2.99 does not warn properly when using an inner describe :symbol

[myronmarston]

See rspec/rspec-rails#1415 for an example user report. 2.99 should warn for this case but it does not. The problem is here:

https://github.com/rspec/rspec-core/blob/v2.99.2/lib/rspec/core/metadata.rb#L184

It still treats symbols as a special case like strings. Compare to:

https://github.com/rspec/rspec-core/blob/v3.3.1/lib/rspec/core/metadata.rb#L298

...which does not treat symbols as a special case.

[xaviershay]

Appropriate warning for this would be https://github.com/rspec/rspec-core/blob/v2.99.2/lib/rspec/core/metadata.rb#L139, right?

[xaviershay]

A warning is triggered currently only if subject is used. So the following will issue a warning (and we have a test case for exactly this):

describe String do
  describe :create do
    it 'works' do
      subject
    end
  end
end

But if you remove subject it does not warn... but also it works just fine in RSpec 3. So I see a couple of options here:

1. Minimal repro provided in Upgrade from 2.99 to 3.3 is causing controller test failures rspec-rails#1415 is not correct and this is actually a different problem. @jeffheifetz might be too long ago for you to care/remember, but tagging you in case some other context occurs to you.
2. rspec-rails is doing something weird to trigger it. I'm not sure how to get a set up to test this within a reasonable amount of time.

If no-one has any bright ideas, I recommend closing this as "cannot reproduce".

@myronmarston @JonRowe

(Thanks to @SalvatoreT who helped me look into this.)

[JonRowe]

rspec-rails is doing something weird to trigger it. I'm not sure how to get a set up to test this within a reasonable amount of time.

How about described_class? Thats what we're using to signal to the Rails stuff which the controlled is, would also change with a nested describe, and might not be triggering the warning.

[xaviershay]

https://github.com/rspec/rspec-core/blob/2-99-maintenance/spec/rspec/core/example_group_spec.rb#L360 suggests that described_class warnings are working as intended.

[myronmarston]

I was on vacation Wednesday until today so I haven't had a chance to look into this yet. I hope to in the next day or so.

[myronmarston]

I think you're right that this is handled properly. My confusion was comparing https://github.com/rspec/rspec-core/blob/v2.99.2/lib/rspec/core/metadata.rb#L184 vs https://github.com/rspec/rspec-core/blob/v3.3.1/lib/rspec/core/metadata.rb#L298 -- it looks like 2.99 is wrongly treating Symbol as a special case in that method. Looking more into it, the Symbol case is handled here instead:

https://github.com/rspec/rspec-core/blob/v2.99.2/lib/rspec/core/example_group.rb#L327-L331

I agree with closing this, barring a clear repro case of a situation where 2.99 does not warn but should.