[rspec-users] should_be_different -- possible implementation

[David Chelimsky]

On 12/4/06, s.ross <cwdinfo at gmail.com> wrote:
> There is an assert_difference helper for Test::Unit (you can read
> about it at http://blog.caboo.se/articles/2006/06/13/a-better-
> assert_difference and to some it should look familiar -- you know who
> you are :), so I took a shot at one for rSpec on Rails.

We've already got this in 0.7.4. It's called assert_change, but it's
the same functionality you're looking for:

http://rspec.rubyforge.org/documentation/expectations.html

Cheers,
David

>
>
> class Object
>      def should_be_different(method = nil, difference = nil)
>          return self.should_not_eql(yield) if method.nil?
>
>          before = self.send(method)
>          yield
>          return self.send(method).should_not_eql(before) if
> difference.nil?
>          (self.send(method) - before).should_eql difference
>      end
> end
>
>
> Usage:
>
>      specify "should be able to create a new user" do
>          Person.create(:name => 'joe')
>          Person.should_be_different :count do
>              Person.create(:name => 'bob')
>          end
>      end
>
> This has probably already been done, but I thought I'd put it out
> here for comment in case it hadn't. Note that if you omit the method,
> the objects are tested for equality (however the object defines it).
> If you include the method but omit the difference, then any amount
> difference is ok. Testing destroy might indicate using -1 for the
> difference to indicate that the count has gone down.
>
> My implementation works, but a more natural syntax would be:
>
> Person.count.should_be_different do
>         blah
> end
>
> and:
>
> Person.count.should_be_different_by 1 do
>         blah, blah
> end
>
> Comments?
>
> Steve
>
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
>