Since I have been spec-driving Rails code, and since I have very limited table space at home right now, I have been experimenting with something I’d never liked before: an electronic test and refactoring list. So far, it’s not too bad.

Like many people, I like to avoid keeping lists in my head, since I tend to forget things. Normally, I use an index card and write tests from the top and refactorings from the bottom, crossing each out as I perform them. It works well. So well, in fact, that I generally question anyone who attempts to use an electronic system, rather than a paper system. The idea, of course, is that we tend to fixate on electronic systems, whereas we use paper systems simply and quickly. Using the computer for anything that doesn’t benefit from automation seems counterproductive to me. Still, since we’re low on table space, I’ve had to try something else, and RSpec gives me the opportunity to maintain a test list without having to implement multiple failing tests (or specs) at once.

With RSpec, one can describe a test with prose without implementing it, simply by invoking it without a block.

describe "The thing I'm spec-driving" do
  it "should do something important"
  it "should do something else important"
  it "should do this third, equally important, thing"
end

When it’s time to implement the spec, add the corresponding block. In the meantime, RSpec reports the other specs as “pending”, rather than attempting to execute them. I had never thought it was a good idea, before, but I’ve been trying it recently, and I’m less annoyed by it than I used to be. That might be because I’m more comfortable spec-driving than I was years ago when I started; but it might just be because my personal taste has changed. I can’t say. What I can say is that I haven’t managed to get lost yet, and it takes little extra effort to describe a handful of specs up front. If it starts bothering me, I’ll be sure to describe my experience.