Why is Capybara unable to see the contents of this page?

Question

I'm trying to write an Rspec request spec using Capybara. It seems that everything is working correctly except that Capybara sees the page as blank.

Good signs:

• The page loads fine in a browser
• Tailing the log shows that the correct view is rendered during the test run
• Any logging statements I put in the views are executed
• If I use assert_select "h1", :text => "hello world", the test passes.

Bad signs:

• If I use page.should have_content('hello world'), it fails, saying Capybara::ElementNotFound: Unable to find xpath "/html"
• If I do $stdout.puts page.html, it's empty except for a doctype

My test looks something like this:

describe "working with foos" do
  it "should have a 'new foo' form" do
    get '/foos/new'
    assert_select 'h1', text: 'hello world' # passes
    page.should have_content('hello world') # fails
    $stdout.puts page.html                  # empty except for a doctype
  end
end

What could be wrong here?

Answer 1

The forehead-slapping answer

... which I'm sharing to save others the same pain.

Capybara uses the visit method to set up its page variable.

visit '/assembly/manage/task_lists/new'

Don't use get with Capybara:

get '/assembly/manage/task_lists/new'