I am working to convert some Selenium tests to use Selenium grid and am running into a problem.
The tests run fine using a browser on the local machine.
I created a new driver like this (in spec_helper.rb):
Capybara.register_driver :selenium_remote do |app|
caps = Selenium::WebDriver::Remote::Capabilities.chrome
Selenium::WebDriver.for(:remote,
:url => 'http://gridhub:4444/grid/driver',
:desired_capabilities => caps)
end
When I run the tests with the new driver the hub is contacted and a node loads the Chrome driver, everything looks ok on the grid side.
The problem is that on the machine initiating the tests I am getting the following exception:
Failure/Error: Unable to find matching line from backtrace
SystemStackError:
stack level too deep
# /home/.rvm/gems/ruby-1.9.3-p194/gems/capybara-2.0.1/lib/capybara.rb:298
I created this bare bones test, and I am still seeing the same exception just running it:
Capybara.default_driver = :selenium_remote
describe 'I can' do
it 'visit Google without throwing an error' do
visit 'http://www.google.com'
end
end
I am using Selenium server standalone 2.26.0, Capybara 2.0.1 and RSpec 2.9.0.
Any ideas what is causing this?
Well, with the help of this blog post (http://www.softr.li/blog/2012/11/08/capybara-selenium-rspec-vagrant-remotely) I got it working.
The above blog post was based on another blog post (http://www.without-brains.net/blog/2012/08/01/capybara-and-selenium-with-vagrant).
It looks like I was registering the Capybara Selenium web driver incorrectly.
Below is the code from the blog post that correctly registers the driver. Just using it as-is fixed my issue, then I modified it for my own use.
The code is also in a gist (https://gist.github.com/4038197#file-capybara_remote-rb).
Create a new Ruby file, add this code, and include it in your spec_helper.rb file.
# SELENIUM_SERVER is the IP address or hostname of the system running Selenium
# Server, this is used to determine where to connect to when using one of the
# selenium_remote_* drivers
SELENIUM_SERVER = "10.10.11.1"
# SELENIUM_APP_HOST is the IP address or hostname of this system (where the
# tests run against) as reachable for the SELENIUM_SERVER. This is used to set
# the Capybara.app_host when using one of the selenium_remote_* drivers
SELENIUM_APP_HOST = "10.10.11.2"
# CAPYBARA_DRIVER is the Capybara driver to use, this defaults to Selenium with
# Firefox
CAPYBARA_DRIVER = "selenium_remote_firefox"
# At this point, Capybara.default_driver is :rack_test, and
# Capybara.javascript_driver is :selenium. We can't run :selenium in the Vagrant box,
# so we set the javascript driver to :selenium_remote_firefox which we're going to
# configure.
Capybara.javascript_driver = :selenium_remote_firefox
RSpec.configure do |config|
config.before(:each) do
if selenium_remote?
Capybara.app_host = "http://#{SELENIUM_APP_HOST}:#{page.driver.rack_server.port}"
end
end
config.after(:each) do
Capybara.reset_sessions!
Capybara.use_default_driver
Capybara.app_host = nil
end
# Determines if a selenium_remote_* driver is being used
def selenium_remote?
!(Capybara.current_driver.to_s =~ /\Aselenium_remote/).nil?
end
end
# CapybaraDriverRegistrar is a helper class that enables you to easily register
# Capybara drivers
class CapybaraDriverRegistrar
# register a Selenium driver for the given browser to run on the localhost
def self.register_selenium_local_driver(browser)
Capybara.register_driver "selenium_#{browser}".to_sym do |app|
Capybara::Selenium::Driver.new(app, browser: browser)
end
end
# register a Selenium driver for the given browser to run with a Selenium
# Server on another host
def self.register_selenium_remote_driver(browser)
Capybara.register_driver "selenium_remote_#{browser}".to_sym do |app|
Capybara::Selenium::Driver.new(app, browser: :remote, url: "http://#{SELENIUM_SERVER}:4444/wd/hub", desired_capabilities: browser)
end
end
end
# Register various Selenium drivers
CapybaraDriverRegistrar.register_selenium_remote_driver(:firefox)
Now any RSpec example with the :js => true metadata will use this driver.