TypeError When dealing with link clicks via Capybara::Session

BIGred053 commented 5 years ago

Expected Behavior

I am attempting to confirm the correct function of a link with target="_blank" by switching to the window opened up when clicking the link, then checking the current_url. This should return the url opened by clicking the link.

Actual Behavior

In reality, I am getting the following error when attempting to switch_to_window:

Failure/Error: switch_to_window(new_window)

     Capybara::Poltergeist::JavascriptError:
       One or more errors were raised in the Javascript code on the page. If you don't care about these errors, you can ignore them by setting js_errors: false in your Poltergeist configuration (see documentation for details).

       TypeError: Attempting to configurable attribute of unconfigurable property.
       TypeError: Attempting to configurable attribute of unconfigurable property.
           at https://www.gstatic.com/recaptcha/api2/v1562567553145/recaptcha__en.js:5 in defineProperty
           at https://www.gstatic.com/recaptcha/api2/v1562567553145/recaptcha__en.js:5 in fZ
           at https://www.gstatic.com/recaptcha/api2/v1562567553145/recaptcha__en.js:16

If I attempt the switch_to_window in binding.pry, it gives this error code, but actually seems to process switching windows, since afterwards I can check current_url and get the correct result and a passing test.

Since I ran into this with versions 1.6.0 and 1.18.1, I decided to try to work around this by executing some javascript to reset the links' target to "self". After this workaround, when testing the specs involved individually, they typically pass. However, when I run the entire file of specs, they fail locally. They do pass in our CI process, though.

If this is an error in something in my implementation, please let me know.

Steps to reproduce

Here is the code I attempt:

new_window = window_opened_by { row_to_click.click_link }
sleep 3

switch_to_window(new_window)
expect(current_url).to include anticipated_url # use include in case routing adds path to root url

Here is my 'workaround' code:

before do
  execute_script("$('.link-to-click').each(function() {$(this).attr('target', '_self');})")
end

scenario 'the user is redirected to the correct path' do
  row_to_click.click_link
  sleep 3 # wait for event logging

  expect(current_url).to include anticipated_url # In case link redirects to more specific path
end

twalpole commented 5 years ago

You're getting errors from the JS in the page you're loading. PhantomJS (which Poltergeist depends on) went EOL a while ago and does a pretty terrible job of replicating a modern browser nowadays, so there's a lot of JS libraries that are no longer compatible with it. Thats a large part of the reason Poltergeist is no longer maintained. You will have a better time if you upgrade to one of the newer, more modern browser using drivers - https://github.com/teamcapybara/capybara#drivers

Additionally - by doing - expect(current_url).to include anticipated_url you're opening yourself to a race condition. You'd be much better off using Capybaras matcher with a regex expect(page).to have_current_path(%r{^http://x\.y\.z/blah})

BIGred053 commented 5 years ago

Thanks for this feedback!

teampoltergeist / poltergeist