Test that you can reopen nested exposures

[Nerian]

Related to #62

The test was actually not right. The exposures inside the :user exposure in the root entity Person are getting overwritten in the children.

class Person < Grape::Entity
##This gets overwritten by the exposure in Student. So there is no `:in_first` exposure.
  expose :user do
    expose(:in_first) { |_| 'value' }
  end
end

class Student < Person
  expose :user do
    expose(:user_id) { |_| 'value' }
    expose(:user_display_id, as: :display_id) { |_| 'value' }
  end
end

I corrected the test to reproduce the issue.

[AMar4enko]

So, this is expected behavior, then nested exposure in subclass merged with parent class?

[dblock]

Looks suspicious. Why would exposures merge with the exposure of a parent class? If you're going to override a method in a class hierarchy, you'd expect that the parent method is not called unless you do something like super, no?

[Nerian]

That's an interesting thought. Maybe we need a super – or merge: true – option?

[dblock]

I think anything that says super will be clear(er).

[Nerian]

Can we merge this? The old behaviour was merging with parent, so this fixes a regression bug. We can change the behaviour on a different PR.

[dblock]

I will merge this since it's a regression. @Nerian would you be so kind to open a new issue for changing this behavior?

[Nerian]

Sure thing. I created it here: #64