rspec-to-conclusion

A node based command for running rspec to conclusion

Background

One time, myself, and my friends-- we found ourselves as JavaScript engineers, building JavaScript products, writing JavaScript Unit Tests.. and it was pretty cool.

But our integration tests? The tests that "drove a browser?" Those tests were written in Ruby.

Ruby?

Ruby!

Ruby.

And RSpec + Capybara.

I :heart: Ruby. And I :blue_heart: RSpec.

But.. the big bummer was that the integration tests failed sporadically and a full test run took as long as 45 minutes.

And.. for one reason or another, we couldn't fix the tests directly or delete the problematic ones.

It was a weird time.

And so, for a lark, I wanted to see what would happen if we wrote a small utility that managed rspec runs with JavaScript: :money_with_wings:.

Usage

If you just run rspec by itself, it will run all of the specs in your specs/ directory.

However, if any of these fail, you have to manually parse through the output to see which ones failed, and then run them again.

If you run rspec-to-conclusion it will run all of the specs in your specs/ directory, and it will also:

• Create an HTML record of the run in rspecToConclusion-1.html, and create a JSON record of the run in rspecToConclusion-1.json
• Parse the JSON and make note passing and failing specs.
• If you had any failures, it will start another rspec run and repeat the process above but for only the failed specs.
• HTML and JSON records of what happened during the second run will be created at: rspecToConclusion-2.html and rspecToConclusion-2.json
• rspec-to-conclusion will keep repeating this process until all of the specs have passed (:thumbsup:) or the maximum number of tries has been reached but you still have failures (:thumbsdown:).
• Once the runs have ended, an ascii summary will be printed. An example:

| File Name and Line Number      | Result  | Tries |
| ------------------------------ | ------- | ----- |
| ./spec/always_pend_spec.rb:2   | PENDING | 0     |
| ./spec/always_pend_spec.rb:6   | PENDING | 0     |
| ./spec/always_pass_spec.rb:2   | PASSED  | 1     |
| ./spec/always_pass_spec.rb:6   | PASSED  | 1     |
| ./spec/flakey_spec.rb:2        | PASSED  | 1     |
| ./spec/really_flakey_spec.rb:2 | PASSED  | 1     |
| ./spec/flakey_spec.rb:6        | FAILURE | 2     |
| ./spec/really_flakey_spec.rb:6 | FAILURE | 2     |

Finally, the status code returned by rspec-to-conclusion will be the number of failing specs. So-- if they all passed-- it'll be zero and signal success.

Options

You can focus the number of specs by supplying a parameter:

target specific specs

rspec-to-conclusion specs/for/a/feature

--tries

By default, the maximum number of tries is 5. You can change this to be more or less:

rspec-to-conclusion --tries 10

ENV variables

Any environment variables will be passed onto rspec:

NOT_ON_MY_WATCH=true rspec-to-conclusion

Installation

You can get rspec-to-conclusion by typing:

npm install -g rspec-to-conclusion

Miscellaneous

rspec must be in your PATH.