[LRUG] Details for your cuketober meeting

Paul Battley pbattley at gmail.com
Fri Oct 8 08:06:25 PDT 2010


I've got 16,082 lines of Cucumber here that have convinced me that
it's a terrible idea. And it's all my fault! It seemed like a good
idea at the time. I could show you a thousand examples of why it's
bad, but, as with intercessionary prayer, I suspect that the answer
would merely be that I'm not cuking hard enough. I don't expect to
convince anyone through argument - it's a level of abject pain you
have to experience for yourself.

Tests are code, and Cucumber is not quite a programming language. The
software world is littered with simple declarative systems that edged
towards Turing completeness. Some, thankfully, are just corpses.
Others, such as PHP or MUMPS[1], persist like zombies, eating the
brains of unwary programmers. I fear that Cucumber is one of these.

There is the core of a good idea in Cucumber: separating the intention
of a test from the fine details of its implementation. All I'll say to
that is that Ruby already provides a means of doing this: methods.
Cucumber won't stop you writing leaky steps, although it will make it
uglier and harder.

As to the idea that "the business" is going to read - let alone write!
- Cucumber features, I call bullshit. I'd recommend talking as a more
effective alternative method of communication.

But I don't even care if anyone else uses Cucumber. It would be nice
if it went away and died, but really, it doesn't matter. As with PHP,
I just want to avoid it in future. If you're looking for an
alternative, I'd say just use Ruby. You can still use Capybara, after
all.

Paul.

1. http://thedailywtf.com/Articles/A_Case_of_the_MUMPS.aspx



More information about the Chat mailing list