Put your tests right inline with your code. Like rdoc for testing.
Fetching latest commit…
Cannot retrieve the latest commit at this time
= An Experiment First let me say up front that this library is an experiment. I don't even know if it is a good idea yet but if you are feeling adventurous I encourage you to take the red pill and see how deep this rabbit hole goes. = Putting Tests In-line So the basic goal of this library is to allow you put your test right next to the code being tested. For example: def multiply(a, b) a * b end Test do assert_equal 6, multiply(2, 3) assert_equal 10, multiply(5, 10) assert_not_equal 42, multiply(6, 9) end Lets go over a few of the key features of in-line testing. == Optional Naming The name of a test is often used just to know what code it is testing. For example under traditional Test::Unit I might call the test "test_multiply". But this name is entirely redundant when you start placing your tests next to the code being tested. It is obvious what is being testing. When a tests fails we are often just looking at the line number so naming the test doesn't really help there. This doesn't mean I am against naming test. For example you may have multiple test blocks that are testing a single method. You want them in multiple blocks because each has a specific purpose. So you may want to give them a name to let others know of that purpose. For example we might write the above as: def multiply(a, b) a * b end Test 'success' do assert_equal 6, multiply(2, 3) assert_equal 10, multiply(5, 10) end Test 'failure' do assert_not_equal 42, multiply(6, 9) end Here we have separated our success testing and our failure testing into separate test blocks. But even here the name is just a label. In fact we may later have another set of tests that have a "success" and "failure" block. This is allowed (multiple test blocks can have the same label). == Test By Running Another advantage to having tests and code together is that you can test a file just by running it (assuming the code is written to be fairly independent). No need to have fancy rake tasks or test suites. Just execute the file you are currently coding and the results are returned to you. Of course you can still run a test suite of your entire project if you want but sometimes it is nice to be able to pass around a file that contains the documentation, the code and the test all together and you can just run it to verify it is working. == Rails integration We also provide Rails integration. Simply include this gem in your Rails project and add the following lines to your Rakefile: require 'test_inline' require 'test_inline/rails_tasks' Now you will be able to run your unit tests with a simple: rake test:inline:units You will be able to run your functional test with: rake test:inline:functionals Or you can run both with just: rake test:inline Finally note that our Rails integration will make the TestCase inherit from the right classes in the Rails framework. So for example you can just call the normal "get, post, etc" methods in your controller tests, you can access the Rails specific assertions (such as assert_redirected_to). You can just place your tests right in with the controllers, helpers and models and it will do the right thing. == Wrapper for Test::Unit We are currently just wrapping Test::Unit. This has the advantage of building on something well used and standard in the Ruby world. It means that any traditional Test::Unit can be converted over to test_inline mostly by just copying and pasting. This does cause problems in the fact that Test::Unit has certain concepts that we don't use (like named test cases and named test methods). For test cases we just generate anonymous classes. For test methods we generate unique name (possibly based on the optional label). This of course opens up the question of would it be a good idea to allow wrapping of other test environments like RSpec, Cucumber, etc. Those questions are left to others as Test::Unit is good enough for most Ruby projects.