How I Started TDD

http://blog.extracheese.org/2009/11/how_i_started_tdd.html

by r11t • 17 years ago

8 7 17 years ago

blog.extracheese.org

sync• 17 years ago

That last step is quite a leap from the previous 6.

wooby• 17 years ago

I agree - this TDD-koan stuff really turns me off. What's preventing the author from just writing fibonacci as an infinite series of tests, and an infinitely long switch statement?

What the article doesn't mention is that full coverage on a function taking all positive integers is impossible unless you're comfortable making some assumptions about the behavior of your program.

The 'leap of faith' here is induction and the transition from simple testability to recursive implementation is huge, and left basically unexplained.

I personally think TDD is a great tool and an awesome way to write code. But it doesn't have to be this mysterious; I just think a function like fib is not a good way to show the benefits.

garybernhardt3• 17 years ago

The thing preventing infinite implementation is a combination of (1) the rules of simple design and (2) the definition of "simple". I write about those in the follow-up to this post: http://blog.extracheese.org/2009/11/the_limits_of_tdd.html

I've said this several times across several forums since publishing this post, but my goal was to write a very brief example of "doing the simplest thing". It's not about doing the entire TDD process on a real-life software problem, because that post would be so long that no one would read it! :) I don't think TDD can't be adequately demystified in any short example.

tjpick• 17 years ago

gah! but all those steps make the whole process harder than just thinking hard and writing down the correct function to start with. Or to write several test cases to start with (because you know many of the expected outputs up front) and writing a single function that passes them all.

TDD should make the process EASIER and MORE efficient.

>>> This isn't a perfect example of TDD, but that's not the point.

Illustrating a process by applying it to a problem that doesn't need it is surely the wrong way to go about it. Unfortunately "the point" is mostly dwarfed by the imperfect example.

garybernhardt3• 17 years ago

This post was aimed at the person who is interested in TDD, but hasn't had their first big insight into how it works. It's about "do the simplest thing", not about a full TDD process to build a working production system. As I said to someone else in this thread, such a blog post would be huge. No one would read it. We wouldn't be here talking about it. :)

How I Started TDD

Related Stories