Adding back the #wait_until to Capybara v2+ gem. Can probably be adapted for cucumber tests.

capybara_wait_until.rb

# add this file capybara_wait_until.rb to your /test directory

module Capybara
  class Session
    ##
    #
    # Retry executing the block until a truthy result is returned or the timeout time is exceeded
    #
    # @param [Integer] timeout   The amount of seconds to retry executing the given block
    #
    # this method was removed in Capybara v2 so adding it back if not already defined
    #
    unless defined?(wait_until)
      def wait_until(timeout = Capybara.default_wait_time)
        Capybara.send(:timeout, timeout, driver) { yield }
      end
    end
  end
end

# Use it this way in /test/integration_test_helper.rb:

require 'capybara_wait_until'
module ActionController
  class IntegrationTest
    def wait_for_ajax_to_complete
      # pass timeout in seconds if you need to override default_wait_time
      page.wait_until { page.evaluate_script('jQuery.active === 0') }
    end
  end
end

Why can't you use #synchronize? http://rubydoc.info/github/jnicklas/capybara/master/Capybara/Node/Base#synchronize-instance_method

When you use find, capybara will be actually waiting for the element to appear (by using synchronize behind the scenes), so there's little reason to use wait_for here.

Moreover, this couples your acceptance tests to your underlying implementation and will be a pain to deal with when you decide to use websockets there, or just remove ajax calls for some reason.

@josepjaume - can you give a code example on using synchronize? I get an error that the method doesn't exist.

When I try to call wait_until, I get:
class or module required for rescue clause

to be more specific:

Failure/Error: page.wait_until {
     TypeError:
       class or module required for rescue clause
     # ./spec/support/capybara/wait_until.rb:13:in `wait_until'