Has there been any development on shared helpers / it_should_behave_like feature
in rspec?
I forget the reasons, but I remember a patch for something like this was
rejected:
it_should_behave_like "an_entry", :locals => { :entry =>
Entry.new }
OR:
before do
@entry = Entry.new
end
it_should_behave_like "an_entry", :locals => lambda {
{
:entry => @entry
}
}
Is there any code which now deals with the variable passing issue in rspec?
Best,
Scott
On 23 Feb 2010, at 17:30, Scott Taylor wrote:> > Has there been any development on shared helpers / > it_should_behave_like feature in rspec? > > I forget the reasons, but I remember a patch for something like this > was rejected: > > it_should_behave_like "an_entry", :locals => { :entry => > Entry.new } > > OR: > > before do > @entry = Entry.new > end > > it_should_behave_like "an_entry", :locals => lambda { > { > :entry => @entry > } > } > > Is there any code which now deals with the variable passing issue in > rspec? > > Best, > > ScottFWIW, what I''ve done in this situation is expect that the ExampleGroup I''ve mixed the shared behaviour into to define a method, and call that from the shared examples - if it''s not defined you''ll find out pretty quickly. Another approach is to just expect the @variable to be set: describe "an entry" do before @entry or raise("You need to set @entry to use this shared example group") end end It''s not great, but it makes the shared example group a little more self-documenting than if you just shared the instance variable without the check. cheers, Matt http://mattwynne.net +447974 430184
I just use a factory method.
describe ''Authorize.net CIM gateway'', :shared => true do
describe ''saving a card'' do
describe ''preconditions'' do
it "should raise an error if the card is not saved" do
lambda {
gateway.save_credit_card(Factory.build(:credit_card, :user =>
Factory(:user)))
}.should raise_error(ArgumentError, /unsaved record/)
end
end
describe ''first card saved on an account'' do
it "should set the authorize_id on the credit card" do
credit_card = Factory :credit_card, :user => Factory(:user)
gateway.save_credit_card credit_card
credit_card.authorize_id.should be_present
end
it "should return true for a successful response" do
credit_card = Factory :credit_card, :user => Factory(:user)
gateway.save_credit_card(credit_card).should be_true
end
it "should mark the card as having a validated card code" do
card = Factory :credit_card, :card_code_validated => false
gateway.save_credit_card card
card.should be_card_code_validated
end
end
# etc etc etc etc
end
end
describe Payment::AuthorizeNetCim::FakeGateway do
def gateway
@gateway ||= Payment::AuthorizeNetCim::FakeGateway.new
end
it_should_behave_like ''Authorize.net CIM gateway''
# some extra custom examples
end
I prefer this to requiring an instance variable, because you get built in
error-handling for free. If you don''t implement the factory method,
you get "undefined method: gateway" rather than the less obvious
undefined method on NilClass error.
Also, your :locals => { :entry => Entry.new } syntax is going to be no
good because you''ll only have one instance of Entry that gets reused
across examples. No bueno. You''d have to do
:locals => lambda { { :entry => Entry.new } }
and now it''s just getting ugly.
If you think defining methods is too heavy (which I wouldn''t agree
with, but okay) then maybe use the let method?
describe Entry do
before do
let(:entry) { Entry.new )
end
it_should_behave_like "an_entry"
end
How''s that look?
Pat
On Feb 23, 2010, at 9:30 AM, Scott Taylor wrote:
>
> Has there been any development on shared helpers / it_should_behave_like
feature in rspec?
>
> I forget the reasons, but I remember a patch for something like this was
rejected:
>
> it_should_behave_like "an_entry", :locals => { :entry =>
Entry.new }
>
> OR:
>
> before do
> @entry = Entry.new
> end
>
> it_should_behave_like "an_entry", :locals => lambda {
> {
> :entry => @entry
> }
> }
>
> Is there any code which now deals with the variable passing issue in rspec?
>
> Best,
>
> Scott
>
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
On Tue, Feb 23, 2010 at 3:41 PM, Matt Wynne <matt at mattwynne.net> wrote:> > On 23 Feb 2010, at 17:30, Scott Taylor wrote: > >> >> Has there been any development on shared helpers / it_should_behave_like >> feature in rspec? >> >> I forget the reasons, but I remember a patch for something like this was >> rejected: >> >> ? it_should_behave_like "an_entry", :locals => { :entry => Entry.new } >> >> OR: >> >> ? before do >> ? ? @entry = Entry.new >> ? end >> >> ? it_should_behave_like "an_entry", :locals => lambda { >> ? ? { >> ? ? ? :entry => @entry >> ? ? } >> ? } >> >> Is there any code which now deals with the variable passing issue in >> rspec? >> >> Best, >> >> Scott > > FWIW, what I''ve done in this situation is expect that the ExampleGroup I''ve > mixed the shared behaviour into to define a method, and call that from the > shared examples - if it''s not defined you''ll find out pretty quickly. > Another approach is to just expect the @variable to be set: > > ?describe "an entry" do > ? ?before > ? ? ?@entry or raise("You need to set @entry to use this shared example > group") > ? ?end > ?end > > It''s not great, but it makes the shared example group a little more > self-documenting than if you just shared the instance variable without the > check. > > cheers, > MattWe do what Matt does, plus we document on top of the shared examples the interface the example group needs: # You need to provide an <invalid_record> method that returns a record without # saving, in an invalid state, for this mixin to work. shared_examples_for "a model that can disable validations" do context "when checking for a record''s validity" do it "always returns true if you disabled validations" do described_class.disable_validations! invalid_record.should be_valid end it "falls back to the usual behavior if you enable validations" do described_class.enable_validations! invalid_record.should_not be_valid end end ... end describe User do let :invalid_record do User.make_unsaved(:invalid) end it_should_behave_like "a model that can disable validations" end Cheers, -foca> http://mattwynne.net > +447974 430184 > > _______________________________________________ > rspec-users mailing list > rspec-users at rubyforge.org > http://rubyforge.org/mailman/listinfo/rspec-users >