thr3ads.net - mocha developer - [mocha-developer] Stubbing constructiors [Dec 2006]

If this information is useful, please help other people find it:
Share via:

Tomas Pospisek

2006-Dec-14 13:27 UTC

[mocha-developer] Stubbing constructiors

This works:

    class X
      def X.initialize( stuff )
      end
    end

    X.initialize("bla")


However stubbing it doesn,t:

    require ''test/unit''
    require ''stubba''

    class X
      def X.initialize( stuff )
      end
    end

   class XTest < Test::Unit::TestCase
     def test_
       X.stubs(:initialize).with("bla")
       X.initialize("bla")
     end
   end

Ruby lets me know that:

   Loaded suite /tmp/foo
   Started
   E(eval):1: warning: removing `initialize'' may cause serious problem

   Finished in 0.002625 seconds.

     1) Error:
   test_(XTest):
   NoMethodError: private method `initialize'' called for X:Class
      /tmp/kak.rb:12:in `test_''

   1 tests, 0 assertions, 0 failures, 1 errors

Routing around the problem is possible:

  require ''test/unit''
  require ''stubba''

  class X
    def X.initialize( stuff )
    end
  end

  class XTest < Test::Unit::TestCase
    def test_
      X.stubs(:initialize).with("bla")
      X.public_class_method :initialize
      X.initialize("bla")
    end
  end

...however a bit awkward...

:-)
*t



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

James Mead

2006-Dec-14 14:15 UTC

head link

[mocha-developer] Stubbing constructiors

On 14/12/06, Tomas Pospisek <tpo2 at sourcepole.ch>
wrote:> This works:
>
>     class X
>       def X.initialize( stuff )
>       end
>     end
>
>     X.initialize("bla")
>
>
> However stubbing it doesn,t:
>
>     require ''test/unit''
>     require ''stubba''
>
>     class X
>       def X.initialize( stuff )
>       end
>     end
>
>    class XTest < Test::Unit::TestCase
>      def test_
>        X.stubs(:initialize).with("bla")
>        X.initialize("bla")
>      end
>    end
>
> Ruby lets me know that:
>
>    Loaded suite /tmp/foo
>    Started
>    E(eval):1: warning: removing `initialize'' may cause serious
problem
>
>    Finished in 0.002625 seconds.
>
>      1) Error:
>    test_(XTest):
>    NoMethodError: private method `initialize'' called for X:Class
>       /tmp/kak.rb:12:in `test_''
>
>    1 tests, 0 assertions, 0 failures, 1 errors
>
> Routing around the problem is possible:
>
>   require ''test/unit''
>   require ''stubba''
>
>   class X
>     def X.initialize( stuff )
>     end
>   end
>
>   class XTest < Test::Unit::TestCase
>     def test_
>       X.stubs(:initialize).with("bla")
>       X.public_class_method :initialize
>       X.initialize("bla")
>     end
>   end
Not being able to stub private methods is a known issue (see elsewhere
on this mailing list). We may add functionality to allow it, but even
if we do I suspect we will want to make the API explicitly express the
fact that you are stubbing a private method which means you are
tightly coupling your test to the implementation details of the class.

Have you considered stubbing the "new" method which is a public
method? It is effectively an implementation detail of the Object class
that the new method calls the initialize method.

-- 
James.
http://blog.floehopper.org

Tomas Pospisek

2006-Dec-14 15:06 UTC

head link

[mocha-developer] Stubbing constructiors

Quoting James Mead <jamesmead44 at gmail.com>:
> On 14/12/06, Tomas Pospisek <tpo2 at sourcepole.ch> wrote:
> > This works:
> >
> >     class X
> >       def X.initialize( stuff )
> >       end
> >     end
> >
> >     X.initialize("bla")
> >
> >
> > However stubbing it doesn,t:
> >
> >     require ''test/unit''
> >     require ''stubba''
> >
> >     class X
> >       def X.initialize( stuff )
> >       end
> >     end
> >
> >    class XTest < Test::Unit::TestCase
> >      def test_
> >        X.stubs(:initialize).with("bla")
> >        X.initialize("bla")
> >      end
> >    end
> >
> > Ruby lets me know that:
> >
> >    Loaded suite /tmp/foo
> >    Started
> >    E(eval):1: warning: removing `initialize'' may cause
serious problem
> >
> >    Finished in 0.002625 seconds.
> >
> >      1) Error:
> >    test_(XTest):
> >    NoMethodError: private method `initialize'' called for
X:Class
> >       /tmp/kak.rb:12:in `test_''
> >
> >    1 tests, 0 assertions, 0 failures, 1 errors
> >
> > Routing around the problem is possible:
> >
> >   require ''test/unit''
> >   require ''stubba''
> >
> >   class X
> >     def X.initialize( stuff )
> >     end
> >   end
> >
> >   class XTest < Test::Unit::TestCase
> >     def test_
> >       X.stubs(:initialize).with("bla")
> >       X.public_class_method :initialize
> >       X.initialize("bla")
> >     end
> >   end
>
> Not being able to stub private methods is a known issue (see elsewhere
> on this mailing list). We may add functionality to allow it, but even
> if we do I suspect we will want to make the API explicitly express the
> fact that you are stubbing a private method which means you are
> tightly coupling your test to the implementation details of the class.
>
> Have you considered stubbing the "new" method which is a public
> method? It is effectively an implementation detail of the Object class
> that the new method calls the initialize method.
Um, well, kind of. The above is of course just a "mock" of the real
situation
I''m confronted with in my code and there I really do have a *class*
method
named "initialize" that I call just like that. That is I am
initializing the
class and not instantiating an object, in which case, as you correctly point
out I''d better stub "new".

It smells a bit, yes, unfortunately I''m only at the "aspiring to
become a Ruby
wizzard" stage...

Thanks,
*t



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Jay

2006-Dec-15 01:37 UTC

head link

[mocha-developer] Stubbing constructiors

Your initialize method is a class method, but the class already has  
an initialize method since classes are objects.  I''d suggest changing  
the name of your method since overwriting the initialize method of  
your class could have side-effects...

Jay

On Dec 14, 2006, at 10:06 AM, Tomas Pospisek wrote:
> Quoting James Mead <jamesmead44 at gmail.com>:
>
>> On 14/12/06, Tomas Pospisek <tpo2 at sourcepole.ch> wrote:
>>> This works:
>>>
>>>     class X
>>>       def X.initialize( stuff )
>>>       end
>>>     end
>>>
>>>     X.initialize("bla")
>>>
>>>
>>> However stubbing it doesn,t:
>>>
>>>     require ''test/unit''
>>>     require ''stubba''
>>>
>>>     class X
>>>       def X.initialize( stuff )
>>>       end
>>>     end
>>>
>>>    class XTest < Test::Unit::TestCase
>>>      def test_
>>>        X.stubs(:initialize).with("bla")
>>>        X.initialize("bla")
>>>      end
>>>    end
>>>
>>> Ruby lets me know that:
>>>
>>>    Loaded suite /tmp/foo
>>>    Started
>>>    E(eval):1: warning: removing `initialize'' may cause
serious
>>> problem
>>>
>>>    Finished in 0.002625 seconds.
>>>
>>>      1) Error:
>>>    test_(XTest):
>>>    NoMethodError: private method `initialize'' called for
X:Class
>>>       /tmp/kak.rb:12:in `test_''
>>>
>>>    1 tests, 0 assertions, 0 failures, 1 errors
>>>
>>> Routing around the problem is possible:
>>>
>>>   require ''test/unit''
>>>   require ''stubba''
>>>
>>>   class X
>>>     def X.initialize( stuff )
>>>     end
>>>   end
>>>
>>>   class XTest < Test::Unit::TestCase
>>>     def test_
>>>       X.stubs(:initialize).with("bla")
>>>       X.public_class_method :initialize
>>>       X.initialize("bla")
>>>     end
>>>   end
>>
>> Not being able to stub private methods is a known issue (see  
>> elsewhere
>> on this mailing list). We may add functionality to allow it, but even
>> if we do I suspect we will want to make the API explicitly express  
>> the
>> fact that you are stubbing a private method which means you are
>> tightly coupling your test to the implementation details of the  
>> class.
>>
>> Have you considered stubbing the "new" method which is a
public
>> method? It is effectively an implementation detail of the Object  
>> class
>> that the new method calls the initialize method.
>
> Um, well, kind of. The above is of course just a "mock" of the
real
> situation
> I''m confronted with in my code and there I really do have a
*class*
> method
> named "initialize" that I call just like that. That is I am  
> initializing the
> class and not instantiating an object, in which case, as you  
> correctly point
> out I''d better stub "new".
>
> It smells a bit, yes, unfortunately I''m only at the "aspiring
to
> become a Ruby
> wizzard" stage...
>
> Thanks,
> *t
>
>
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> _______________________________________________
> mocha-developer mailing list
> mocha-developer at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mocha-developer

Reasonably Related Threads

Search for more reasonably related threads

mocha developer - Dec 2006 - Stubbing constructiors

[mocha-developer] Stubbing constructiors

[mocha-developer] Stubbing constructiors

[mocha-developer] Stubbing constructiors

[mocha-developer] Stubbing constructiors

Reasonably Related Threads