Skip to main content

Cucumber - my perspective

I have been using Cucumber (http://cukes.info/) at my current gig for about a year now. My initial reaction was that I absolutely hated it. It didn't seem to make sense for a programmer to write out tests (features) in plain English and then write out a bunch of regular expressions to turn that plain English into runnable code. What a palaver!

The other problem, is that the Cucumber tests were extremely fragile. Even making text and/or HTML changes would break things in lots of random places.

Anyways, as it turns out, I don't really hate Cucumber, I just hate the way it is implemented in my current gig. Here are some lessons I learnt on the way...

1) Features are not supposed to be written by programmers.
You can write features as a programmer, but you are not the intended audience. The reason why features are written in plain text is that they are supposed to be written by business owners. As a programmer though, you can use features to organize your thoughts in plain English.

For example

Given I am a shopper
When I add something to the cart
Then I should see it in the cart

2) Keep all implementation details out of the features.
The features are not for you (the programmer). They are for the business owners. You are just supposed to make them pass and use them as a guide.

3) Don't aim to re-use features
Testing is about the one place where DRY (Do not Repeat Yourself) does not apply. When you re-use the same tests/step definitions in multiple places, you are creating hidden dependencies between 2 sections of code which probably don't need to be there. A lot of developers also fall into this trap with CSS. As developers we are trained to look for similar behaviours and abstract them out into their own methods, but unless the business owner really decides the two features are linked, they should be kept separate. The one area where I don't agree with Cucumber is that all the step definitions are shared globally (I personally think each feature should have its own step definition file).

4) Features are not supposed to cover every possible edge case
Specs are used to make sure all your edge cases are covered. Features should only cover what is asked for by the business owners.

5) Do not tie your step definitions to HTML elements
This makes it impossible to do any redesign. It is a bit annoying that copy (or text) changes can break your tests, but at the end of the day, business owners are the ones who are in charge of copy and they should be aware that copy changes will come at a cost.

So at the end of the day, keep your features concise, don't reuse steps and write them like you are not a programmer and then you should be happy with Cucumber.

Comments

Popular posts from this blog

Freezing Gems

What is a gem and why would you want to freeze it?

In Ruby, there are times when you want to access pieces of functionality that other people of written (3rd party libraries) and you normally have 2 options. You can install a plug in or install a gem. Normally the method you use is determined by which ever is made available by the author.

Gems are installed on the host machine and are pretty handy when you want to run things in the command line or else across lots of projects, but their downside is that if you use a gem in a Rails project there is no automatic publishing mechanism when you deploy your site. You will need to log onto the remote host machine and install the gem manually.

Plugins are specific to Rails and are similar to gems in that they are also 3rd party libraries. However they are associated with your Rails project as opposed to your machine so they will get posted to the server on a regular deploy.

Freezing a gem is the process of transforming a gem into a plug in. Essen…

Unit/Functional Testing RubyAMF

One of my current projects is using RubyAMF to communicate with Flash (http://rubyforge.org/projects/rubyamf/). On the whole this is really nice because it allows you to transfer Ruby objects directly to ActionScript ones (as opposed to translating the object into XML, sending the XML and then recreating the object in ActionScript).
However, Rails does not provide a built in transport mechanism for AMF, so we cannot run functional testing directly on the data call (as we could for an XML or HTML transport layer). This is a show stopper for a lot of people (Rails w/o Unit testing = a big mess of trouble when something goes wrong).
We can though serve both the HTML and the AMF formats depending on the request format. This means that we can test the object instantiation logic and make sure there are no errors in the controllers (though we cannot check the actual format of the data being served). In the controller, instead of rendering AMF alone, do the following respond_to do |format|

Comparing Rails' Active Record Pattern with Phoenix/Elixir/Ecto

Rails has a very well established Active Record pattern for dealing with the database. You have an Active Record model which maps to the database table, the schema of the model comes directly from the database schema and you place your model specific methods on the Active Record model. This file is also where you set your model relationships (e.g. has_many, has_one, belongs_to). Your instance of the model has all the methods built in.

In Ecto/Phoenix it's a little different. First of all, the database schema doesn't automatically map to the "model". In fact we don't really have models (as Elixir is a functional paradigm). What happens in one file in Rails, happens in essentially two (or more). You have a schema file (where you have to list out all the attributes and relationships). Using the schema file, your "instance" is essentially a data structure (with no methods on it). If you want to transform the data on your struct, you would use a context modu…