[rspec-users] spec'ing calls to super (or other Ruby keywords)

Scott Taylor mailing_lists at railsnewbie.com
Sat May 3 16:51:51 EDT 2008

On May 3, 2008, at 11:09 AM, Al Chou wrote:

> In the particular case of super, another approach, though perhaps  
> not using the spec framework, would be to assert (a la Test::Unit)  
> that your class is a subclass of the intended superclass.  To be  
> truly anal, also assert that the superclass has a method with the  
> same name as the subclass's method of interest and that the subclass  
> has that method, too.  But all this seems deeply into the xUnit  
> world, and far from the BDD way.
> Al

The truth is that you shouldn't be spec'ing calls to the super  
"method", but rather the behaviour of super.  One quick way to do this  
is as so:

[BaseClass, Subclass].each { |klass|
	it "should do whatever super does" do
		klass.new.foo_bar.should == :baz

Your original example is a little trickier, because you didn't write  
the base class, and you probably aren't going to spec it out.  There's  
obviously great complexity in that call to super.  Now you know why  
none of us like writing specs against rails apps.


> ----- Original Message ----
> From: John D. Hume <duelin.markers at gmail.com>
> To: rspec-users <rspec-users at rubyforge.org>
> Sent: Saturday, May 3, 2008 7:49:11 AM
> Subject: Re: [rspec-users] spec'ing calls to super (or other Ruby  
> keywords)
> I believe calls to super are sufficiently internal to the Ruby  
> interpreter that a mocking framework can't intercept them without  
> doing separate implementations for separate Ruby interpreters (and  
> likely even separate versions).  I could be wrong, but even so I'd  
> recommend a different approach.
> If your need is really as simple as your example, what you have is  
> just a method that has to get two things done: the base save and one  
> additional call.  You can write one (or more) example for each of  
> those two things without your spec knowing that one of those things  
> gets done by calling super.  (You might object that by spec'ing the  
> base #save behavior you're spec'ing the framework.  I'd say you're  
> USING the framework to spec something your code does.  To be clear,  
> I'm not suggesting you spec every detail of what save does: just  
> something to make sure the record actually lands in the db.)
> (Sidebar: Keep in mind the return value if you're really overriding  
> #save like that.)
> If you're dead set on spec'ing that the super method gets called,  
> there are a couple of hideous ways of doing it that will leak out of  
> your example.  Namely, you can (in your spec) redefine the method in  
> the superclass and verify it gets called or (also in your spec, and  
> this one's a little less leaky) have the class under test include a  
> module that defines the same method and verify it gets called.   
> Don't do either of those though (unless it's just to prove to  
> yourself that they're possible).
> -hume.
> On Fri, May 2, 2008 at 5:17 PM, Matt McNeil <nabble.108 at xoxy.net>  
> wrote:
> Hi there,
> How does one spec an invocation of a Ruby keyword, such as super in  
> this
> case?
> class User < ActiveResource::Base
>  # faking the ActiveRecord before/after_save observers
>  def save
>    super
>    UserMailer.deliver_activation(self) if recently_activated?
>  end
> end
> Does the solution look anything like the following?
> describe User do
>  describe '#save' do
>    it "should call save on the parent class" do
>      # something.should_receive(:something)
>      @user.save
>    end
>  end
> end
> Any thoughts?
> Thanks much,
> Matt
> Be a better friend, newshound, and know-it-all with Yahoo! Mobile.  
> Try it now._______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/rspec-users/attachments/20080503/68c6543d/attachment-0001.html>

More information about the rspec-users mailing list