RSpec failed_with_reason

[inkstak]

Hi @palkan

I'm starting to use ActionPolicy reasons and it's very helpful to debug policies when used with RSpec DSL :

class FooPolicy < ActionPolicy::Base
  def manage?
    deny!(:reason_a) unless condition_a?
    deny!(:reason_b) unless condition_b?
    true
  end
end

describe FooPolicy do
  describe_rule :manage? do
    context "when condition A is not met" { failed }
    context "when condition B is not met" { failed }
    context "when all conditions are met" { succeed }
  end
end

1) FooPolicy#manage? when all conditions are met
Failure/Error:  ...

Expected to succeed but failed:
<FooPolicy#manage?: false (reasons: {:"foo"=>[:reason_b]})>

Now, I'm thinking it will also be helpful to ensure a test failed for the right reason :

describe FooPolicy do
  describe_rule :manage? do
    context "when condition A is not met" { failed_with_reason(:reason_a) }
    context "when condition B is not met" { failed_with_reason(:reason_b) }
    context "when all conditions are met" { succeed }
  end
end

1) FooPolicy#manage? when condition B is not met
Failure/Error:  ...

Expected to failed with `:reason_b` but actually failed for another reason:
<FooPolicy#manage?: false (reasons: {:"foo"=>[:reason_a]})>

When rules becomes more complicated, it allows to identify false-positives.
What do you think ?

I'm already starting to prepare a PR.

[inkstak]

Hi @palkan

The PR is on its way : master...inkstak:action_policy:feature-rspec_failed_with_reason
I still need to write some tests and documentation.

[palkan]

Hey @inkstak,

Yeah, this feature can be useful. I'm not sure about the API though. I'd like to avoid adding more DSL methods; maybe, we can incorporate something into existing ones? A quick idea is as follows:

describe FooPolicy do
  describe_rule :manage? do
    failed "when condition A is not met", reason: :reason_a

    # for nested policies
    failed "when condition A is not met", reason: {bar: :reason_b}
  end
end

WDYT?