We're going to take a trip back in time… to the land of Ruby Todos. It is a magical and mundane realm with clearly defined expectations and many opportunities to practice object-oriented design.
You are already familiar with the premise, and you have probably written solution code. But is your code tested?
In this challenge, you'll be writing the specs for a simplified version of the Ruby Todos challenge. Don't worry, you won't be testing the user interface. You'll just be writing unit tests.
A word of warning on writing object-oriented specs When you are writing your examples, keep in mind that writing a test is equivalent to adding dependencies on your code. As a general principle, you want to avoid dependencies.
To be both effective and well-designed, your specs need to enforce the contracts for your classes. They should still treat your objects with the same deference and respect that the other objects in your application code do.
You want your tests to be as loosely coupled to the application code as possible. By necessity, they need to know the name of the class, its methods, and the inputs and return values of those methods.
When writing unit tests, be wary of the temptation to test too deeply, i.e. to test the implementation instead of the interface. Remember, you don't care how your code works, you just want to ensure that it works as expected.
Download the files from the gist. You will find two application files: list.rb and task.rb and two spec files list_spec.rb and task_spec.rb.
You can run the spec suite like so:
$ rspec task_spec.rb
# [ test output ]
$ rspec list_spec.rb
# [ test output ]Before you write any examples, be sure to read through all of the code and the specs.
You'll notice that the specs for Task are already written, whereas for List there are no specs.
Once you understand how the code works, you can begin writing specs for List.
Write a full spec suite for the List class. Your examples should test the full public interface of List.
Again, ensure that your specs actually test the code by checking that they fail when you comment out the relevant application code.
For reference, here are some good questions to ask when writing tests:
- What constitutes valid vs. invalid inputs?
- What are the expected and unexpected return values?
- What other effects does the method have on the state of the program?
Over the course of writing your tests, you may find yourself discovering inconsistencies or irregularities in the code. That's great! That is what is supposed to happen! Writing specs are a great way to analyze your code more closely.
As you write the specs for List and come across an oversight in the program architecture, don't be afraid to make improvements to the code. You shouldn't change the public interface, but you can certainly change the implementation of the public methods.
As an example, take the #complete_task method:
def complete_task(index)
tasks[index].complete!
endWhen you are writing the examples for this method, you will eventually come to the question of "What constitutes invalid input?" One answer to this question is that the index is beyond the scope of the array. As the code is written now, when this happens it will be call complete! on nil, which is nor so good.
In this case, the spec has exposed a potential flaw in our program. It would be better to catch this potential error with a refactor along these lines:
def complete_task(index)
return false unless tasks[index]
if tasks[index].complete!
return true
end
endThis code is an improvement in two ways:
- Invalid input is caught and will return
false, which we can now test for. - It no longer returns the return value of
#complete!, which for all we care could be anything at all. Instead, it merely checks to be sure that#complete!returns a truthy value and then returns true. This ensures that we have control over the return value of#complete_taskregardless of what#complete!returns.
Be on the lookout for other potential improvements like this one.