rspec users - Sep 2009 - Expectation test: error says it's different, but it's printed the same way

I put an expectation test, but it complains two things are different
when it prints it out in the error in the exact same way:
http://pastie.org/618481

or below:

## the test
  describe ".process_aftersale" do
    before do
      @orders = [Order.make, Order.make]
      Order.stub!(:paid).and_return(@orders)
      @orders.stub!(:updated_on).and_return(@orders)
    end

    ....

    it "should send 1 week after sale letters" do
      @orders.should_receive(:updated_on).once.with([1.week.ago])
      Order.process_aftersale
    end

    ...
  end

## The error - note that the dates are the same
1)
Spec::Mocks::MockExpectationError in ''Order.process_aftersale should
send 1 week after sale letters''
#<Order:0xb4e8b3a4>#<Order:0xb4e69448> expected :updated_on with
([Wed, 09 Sep 2009 14:59:59 HKT +08:00]) but received it with ([Wed,
09 Sep 2009 14:59:59 HKT +08:00])
./spec/models/order_spec.rb:42:


## the method in Order
  def self.process_aftersale
    self.paid.updated_on(1.week.ago).each do |o|
      MailingsWorker.asynch_deliver_order_aftersale_to_inquire(:order_id
=> o.id)
    end
  end

=============
I may be doing things wrong, so please let me know if I am.

Thanks!
Ramon Tayag

On Wed, Sep 16, 2009 at 3:34 AM, Ramon Tayag <ramon.tayag at gmail.com>
wrote:
> I put an expectation test, but it complains two things are different
> when it prints it out in the error in the exact same way:
> http://pastie.org/618481
Printing is printing. Time is _not_ time. The problem is that the
times are off by milliseconds that aren''t accounted for in time.to_s.

The usual solution is to either stub Time.now or introduce a time
generator, but you''re not using Time.now, so that won''t work.
Options
include:

1. examine the submitted argument directly

@orders.should_receive(:updated_on) do |actual|
  expected = 1.week.ago.to_i
  actual.to_i.should be_between(expected - 1000, expected)
end

That will expect the time within 1 second of 1.week.ago.

2. introduce a custom argument matcher

class Roughly
  def initialize(expected)
    @expected = expected.to_i
  end

  def ==(actual)
    actual.to_i.between?(@expected - 1000, @expected)
  end
end

def roughly(expected)
  Roughly.new(expected)
end

@orders.should_receive(:updated_on).with(roughly(1.week.ago))

And for extra credit, you can even make this one more flexible and
provide a lower and upper bound using a fluent interface:

@orders.should_receive(:updated_on).with(within(1.second).of(1.week.ago))

HTH,
David
>
> or below:
>
> ## the test
> ?describe ".process_aftersale" do
> ? ?before do
> ? ? ?@orders = [Order.make, Order.make]
> ? ? ?Order.stub!(:paid).and_return(@orders)
> ? ? ?@orders.stub!(:updated_on).and_return(@orders)
> ? ?end
>
> ? ?....
>
> ? ?it "should send 1 week after sale letters" do
> ? ? ?@orders.should_receive(:updated_on).once.with([1.week.ago])
> ? ? ?Order.process_aftersale
> ? ?end
>
> ? ?...
> ?end
>
> ## The error - note that the dates are the same
> 1)
> Spec::Mocks::MockExpectationError in ''Order.process_aftersale
should
> send 1 week after sale letters''
> #<Order:0xb4e8b3a4>#<Order:0xb4e69448> expected :updated_on
with
> ([Wed, 09 Sep 2009 14:59:59 HKT +08:00]) but received it with ([Wed,
> 09 Sep 2009 14:59:59 HKT +08:00])
> ./spec/models/order_spec.rb:42:
>
>
> ## the method in Order
> ?def self.process_aftersale
> ? ?self.paid.updated_on(1.week.ago).each do |o|
> ? ? ?MailingsWorker.asynch_deliver_order_aftersale_to_inquire(:order_id
> => o.id)
> ? ?end
> ?end
>
> =============>
> I may be doing things wrong, so please let me know if I am.
>
> Thanks!
> Ramon Tayag
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
>

On 2009-09-16, at 08:08 , David Chelimsky wrote:
> On Wed, Sep 16, 2009 at 3:34 AM, Ramon Tayag <ramon.tayag at
gmail.com>
> wrote:
>> I put an expectation test, but it complains two things are different
>> when it prints it out in the error in the exact same way:
>> http://pastie.org/618481
>
> Printing is printing. Time is _not_ time. The problem is that the
> times are off by milliseconds that aren''t accounted for in
time.to_s.
>
> The usual solution is to either stub Time.now or introduce a time
> generator, but you''re not using Time.now, so that won''t
work. Options
> include:
>
> 1. examine the submitted argument directly
>
> @orders.should_receive(:updated_on) do |actual|
>  expected = 1.week.ago.to_i
>  actual.to_i.should be_between(expected - 1000, expected)
> end
>
> That will expect the time within 1 second of 1.week.ago.
>
> 2. introduce a custom argument matcher
>
> class Roughly
>  def initialize(expected)
>    @expected = expected.to_i
>  end
>
>  def ==(actual)
>    actual.to_i.between?(@expected - 1000, @expected)
>  end
> end
>
> def roughly(expected)
>  Roughly.new(expected)
> end
>
> @orders.should_receive(:updated_on).with(roughly(1.week.ago))
>
> And for extra credit, you can even make this one more flexible and
> provide a lower and upper bound using a fluent interface:
>
> @orders.should_receive(:updated_on).with(within(1.second).of 
> (1.week.ago))
I highly recommend a small refactoring to increase the clarity of your  
code and reduce the coupling with the system clock. I posted my  
suggestion at http://pastie.org/619142.

By doing this, you make it easy to stub/expect @orders in your  
#process_aftersale test, and make it easy to test  
#ready_for_aftersale_letter without wanting to use a stub or mock. To  
test #ready_for_aftersale_letter, simply use 1.week.ago, 8.days.ago  
and 6.days.ago as your test cases. Noticing this, you probably want to  
try 7.days.ago + 1.millisecond and make sure that behaves the way  
you''d expect, because you probably want to ensure that you use  
1.week.ago.at_midnight, and not just 1.week.ago.

Question: if I buy from you on September 15, 2009 at 17:50, when is my  
order ready for an aftersale letter? (a) September 22, 2009 at 17:50,  
(b) September 22, 2009 at 00:00, (c) September 22, 2009 at 09:00, (d)  
September 23, 2009 at 00:00, or some other time entirely? Write a test  
for that.

Take care.
----
J. B. (Joe) Rainsberger :: http://www.jbrains.ca
Your guide to software craftsmanship
JUnit Recipes: Practical Methods for Programmer Testing
2005 Gordon Pask Award for contributions to Agile Software Practice