• ruby app.rb to start the server to port 3000
• ruby integration_tests.rb to run tests.
• I used Sinatra as it serves perfectly the purpose of this assignment: this way we can keep the application code to the minimum and avoid adding extra complexity to it.
• I decoupled code by having the app.rb file which only serves as the controller. It has no knowledge about GitHub and how to retrieve recipes.
• Business logic is encapsulated inside the Search class which handles GitHub API call and returns either Success of Failure.
  • The returned search object exclusively exposes data needed across the application - therefore, we avoid using raw data as returned by Github in our views. It's not needed in such a small project, but I wanted to show the approach of code decoupling/separation (in other words, avoiding using objects returned by GitHub outside of the Search class). A better solution, to be put in place in bigger projects, would be to define our Struct (ie: Dry::Struct) which encapsulates data we need across the application.
• I created a SearchPresenter which is a decorator taking care of all view logic. This way, the view only defines the structure while contents are defined by the presenter.
• Views are in HAML: normally I use ERB (because it is more of a standard) but for such small projects HAML helps with programming speed, therefore I used it.
• I created integration tests to make sure the pages are working. In a real life scenario, I would have written unit tests, too, but, given the time constraint, I have chosen to write ony the most essential tests (which are integration tests because they prove that the whole system works).
• No GitHub credentials are needed to list public repositories.
• TODO: the application should better handle the case of a user manipulating the URL with a wrong page number. Currently it shows an empty list but in a real life scenario it should redirect to the first or last page.