Undefined const regression on master

[Blacksmoke16]

It seems #11208 caused a regression in master, for example from nightly CI:

src/components/framework
  Showing last frame. Use --error-trace for full trace.

  In src/components/framework/spec/controllers/query_param_controller.cr:74:7

   74 | @[ATHR::Time::Format(format: "%Y--%m//%d  %T")]
          ^-----------------
  Error: Error: undefined constant ATHR::Time::Format

https://github.com/athena-framework/athena/actions/runs/8682447817/job/23806982620

Best I can tell its because the annotation is defined via a nested macro?

macro configuration_annotation(name, *args, &)
  annotation {{name.id}}; end
end

struct MyApp::ValueResolvers::Time
  macro configuration(name, *args)
    configuration_annotation ::{{@type}}::{{name.id}}
  end

  configuration Format
end

class MyClass
  @[MyApp::ValueResolvers::Time::Format(format: "%Y--%m//%d  %T")]
  def time
    {{pp @def.annotations}}
  end
end

pp MyClass.new.time

On 0efbf53cc2762431146eb028de95691cac7db039:

$ ccrystal test.cr
Using compiled compiler at .build/crystal
Showing last frame. Use --error-trace for full trace.

In test.cr:14:5

 14 | @[MyApp::ValueResolvers::Time::Format(format: "%Y--%m//%d  %T")]
        ^----------------------------------
Error: undefined constant MyApp::ValueResolvers::Time::Format

On commit before that:

$ crystal test.cr
[@[MyApp::ValueResolvers::Time::Format(format: "%Y--%m//%d  %T")]]
nil

[straight-shoota]

The issue seems to be that name.id drops the global prefix.

macro foo(path)
  {% p path.id %} # => Foo
end

foo ::Foo

[straight-shoota]

So I guess running athena CI against Crystal nightly has already payed off 🚀 😄

[crysbot]

This issue has been mentioned on Crystal Forum. There might be relevant details there:

https://forum.crystal-lang.org/t/having-hard-time-trying-to-maintain-my-library/7118/4