rspec users - Aug 2011 - [JRuby] simulating a java exception in a spec

JRuby 1.6.2
rspec-core (2.6.4)
rspec-expectations (2.6.0)
rspec-mocks (2.6.0)
rspec-rails (2.6.1)

I''m sure this has more to do with the way JRuby wraps Java exceptions
but I figured I''d post here in case anyone here has any insight or
pointers. In the context of writing a spec for a model like thing that wraps
legacy Java code, I found myself attempting to stub a method on a Java Exception
rescued in the implementation.

eg.

describe ''#valid?'' do
....
it ''adds validation exceptions raised by service to #errors''
do
      ve = ValidationException #a java exception
      ve.stub(:localized_message).and_return(''a bunch of
errors'')

      service.stub(:validateTaskForSave).and_raise(ve)

      subject.valid?

      subject.errors.should == [''a bunch of errors'']
end

the above example fails because the :localize_message stub is ignored and
instead the real implementation receives the message. I know this typically
screams typo but not in this case. Here is a more boiled down version:

specify ''rescued exception message should be "bar" because I
stubbed it'' do
      begin
        e = Java::java.lang.NullPointerException.new(''foo'')
        e.stub(:message).and_return(''bar'')
        raise e
      rescue Java::java.lang.NullPointerException => e
        e.message.should == ''bar''
      end
end

Failure/Error: e.message.should == ''bar''
       expected: "bar"
            got: "foo" (using ==)

That seemed odd to me but maybe moot anyway since in reality I would be need to
rescue a NativeException masquerading as my target exception.

e.g. This code

def valid?
    begin
      service.validateTaskForSave(task)
    rescue ValidationException => e
      puts "rescued exception: #{e.class.name}"
      .....

prints "rescued exception: NativeException"  when hit through the
console.

So how would one simulate a NativeException if needed (i.e. you want to stub
methods on it)?

-lenny

it ''adds validation exceptions raised by service to #errors''
do
     ve = ValidationException #a java exception
     ve.stub(:localized_message).and_return(''a bunch of
errors'')
    ....


Just clarifying, but did you mean
    ve = ValidationException.new


I tried replicating your spec but with java.lang.RuntimeException
instead of ValidationException and I got
   exception class/object expected

because raise expects an instance of the exception, not the exception
class. Am I missing something?

Best,
Sidu.
http://c42.in
http://blog.sidu.in



On 23 August 2011 03:20, Lenny Marks <lenny at aps.org>
wrote:
> service.stub(:validateTaskForSave).and_raise(ve)

> 
> On Aug 22, 2011, at 5:50 PM, Lenny Marks wrote:
> 
>> JRuby 1.6.2
>> rspec-core (2.6.4)
>> rspec-expectations (2.6.0)
>> rspec-mocks (2.6.0)
>> rspec-rails (2.6.1)
>> 
>> I''m sure this has more to do with the way JRuby wraps Java
exceptions but I figured I''d post here in case anyone here has any
insight or pointers. In the context of writing a spec for a model like thing
that wraps legacy Java code, I found myself attempting to stub a method on a
Java Exception rescued in the implementation.
>> 
>> eg.
>> 
>> describe ''#valid?'' do
>> ....
>> it ''adds validation exceptions raised by service to
#errors'' do
>>      ve = ValidationException #a java exception
>>      ve.stub(:localized_message).and_return(''a bunch of
errors'')
>> 
>>      service.stub(:validateTaskForSave).and_raise(ve)
>> 
>>      subject.valid?
>> 
>>      subject.errors.should == [''a bunch of errors'']
>> end
> it ''adds validation exceptions raised by service to
#errors'' do
>     ve = ValidationException #a java exception
>     ve.stub(:localized_message).and_return(''a bunch of
errors'')
>    ....
> 
> 
> Just clarifying, but did you mean
>    ve = ValidationException.new
> 
> 
> I tried replicating your spec but with java.lang.RuntimeException
> instead of ValidationException and I got
>   exception class/object expected
> 
> because raise expects an instance of the exception, not the exception
> class. Am I missing something?
> 
> Best,
> Sidu.
> http://c42.in
> http://blog.sidu.in
You have to raise an exception instance, not the class of the exception as is
typical with Ruby exceptions. Check out the "boiled down" example
below to reproduce.
>> 
>> the above example fails because the :localize_message stub is ignored
and instead the real implementation receives the message. I know this typically
screams typo but not in this case. Here is a more boiled down version:
>> 
>> specify ''rescued exception message should be "bar"
because I stubbed it'' do
>>      begin
>>        e =
Java::java.lang.NullPointerException.new(''foo'')
>>        e.stub(:message).and_return(''bar'')
>>        raise e
>>      rescue Java::java.lang.NullPointerException => e
>>        e.message.should == ''bar''
>>      end
>> end
>> 
>> Failure/Error: e.message.should == ''bar''
>>       expected: "bar"
>>            got: "foo" (using ==)
>> 
>> That seemed odd to me but maybe moot anyway since in reality I would be
need to rescue a NativeException masquerading as my target exception.
>> 
>> e.g. This code
>> 
>> def valid?
>>    begin
>>      service.validateTaskForSave(task)
>>    rescue ValidationException => e
>>      puts "rescued exception: #{e.class.name}"
>>      .....
>> 
>> prints "rescued exception: NativeException"  when hit through
the console.
The more I think about this, the more I wish it was different(more explicit) in
JRuby. IMO, the above is very unintuitive (i.e. if I rescue a specific exception
class then I would expect the exception instance to be an instance of that
class). Any attempt to reference a custom method on a rescued java exception
results in "undefined method".

rescue MyJavaException => e
  # e.some_method  "undefined method"
  e.cause.some_method # need to unwrap
end

AFAICT, the JRuby wiki makes no mention of this behavior (
https://github.com/jruby/jruby/wiki/CallingJavaFromJRuby ). Of course
that''s more suited for the JRuby mailing list, but if it is just a
matter of filling in the docs then there should still be an easy way to simulate
such an exception with #and_raise such that custom methods on the exception can
be stubbed. No??

-lenny
>> 
>> So how would one simulate a NativeException if needed (i.e. you want to
stub methods on it)?
>> 
>> -lenny
>> 
>> 
>> 
>> 
>> 
>> _______________________________________________
>> 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/20110823/9e98e0ad/attachment.html>

Please ignore my previous question - I just realised you''re doing
Java::java.lang.NullPointerException.new in your second example. I''m
seeing the same behaviour and can''t think of any way around it off the
top of my head.

You could try redefining Java::JavaLang::NullPointerException to be a
Ruby RuntimeError within the scope of your spec but I''m not sure how
this will work (if at all) when Java::JavaLang::NullPointerException
is raised from native code.

Best,
Sidu.
http://c42.in

On 23 August 2011 22:59, Sidu Ponnappa <ckponnappa at gmail.com>
wrote:
> it ''adds validation exceptions raised by service to
#errors'' do
> ? ? ve = ValidationException #a java exception
> ? ? ve.stub(:localized_message).and_return(''a bunch of
errors'')
> ? ?....
>
>
> Just clarifying, but did you mean
> ? ?ve = ValidationException.new
>
>
> I tried replicating your spec but with java.lang.RuntimeException
> instead of ValidationException and I got
> ? exception class/object expected
>
> because raise expects an instance of the exception, not the exception
> class. Am I missing something?
>
> Best,
> Sidu.
> http://c42.in
> http://blog.sidu.in
>
>
>
> On 23 August 2011 03:20, Lenny Marks <lenny at aps.org> wrote:
>> service.stub(:validateTaskForSave).and_raise(ve)
>