Fix bug where routes helpers cannot be used in tests that use an anonymous controller.

[thomasfedb]

Changed orig_routes to @orig_routes as it is referred to in method_missing on line 138. Bug introduced in 86aea05.

I attempted to write a failing test, but it was beyond me how to do so. It does, however resolve the incorrectly failing test in my application.

[thomasfedb]

Note: I have made this change against the commit for 3.0.0.beta1 so the branch can be used as a straight replacement. The change should apply cleanly to master however.

[JonRowe]

If you can find a way to make this change that doesn't cause warnings to be issued (the cause of 86aea05) then I'll merge it, otherwise we need to find a way to fix this that doesnt generate warnings.

[thomasfedb]

This doesn't currently cause any warnings for me. (ruby 2.0.0p353)

[JonRowe]

I'll check it out later, but be mindful that we support numerous versions of Rails and Ruby.

[thomasfedb]

@JonRowe Yes, I saw the list on Travis - impressive.

However, personally I'd take fixing a bug over avoiding a warning, especially when the bug breaks tests that previously passed.

[chuyeow]

I can confirm that this fixes a test that broke after I upgraded to 3.0.0.beta1 from 2.99.0.beta1.

FWIW the test that now fails looks something like this:

describe ApplicationController do
  controller do
    def index
      render nothing: true
    end
  end

  describe '#authenticate_user!' do
    it 'redirects to a confirmation controller' do
      get :index
      expect(controller).to redirect_to(new_confirmation_path) # new_confirmation_path is undefined
    end
end

[thomasfedb]

Very similar spec to the one I have @chuyeow. Can you confirm that it works against my branch?

[chuyeow]

Yup the test passes after I pull your patch.

[chuyeow]

FWIW again, +1 to this PR. I really want to get rid of that single failing test in my test suite :)

I agree with @thomasfedb that fixing this bug is more important than the warnings.

[JonRowe]

When this has a test that fails, and it fixes the warning too I'll merge it

On Friday, 20 December 2013, Cheah Chu Yeow wrote:

FWIW again, +1 to this PR. I really want to get rid of that single failing
test in my test suite :)

I agree with @thomasfedb https://github.com/thomasfedb that fixing this
bug is more important than the warnings.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/881#issuecomment-30994153
.

[JonRowe]

Ok, I've fixed up a lot of warnings on rspec-rails and enabled our warning detection from rspec-support over in #892, can you rebase this off that, and add a spec showing your bug, then I'll merge this if it's green.

[thomasfedb]

Please continue at #905.