BDD test framework Shoulda

Behavior Driven Development is an interesting slant on testing. There appear to be many opinions on its advantages and disadvantages. At the end of the day, if you are testing, you have already won.

So, the question is, how can you tweak your method to improve on your bugs-per-line-of-code metric. This post isn't about that because I have no idea if BDD or Shoulda will help.

Shoulda is an alternative BDD framework to RSpec that has been argued about for so long. I can't help you with that either.

Instead, I just started using Shoulda in a Rails app so thought I would share some complaints, because being constructive is just too much work.

I switch a simple controller test over to Shoulda and it ran and passed but the original 3 tests that used to pass are now reported as "0 tests, 0 failures, 0 errors". While I proved that if there was an error it did explain precisely what was wrong to me, this 0 tests thing is severely demotivating. It is amazing how watching the number of tests grow in your project can encourage you to continue to strive for perfection.

IDE integration is always a problem with new stuff. NetBeans allows me to run a single test case with ctrl-shift F6 which is convenient if you have few that are suddenly failing. No such luck with the Shoulda version, all tests all the time.

However, all this is trivia compared to being able to do something like this:

require 'test/unit'
require 'shoulda'

class Lion
def speak
"rarrrr hickup"
end
end

class TestLion < Test::Unit::TestCase

context "the lion" do
setup do
@lion = Lion.new
end

should "say rarrr" do
assert_equal "rarrr", @lion.speak
end
end
end

and get this explanation of what you did wrong.

1) Failure:
test: the lion should say rarrr. (TestLion):
<"rarrrr"> expected but was <"rarrr hickup">.

Craft your feedback as carefully as you craft your output.