RSPEC Test Passing that should fail

I'm using rspec, rails, guard and sorcery for my authentication and testing.

I have a test that is testing the length of an email. I want to reject emails that are too long. Here is the test I wrote for spec/models/user_spec.rb

 开发者_如何转开发   require 'spec_helper'

    describe User do
     before(:each) do
       @attr = { :email => "testuser@example.com", :password => "password", :password_confirmation => "password" }
     end


     it "should reject emails that are too long" do
       long_email = "a" * 101 + "gmail.com"
       long_email = User.new (@attr.merge(:email => long_email))
       long_email.should_not be_valid
     end

Here is the model validations I have in place:

class User < ActiveRecord::Base
  authenticates_with_sorcery!

  attr_accessible :email, :password, :password_confirmation

  validates_presence_of :password, :on => :create
  validates :password, :confirmation => true,
                       :length       => { :within => 6..100 }

  email_regex = /\A[\w+\-.]+@[a-z\d\-.]+\.[a-z]+\z/i
  validates :email, :presence        => true,
                    :format          => { :with => email_regex },
                    :uniqueness      => {:case_sensitive => false},
                    :length          => { :within => 5..100 }
end

I'm a noob with this stuff, so any help would be greatly appreciated. The test is green right now, it goes red if I change the line to long_email.should be_valid. Thanks in advance!

As @apneadiving points out, the format of test is important. It not only makes it easier to read, but also shows up places you haven't tested:

require 'spec_helper'

describe User do
  let(:attr){ 
    {:email => "testuser@example.com", :password => "password", :password_confirmation => "password" }
  }

  context "When given an email address" do
    context "That is too long" do
      let(:long_email){ "a" * 101 + "gmail.com" }
      subject{ User.new attr.merge(:email => long_email) }

      specify{ subject.should_not be_valid }
    end
    context "That is too short" do
      pending
    end
    context "That is between 5 and 100 characters long"
      pending
    end
  end

end

As you can see, rewriting it has done several things:

• made it easier to read
• made it easier to add more tests without having earlier tests impact later ones (by using let instead of before)
• shown you that you've only specified a failing case, and not tested for the positive case (or other cases)

If you write the positive case and that passes too then you really know something is wrong!