rspec users - Jun 2007 - Is it still possible to use should_have_rjs

Hi

I''ve come back to a Rails project after about a month doing other  
work, and there''s a lot of changes.  I was on the 0.7 branch before  
and I''ve upgraded to RSpec 1.0.5.  I used the spec translator, fixed  
the cases where it barfed on unusual line contents, and all my model  
and controller specs now pass. But I''m getting 53 failures on my view  
specs, all along the lines of:

should have visual effect
Unknown RJS statement type effect
/Users/ashleymoran/Documents/Development/YourMoney/trunk/src/spec/ 
views/admin/add_dealer_rjs_view_spec.rb:33
31
32  it "should have visual effect" do
33    response.should have_rjs(:effect, :blind_down, "add_dealer")
34  end
35

I''m trying to revert these to should_have_rjs but I''ve looked
through
the RSpec source and it seems should_have_rjs was silently removed  
between 0.8 and 0.9.  At least I can''t find any reference to it in  
the 0.9.4 source or the 1.0.5 source.  Am I missing something?  I  
really need should_have_rjs- those 53 failures represent about 20% of  
all my view specs.

Thanks
Ashley

On 6/15/07, Ashley Moran <work at ashleymoran.me.uk>
wrote:
> Hi
>
> I''ve come back to a Rails project after about a month doing other
> work, and there''s a lot of changes.  I was on the 0.7 branch
before
> and I''ve upgraded to RSpec 1.0.5.  I used the spec translator,
fixed
> the cases where it barfed on unusual line contents, and all my model
> and controller specs now pass. But I''m getting 53 failures on my
view
> specs, all along the lines of:
>
> should have visual effect
> Unknown RJS statement type effect
> /Users/ashleymoran/Documents/Development/YourMoney/trunk/src/spec/
> views/admin/add_dealer_rjs_view_spec.rb:33
> 31
> 32  it "should have visual effect" do
> 33    response.should have_rjs(:effect, :blind_down,
"add_dealer")
> 34  end
> 35
>
> I''m trying to revert these to should_have_rjs but I''ve
looked through
> the RSpec source and it seems should_have_rjs was silently removed
> between 0.8 and 0.9.  At least I can''t find any reference to it in
> the 0.9.4 source or the 1.0.5 source.  Am I missing something?
You''re not missing anything. I switched from a direct port of ARTS to
wrapping assert_select_rjs in 0.9. Neither library covers the full
breadth of rjs statements and I went w/ the wrapper because it meant
that rspec would reap the benefits of improvements to
assert_select_rjs. Sadly, it seems as though the only improvements to
assert_select_rjs have been those that I''ve personally contributed to
Rails, and they haven''t covered every situation.

As for your immediate problem, Spec::Rails provides access to
Test::Unit assertions for free. I''d recommend that you install the
ARTS plugin (from which should_have_rjs was ported) and use it
directly as/is (assert_rjs) right in your examples. It won''t be as
pleasing syntactically, but at least it should work for you.

http://glu.ttono.us/articles/2006/05/29/guide-test-driven-rjs-with-arts

There are a couple of things that ARTS didn''t cover that we had added
to RSpec. Those you''ll have to do manually with any of these:

response.should have_text("full text of the response")
response.should have_text(/regexp representing a substring of the response/)
response.body.should == "full text of the response"
response.body.should =~ /regexp representing a substring of the response/

In terms of a longer term solution for rspec, I''m open to suggestions.
ARTS and assert_select_rjs both make assumptions about what options
are available, which means they both need to be updated as new options
appear. I had hoped that this would be maintained in rails, but it
doesn''t appear to be happening.

We could resurrect ARTS in RSpec, but that would mean either coming up
with a new matcher (include_rjs?) as not to break existing specs that
use "should have_rjs.", and it would put more of a burden on
Spec::Rails to keep up with additions to rjs.

Other ideas?

David
>  I
> really need should_have_rjs- those 53 failures represent about 20% of
> all my view specs.
>
> Thanks
> Ashley
>
>
>
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
>

On 15 Jun 2007, at 12:42, David Chelimsky wrote:
> You''re not missing anything. I switched from a direct port of ARTS
to
> wrapping assert_select_rjs in 0.9. Neither library covers the full
> breadth of rjs statements and I went w/ the wrapper because it meant
> that rspec would reap the benefits of improvements to
> assert_select_rjs. Sadly, it seems as though the only improvements to
> assert_select_rjs have been those that I''ve personally contributed
to
> Rails, and they haven''t covered every situation.
Hmm, surely RJS testing is a pretty essential part of Rails, seeing  
how everyone''s going Web2.0-crazy?  I''m surprised
assert_select_rjs
doesn''t cover everything.

> As for your immediate problem, Spec::Rails provides access to
> Test::Unit assertions for free. I''d recommend that you install the
> ARTS plugin (from which should_have_rjs was ported) and use it
> directly as/is (assert_rjs) right in your examples. It won''t be as
> pleasing syntactically, but at least it should work for you.
>
> http://glu.ttono.us/articles/2006/05/29/guide-test-driven-rjs-with- 
> arts
Thanks for the link.  I tried this:

describe "A rendered add_dealer view" do
   before(:each) do
     render ''admin/add_dealer''
   end

   it "should have visual effect" do
     assert_rjs(:effect, :blind_down, "add_dealer")
   end
end

But it doesn''t work:

should have visual effect
ActionView::Helpers::JavaScriptProxy#to_str should return String
/Users/ashleymoran/Documents/Development/YourMoney/trunk/src/ 
config/../vendor/plugins/arts/lib/arts.rb:98 :in `index''
/Users/ashleymoran/Documents/Development/YourMoney/trunk/src/ 
config/../vendor/plugins/arts/lib/arts.rb:98 :in  
`assert_response_contains''
/Users/ashleymoran/Documents/Development/YourMoney/trunk/src/ 
config/../vendor/plugins/arts/lib/arts.rb:12 :in `assert_rjs''
./spec/views/admin/add_dealer_rjs_view_spec.rb:33
96
97  def assert_response_contains(str, message)
98     assert @response.body.to_s.index(str), message
99  end
100

My naive attempt to fix it...

module ActionView
   module Helpers
     class JavaScriptProxy
       def to_str
         to_s
       end
     end
   end
end

achieves nothing.  (Probably because it''s completely wrong.)  Any  
ideas on this one?
> There are a couple of things that ARTS didn''t cover that we had
added
> to RSpec. Those you''ll have to do manually with any of these:
>
> response.should have_text("full text of the response")
> response.should have_text(/regexp representing a substring of the  
> response/)
> response.body.should == "full text of the response"
> response.body.should =~ /regexp representing a substring of the  
> response/
>
> In terms of a longer term solution for rspec, I''m open to
suggestions.
> ARTS and assert_select_rjs both make assumptions about what options
> are available, which means they both need to be updated as new options
> appear. I had hoped that this would be maintained in rails, but it
> doesn''t appear to be happening.
Hmm, surely a framework feature without a means to test it is next to  
useless?  Or maybe I am being a too demanding (I''m known for it).

> We could resurrect ARTS in RSpec, but that would mean either coming up
> with a new matcher (include_rjs?) as not to break existing specs that
> use "should have_rjs.", and it would put more of a burden on
> Spec::Rails to keep up with additions to rjs.
Is there any way to be clever with have_rjs and decide where to pass  
the request based on the type of match?  That way you could use ARTS  
as a prop and migrate test support to assert_select_rjs as and when  
things as included?  I haven''t written any custom matchers yet -  
don''t know much about the new style specs thanks to lack of time  
spent coding - so again I''m probably being naive.

Ashley