gistfile1.txt

(posted at http://groups.google.com/group/cukes/browse_thread/thread/4e01cc6e1a7071d4)

Hi Guys, 

I'm the author of pickle, and despite this I broadly agree with Jonas' point of view. 
The power of capybara is best realised when you do away with web_steps.rb, and start writing your own steps using capybara's dsl.

In my opinion, pickle is likewise.  pickle_steps.rb is a general set of steps to both introduce the dev to the dsl, and enable rapid cuke writing in the early stages of a project. 

I agree with Nick - the main point of pickle is to provide a general way of storing models for the lifetime of a scenario, and also to provide a general way of referring to them, so that you don't have to have step defs littered with variable names 

In Jonas' example, I would keep his step defs, but write them with pickle's DSL, for example 

  Given /^there is a person called "Fred"$/ do 
    create_model('person "Fred"') 
  end 

And then probably refactor this to (after I have writen the Ethel step) 

  Given /^there is a person called "(.+?)"$/ do |name| 
    create_model("person \"#{name}\"') 
  end 

This then means I need to write just one step to, say go to a person's account page 

  When /^I go to "(.+?)"'s account page$/ do |name| 
    visit "/accounts/#{model!("person \"#{name.id}\")} # or whatever 
  end 

instead of writing 2 (or 3 or 4) steps, each the same except for the ivar name. 

Similarly for the fatherhood step 

  Given /^"(.+?)" is the father of "(.+?)"$/ do |father, child| 
    Fatherhood.create! :father => model!(father), :child => model!(child) 
  end 

So, I think that keeping code out of features is right, but I don't think that pickle users are required to do this, any more than capybara users are required to use css selectors.  In other words, pickle_steps.rb is an invitation to use pickle's dsl to write reusable steps of your own. 

However, I think that pickle's docs should make this clearer, and the dsl could be improved (both things that we are working on). 

Cheers, 
Ian