Get master passing again

[BookOfGreg]

Master branch does not pass travis on it's own, making it difficult to tell when new branches are complete.
I plan on cleaning up the appraisal setup with the aim of making this slightly easier to work on.

• Handle the Rails 5.1.0.rc1 compatibility
• Handle turbolinks better
• Fix broken test on BUNDLE_GEMFILE=/Users/e0052727/source/react-rails/gemfiles/rails_3.2.gemfile bundle exec rake test at test_it_reloads_when_new_jsx_files_are_added_to_the_asset_pipeline Seems to be a flakey test, passes sometimes?
• Consistently name tests from test "foo bar" do to def test_foo_bar
• Rubocop with the standard ruby style
• Fix broken test on rails 5.0 + webpacker 1.2 around manifest_path

[swrobel]

Why not use the test keyword? Rails still recommends it in their edgeguides and says the following: "Although you can still use regular method definitions, using the test macro allows for a more readable test name."

[BookOfGreg]

When a test fails with "FAILED: ControllerTest#test_something_happened", you can global find it if the test name matches test_something_happened but not test "some-thin'g ()happened".

Perhaps this is personal preference but it's much much faster for me to fix tests if I can find them.

Edit: Also when you pass it on the CLI with --name, it also takes it in the form TESTOPTS="--name test_something_happened"

[BookOfGreg]

Think I found the cause of some of my remaining test failures. I'd been running Webpacker 1 against Webpack 2, which is why all my test files keep ending up in the wrong positions. Will probably implement a version check like https://github.com/shakacode/react_on_rails/blob/master/lib/react_on_rails/version_checker.rb to make at least the tests a bit more stable.

I'm a little concerned at how much work was involved at getting the master branch passing so far though. Am I doing something wrong or missing something obvious?

[rmosolgo]

@BookOfGreg Thanks digging in on this!

how much work

I'm not surprised 😬 It was a lot of work to get webpacker working in the first place, and since then, the Webpacker APIs have changed a lot, so I bet it will be hard to get things working properly again 😖

[BookOfGreg]

FYI The last fail on travis is in the webpacker:compile Rake task which is in capture_io, hence why there is no error output.
It's complaining about looking for the manifest in packs/packs when webpack put it in public/packs.
Honestly not sure what to do about that.

[BookOfGreg]

Closing in favour of #777