Querying "attributes" of ResourceGroup hints misleadingly

[Yingzhe Zeng]

1. Create a stack with following template:
-----------------------------------------------------------------------------
heat_template_version: 2015-04-30
description: create cinder volume

resources:
    res-group:
        type: OS::Heat::ResourceGroup
        properties:
            count: 3
            resource_def:
                type: OS::Cinder::Volume
                properties:
                    name: volres_%index%
                    size: 1

outputs:
    attrs-outputs:
        value: { get_attr: [res-group, attributes] }
-----------------------------------------------------------------------------

2. Use "output-show" command to query outputs of this stack, as shown:
-----------------------------------------------------------------------------
  {
    "output_value": null,
    "output_error": "The Referenced Attribute (res-group attributes) is incorrect.",
    "description": "No description given",
    "output_key": "attrs-outputs"
  }
-----------------------------------------------------------------------------

3. The "output_error" message is "The Referenced Attribute (res-group attributes) is incorrect.", which is kind of misleading,
   because "attributes" is a valid attribute of ResourceGroup resource.

4. The real reason is that should configure the querying attribute of child resources of ResourceGroup, but the example above doesn't.

5. The code related to this problem is in FnGetAtt of ResourceGroup, as shown:
-----------------------------------------------------------------------------
if key == self.ATTR_ATTRIBUTES:
 if not path:
  raise exception.InvalidTemplateAttribute(
   resource=self.name, key=key)
-----------------------------------------------------------------------------

Change the raised exception here to hint users correctly, perhaps something like ChildResourceAttrNotSpecify.

[Kairat Kushaev (kkushaev)]

So I added the fix that introduces such kind of exception as subtype of Invalid TemplateAttribute:
https://review.openstack.org/#/c/209544/

[Yingzhe Zeng (zengyingzhe)]

Yes, it's already fixed. Very great.
Thanks Kushaev.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Kairat Kushaev (<email address hidden>) on branch: master
Review: https://review.openstack.org/209544
Reason: Sorry, I don't have enough time to work on this in Mitaka.
So feel free to pick up the work.