.bind('reset', spy)

If you have to use ERB to inject information into the JavaScript or CoffeeScript files in your project, I recommend that you move those injections to a file that is included separately from the code, or include them in application.*.erb like this:

# File: app/assets/javascripts/application.coffee.erb

#= require 'jquery'
#= require 'my_library'

MyLibrary.root_url = <%= api_root_path %>

Sprockets support is still pretty new, so as myself and others discover the best way to set up code that can be used in both places, those practices will be outlined here.

Caching, caching, caching

jasmine-headless-webkit does two things that are CPU intensive (besides running tests): compiling CoffeeScript and analyzing spec files to get line number information for nicer spec failure messages (did I mention you get really nice spec failure messages with jasmine-headless-webkit, too?). These two operations are cached into the .jhw-cache/ folder from where the runner is executed. When this cache is combined with running tests continuously using Guard, runtime overhead is reduced to almost nothing.

Of course, being a cache, it takes time to warm up. The first time you run jasmine-headless-webkit on a big project, it can take several seconds to warm the cache. After that, enjoy an almost 20% speedup in runtime (tested on exactly one project's runtime, YMMV).

What else works?

alert() and confirm() work, though the latter always returns true. You should be mocking calls to confirm(), of course:

spyOn(window, 'confirm').andReturn(false)

console.log() also works. It uses one of three methods for serializing what you've provided:

If you need a heavy-weight object printer, you also have console.pp(), which uses Jasmine's built-in pretty-printer if available, and falls back to JSON.stringify() if it's not.

You also get an additional method, console.peek(), which calls console.log() with the provided parameter, then passes the parameter back along so you can continue to work with it. It's the equivalent of .tap { |o| p o } in Ruby.

Running the runner

jasmine-headless-webkit [ -c / --colors ]
                        [ --no-colors ]
                        [ --no-full-run ]
                        [ --keep ]
                        [ -l / --list ]
                        [ --report <report file> ]
                        [ --runner-out <html file> ]
                        [ --use-server ]
                        [ --server-port <port number> ]
                        [ -j / --jasmine-config <path to jasmine.yml> ]
                        [ --seed <random seed> ]
                        <spec files to run>

The runner will return one of three exit codes:

Setting default options

Much like RSpec, you can define the default options for each run of the runner. Place your global options into a ~/.jasmine-headless-webkit file and your per-project settings in a .jasmine-headless-webkit file at the root of the project.

Listng what files jasmine-headless-webkit will include

If your tests are not picking up a file you thought they should be, or they're being included in the wrong order, run with the -l flag to get a list of the files that jasmine-headless-webkit will include in the generated HTML file. Very handy for making sure your Sprockets requires are working correctly.

Coloring the output

jasmine-headless-webkit will not color output by default. This makes it easier to integrate with CI servers. If you want colored output, use the -c flag. With colored output, your tests will look like this:

Colored Output

If you have colors turned on globally, you can turn them off per-project or per-run with --no-colors.

Preserving compiled output on errors

CoffeeScript logic errors can be hard to track down. Keep the generated HTML files with the --keep flag and you'll get specrunner.$$.html files in your working directory.

Writing out a machine-readable report

Use the --report option to create a detailed report file:

ERROR||Uh oh||file.js:23

guard-jasmine-headless-webkit uses this for the Growl notifications. You can also use it in your own setups, to run specs remotely and stick the results into a CI system. You can use Jasmine::Headless::Report to interpret the file and transform the output.

Using a different jasmine.yml file

If for some reason you're not using the default path for a jasmine.yml file (which is spec/javascripts/support/jasmine.yml), you can provide that path with -j.

Randomizing the order of spec files

Spec files are shuffled into a random order before each run. This lets you find issues where spec files may depend on state established in prior executed spec files -- a bad thing. After each run, you'll get the random seed used to randomize the files:

Test ordering seed: --seed 1234

If you're getting weird results related to the particular order of a run of specs, pass that same seed value back in and get to work!

Running only certain spec files

By default, if no files are passed into jasmine-headless-webkit, all possible spec files in the spec_files definition will be run. You can limit the run to only certain files by passing those to jasmine-headless-webkit:

jasmine-headless-webkit spec/javascripts/models/node_viewer.coffee

Serving files from an HTTP server

The --use-server flag will start up a simple WEBrick server for serving files to the WebKit runner, as opposed to serving them from the filesystem. This also lets you test things like HTML5 history stuff and other things that require an http:// URL when loading files. Normally, the port this server runs on is random, but you can specify it with --server-port.

Filtered runs and full runs

Typically, targeted spec running is done by a tool like Guard, and the order of running goes like this:

Having your test running tool re-run jasmine-headless-webkit is fast, but there's still the cost of instantiating QtWebKit and Ruby with each run. Versions of jasmine-headless-webkit 0.3.0 and greater will do this for you, keeping the widget in memory and running Jasmine tests on first the filtered suite, and then the complete suite. The results you'll get are for the last run that's executed, which is typically what you want to know anyway. Newer versions of guard-jasmine-headless-webkit also support this behavior. This trims valuable seconds off of testing with every run, saving you enough time every day to run to the coffee shop and get some delicious brew!

If you don't want this behavior, pass in --no-full-run and filtered runs will be the only thing that runs when you request one.

Writing the HTML runner to another location

If you want to use the runner file in other places, use the --runner-out parameter with the name of the target file. The HTML produced uses the Jasmine HtmlReporter if not loaded in jasmine-headless-webkit, so you should be able to just open it in a browser and have it work.

If you always want the reporter written to a particular location, you can define that location in jasmine.yml:

runner_output: "runner.html"

Running the runner from a Ruby program

You can call the runner from Ruby:

require 'jasmine-headless-webkit'

status_code = Jasmine::Headless::Runner.run(
  :colors => false, 
    #=> true to get colors
  :remove_html_file => true, 
    #=> false to keep specrunners on failure
  :jasmine_config => 'spec/javascripts/support/jasmine.yml',
    #=> run a different config
  :report => false, 
    #=> filename if a report file should be written
  :full_run => true, 
    #=> false to not run a full run after a targeted run
  :files => ['file_one_spec.js', 'file_two_spec.coffee']
    #=> files to use for a targeted run, [] to run all

Automated testing during development

jasmine-headless-webkit works best when it's running all the time, re-running tests when you update the appropriate files. If you use Guard, install guard-jasmine-headless-webkit and run guard init jasmine-headless-webkit to add the necessary bits to your Guardfile to test a Rails 3.1 (or a well-structured Rails 3.0) app.


With Sprockets support now in jasmine-headless-webkit, there's less of a need for most users for guard-rails-assets, unless you really need to get at those ERB files in your project. guard-rails-assets is what you want to use in this case. It will watch your app's code for changes and rebuild your pipelined JS code, ready to be tested with jasmine-headless-webkit:

guard 'rails-assets' do

guard 'jasmine-headless-webkit' do

Jammit for JS templates

If you're still using Jammit it shove your JS templates into one file, you can use a Guard for that, too! guard-jammit provides Jammit watching support, but the current version (as of 2011-06-18) does not support some changes to Jammit's internals. Use my fork until that gets fixed.

guard 'jammit' do

guard 'jasmine-headless-webkit' do

Rake tasks

You can create a Rake task for your headless Jasmine specs:

require 'jasmine-headless-webkit'

Jasmine::Headless::Task.new('jasmine:headless') do |t|
  t.colors = true
  t.keep_on_error = true
  t.jasmine_config = 'this/is/the/path.yml'

If you've bundled jasmine-headless-webkit in with Rails, you'll also get a basic task for running your Jasmine specs. Be sure to include the gem in the development group so you get with a normal call to rake -T:

group :test, :development do
  gem 'jasmine-headless-webkit'
# rake -T

rake jasmine:headless     # Run Jasmine specs headlessly

This is the same as running jasmine-headless-webkit -c.

Continuous integration & testing using Xvfb

Since most continuous integration servers do not have a display, you will need to use Xvfb or virtual framebuffer Xserver for Version 11. If you elect not to use Xvfb, you will need to have a browser and graphical display to run jasmine-headless-webkit -c.

Reference: Xvfb Manpages

Install Xvfb

  sudo apt-get install xvfb

Resolve missing dependencies

To resolve missing dependencies, you will need to know what to install. $ Xvfb :99 -ac You will see a long list of warning messages:

 [dix] Could not init font path element /usr/share/fonts/X11/misc,
 removing from list!
 [dix] Could not init font path element /usr/share/fonts/X11/cyrillic,
 removing from list!
 [dix] Could not init font path element 
 /usr/share/fonts/X11/100dpi/:unscaled, removing from list!
 [dix] Could not init font path element
  /usr/share/fonts/X11/75dpi/:unscaled, removing from list!
 [dix] Could not init font path element 
 /usr/share/fonts/X11/Type1, removing from list!
 [dix] Could not init font path element 
 /usr/share/fonts/X11/100dpi, removing from list!
 [dix] Could not init font path element 
 /usr/share/fonts/X11/75dpi, removing from list!
 sh: /usr/bin/xkbcomp: not found
 (EE) Error compiling keymap (server-42)
 (EE) XKB: Couldn't compile keymap
 [config/dbus] couldn't take over org.x.config:
 (Connection ":1.74" is not allowed to
 own the service "org.x.config.display99" 
 due to security policies in the configuration file)    

Installing the following packages would resolve the above warning messages. Your missing packages may be different depending on the packages you have installed. sudo apt-get install x11-xkb-utils sudo apt-get install xfonts-100dpi xfonts-75dpi sudo apt-get install xfonts-scalable xfonts-cyrillic sudo apt-get install xserver-xorg-core

Once you have resolved these dependencies, you should see: [dix] Could not init font path element /usr/share/fonts/X11/misc, removing from list! [dix] Could not init font path element /usr/share/fonts/X11/cyrillic, removing from list! [dix] Could not init font path element /usr/share/fonts/X11/100dpi/:unscaled, removing from list! [dix] Could not init font path element /usr/share/fonts/X11/75dpi/:unscaled, removing from list! [dix] Could not init font path element /usr/share/fonts/X11/Type1, removing from list! [dix] Could not init font path element /usr/share/fonts/X11/100dpi, removing from list! [dix] Could not init font path element /usr/share/fonts/X11/75dpi, removing from list!

Run with Xvfb

...as a Rake task

 xvfb-run rake jasmine:headless 
 # ...or...
 xvfb-run jasmine-headless-webkit -c



First run Xvfb in the background:

Xvfb :0 -screen 0 1024x768x24 > /dev/null 2>&1 &

Then, set your DISPLAY to point at the Xvfb instance. Putting all this in your .bash_profile or equivalent startup script makes this a lot easier:

xdpyinfo -display :0 &>/dev/null && export DISPLAY=:0

See Paul Goscicki's post for more details on the setup. Thanks, Paul!


RubyMine may throw an error when running rake spec, you will need to provide a JavaScript runtime environment.

 rake aborted!
 Could not find a JavaScript runtime.
 See https://github.com/sstephenson/execjs
 for a list of available runtimes.

To resolve this problem, install and use the therubyracer gem, which is the embed V8 JavaScript interpreter into Ruby. Additionally, you can set the EXECJS_RUNTIME environment variable to a valid ExecJS runtime name.


I have a problem or helpful suggestion, good sir.

Here's what you can do:

Credits & License

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.