For anyone curious: Behavior-Driven Development (BDD) is a business logic alternative to Test-Driven Development (TDD). It uses more of a conversational tone to spell out tests, so you write what you would like the app to do rather than sitting there scratching your head trying to figure out which tests to write. So it places emphasis on the "what" you want the program to do, and fill in the "how" behind the scenes. You can actually write all of your BDD gherkin and test fly your app with known inputs and outputs and then add the logic for arbitrary input and output later. Here is an example implementation for PHP:
Oh, boy. As usual anyone trying to define the difference between BDD and TDD gets it wrong. The difference between TDD and BDD isn't the tone in which you describe your test examples. The difference lies in the depth of the layer you start developing a system (or subsystem) at. Some refer to this as Outside-In versus Inside-Out. BDD and TDD have nothing to do with DSLs or the tone in which you describe your test examples.
As an example, I practice BDD but I use MiniTest, a "TDD" testing library.
BDD's difference in comparison to TDD absolutely has to do with tone - more likely philosophy of how one programs. It's about whether you're more comfortable verifying behavior or state. Mockist testing and BDD is the former , Classic TDD is the latter.
TDD doesn't care how you write the test since it's about verifying states. Kent mostly used it for inside-out testing and never uses mocks, but sometimes plays with middle-out or outside-in interactions using programmer tests, fake objects, and stubs.
Only BDD was about specifying the test as a behavioural description in a DSL, and is always outside-in. It's pretty clear cut.
It seems you're not reading what I (or anyone else) wrote.
I never mentioned syntax. I mentioned a DSL. Using a DSL is about much more than syntax, it is about expressing verifiable behavior tests in a ubiquitous language. I said nothing about any specific syntax of external or internal DSL, heck people contort FitNesse to fit BDD for whatever reason.
I have been on Ward's wiki since 1998, this isn't exactly new. It's also not for everyone.
I wouldn't call this a BDD framework yet. It's speckled with both standard assertions and semantic assertions, and an "It" function that simply takes a block and a string. I can replicate this with using underscores in my test function names and avoid the complexity. Not trying to be a jerk, but what we should be driving for is something closer to:
func Describe("The system I want to Test"){
When("I click the red button")
Then("I should see the light turn on")
And("I should see 4 lights")
}
func Step("I click the red button"){
buttons["red"].click
}
func Step("I should see the light turn on"){
lights.on.should().be_true
}
func Step("I should see (\d+) lights"){
lights.count.should().be(4)
}
it depends on what you consider "BDD". The linked library is nearly identical to Cedar ( https://github.com/pivotal/cedar ), which is, in turn, nearly identical to rspec ( http://rspec.info/ ). Both of these libraries are very much considered BDD. You seem to be advocating something closer to cucumber/gherkin?
Maybe somewhere in the middle, but fair point. I have my bias for the style change, though that was a distraction from my point. It's premature, and in it's current form, doesn't add much value.
15 comments
[ 2.2 ms ] story [ 36.6 ms ] threadhttp://codeception.com
As an example, I practice BDD but I use MiniTest, a "TDD" testing library.
TDD doesn't care how you write the test since it's about verifying states. Kent mostly used it for inside-out testing and never uses mocks, but sometimes plays with middle-out or outside-in interactions using programmer tests, fake objects, and stubs.
Only BDD was about specifying the test as a behavioural description in a DSL, and is always outside-in. It's pretty clear cut.
http://c2.com/cgi/wiki?BehaviourDrivenDevelopment
Again, nothing to do with silly syntax. It's about acceptance testing.
I never mentioned syntax. I mentioned a DSL. Using a DSL is about much more than syntax, it is about expressing verifiable behavior tests in a ubiquitous language. I said nothing about any specific syntax of external or internal DSL, heck people contort FitNesse to fit BDD for whatever reason.
I have been on Ward's wiki since 1998, this isn't exactly new. It's also not for everyone.
So by your logic, a language that doesn't have good ability to create DSLs can't do BDD?
Should be "BDD-style framework for swift http://railsware.com/blog/2014/07/04/bdd-style-testing-in-sw...