Re: AS 2.0 Test Suite Approach

On 03/31/2015 05:14 PM, James M Snell wrote:
> I've been giving some thought to how we can design and develop a test
> framework for the Activity Streams 2.0 format. Here are some initial
> thoughts.
> 
> Activity Streams 2.0 is fundamentally about documenting and describing
> Activity. To that end, ensuring that implementations are capable of
> consistently documenting and describing specific kinds of Activities
> is critical. So, an individual test case would take the form of a
> Short series of one or more activity statements plus supporting data
> that an implementation would be expected to produce in AS2 form. A
> second implementation ought to be able to pull in that serialization
> and be capable of answering certain specific questions about the
> Activities that have occurred.
Sounds like path worth to explore!


> 
> The following is an example:
> 
> Story 1: SWAT0
> 
> 1. With his phone, Dave takes a photo of Tantek and uploads it using a service
> 2. Dave tags the photo with Tantek
> 3. Tantek gets a notification on another service that he's been tagged
> in a photo
> 3. Evan, who is subscribed to Dave, sees the photo on yet another service
> 4. Evan comments on the photo
> 5. David and Tantek receive notifications that Evan has commented on the photo
> 
> There are six distinct Activities in this story.
> 
> A. Dave takes a photo of Tantek
> B. Dave adds the photo to a collection
> C. Dave tags Tantek in the photo
> D. Tantek and Evan are notified about the photo
> E. Evan comments on the photo
> F. Dave and Tantek are notified about the comment
1) The notifications part sound for me more like federation plumbing not
the actual social interaction.
2) One could model A and B as single activity - *Post*(Publish) a photo
to an Album. Relevant github issue -
https://github.com/jasnell/w3c-socialwg-activitystreams/issues/88


> 
> Ignoring the multiple-platforms aspect of this story, an
> implementation describing these activities should produce a set of AS2
> Activity statements that accurately describe the story. Assume that
> Dave, Tantek and Evan have user identifiers of
> "http://dave.example.org", "http://tantek.example.org", and
> "http://evan.example.org", respectively.
> 
> A. {
>   "@context": "http://www.w3.org/ns/activitystreams",
>   "@type": "Create",
>   "published": "2015-03-31T12:34:56Z",
>   "actor": {
>     "@type": "Person",
>     "@id": "http://dave.example.org",
>     "displayName": "Dave"
>   },
>   "object": {
>     "@type": "Image",
>     "@id": "http://i.example.org/images/abc.jpg",
>     "url": {
>       "mediaType": "image/jpeg",
>       "href": "http://i.example.org/images/abc.jpg"
>     }
>   }
> }
Do we talk here about offline activity of 'Taking a photo' or activity
of posting/publishing it online? (which may happen later in a future)


> 
> B. {
>   "@context": "http://www.w3.org/ns/activitystreams",
>   "@type": "Add",
>   "published": "2015-03-31T12:45:56Z",
>   "actor": {
>     "@type": "Person",
>     "@id": "http://dave.example.org",
>     "displayName": "Dave",
>   },
>   "object": {
>     "@type": "Image",
>     "@id": "http://i.example.org/images/abc.jpg",
>     "url": {
>       "mediaType": "image/jpeg",
>       "href": "http://i.example.org/images/abc.jpg"
>     }
>   },
>   "target": {
>     "@type": "Album",
>     "@id": "http://i.example.org/images",
>     "displayName": "Dave's Photos"
>   }
> }
Possible alternative (assuming Post == Publish)
{
  "@context": "http://www.w3.org/ns/activitystreams",
  "@type": "Post",
  "published": "2015-03-31T12:45:56Z",
  "actor": {
    "@type": "Person",
    "@id": "http://dave.example.org",
    "displayName": "Dave",
  },
  "object": {
    "@type": "Image",
    "@id": "http://i.example.org/images/abc.jpg",
    "url": {
      "mediaType": "image/jpeg",
      "href": "http://i.example.org/images/abc.jpg"
    }
  },
  "target": {
    "@type": "Album",
    "@id": "http://i.example.org/images",
    "displayName": "Dave's Photos"
  }
}

> 
> C. {
>   "@context": "http://www.w3.org/ns/activitystreams",
>   "@type": "Add",
>   "published": "2015-03-31T12:56:56Z",
>   "actor": {
>     "@type": "Person",
>     "@id": "http://dave.example.org",
>     "displayName": "Dave",
>   },
>   "object": {
>     "@type": "Mention",
>     "href": "http://tantek.example.org"
>   },
>   "target": {
>     "@type": "Image",
>     "@id": "http://i.example.org/images/abc.jpg",
>     "url": {
>       "mediaType": "image/jpeg",
>       "href": "http://i.example.org/images/abc.jpg"
>     }
>   }
> }
We could distinguish *mentioning* in the accompanying text from
*tagging* on a photo.

> 
> D. {
>   "@context": "http://www.w3.org/ns/activitystreams",
>   "@type": "Announce",
>   "published": "2015-03-31T13:34:56Z",
>   "actor": {
>     "@type": "Person",
>     "@id": "http://dave.example.org",
>     "displayName": "Dave"
>   },
>   "object": {
>     "@type": "Image",
>     "@id": "http://i.example.org/images/abc.jpg",
>     "url": {
>       "mediaType": "image/jpeg",
>       "href": "http://i.example.org/images/abc.jpg"
>     },
>     "attachment": {
>       "@type": "Mention",
>       "href": "http://tantek.example.org"
>     }
>   },
>   "to": {
>     "@type": "Person",
>     "@id": "http://tantek.example.org",
>     "displayName": "Tantek"
>   },
>   "cc": {
>     "@type": "Person",
>     "@id": "http://evan.example.org",
>     "displayName": "Evan"
>   }
> }
once again, looks bit like a plumbing and i don't feel convinced that we
want to take such path...


> 
> E. {
>   "@context": "http://www.w3.org/ns/activitystreams",
>   "@type": "Create",
>   "published": "2015-03-31T13:45:56Z",
>   "actor": {
>     "@type": "Person",
>     "@id": "http://evan.example.org",
>     "displayName": "Evan"
>   },
>   "object": {
>     "@type": "Note",
>     "@id": "http://c.example.org/notes/1",
>     "content": "What a great photo",
>     "inReplyTo": "http://i.example.org/images/abc.jpg"
>   }
> }
again Create vs. Post/Publish ?

> 
> F. {
>   "@context": "http://www.w3.org/ns/activitystreams",
>   "@type": "Announce",
>   "published": "2015-03-31T13:56:56Z",
>   "actor": {
>     "@type": "Person",
>     "@id": "http://evan.example.org",
>     "displayName": "Evan"
>   },
>   "object": {
>     "@type": "Note",
>     "@id": "http://c.example.org/notes/1",
>     "content": "What a great photo",
>     "inReplyTo": "http://i.example.org/images/abc.jpg"
>   },
>   "to": [
>     {
>       "@type": "Person",
>       "@id": "http://tantek.example.org",
>       "displayName": "Tantek"
>     },
>     {
>       "@type": "Person",
>       "@id": "http://dave.example.org",
>       "displayName": "Dave"
>     }
>   ]
> }
looks like one of possible federation plumbings to me once again!

> 
> Note that there are a number of acceptable variations in the way
> implementations can serialize these stories so the test case needs to
> take those into consideration. A consuming implementation would need
> to be able to answer specific questions about the stories:
> 
> 1. What did Dave create?
> 2. Who was tagged in the photo?
> 3. Who was notified about the photo?
> 4. Who commented on the photo?
> 5. What was the text of the comment?
> 6. Who was notified about the comment?
> 7. When was the photo created?
> 8. When was Tantek tagged in the photo?
> 9. When were Tantek and Evan tagged?
yes! we *should* test our data model by querying it :)

> 
> This example case can be improved but it illustrates the point: Given
> a set of commonly understood Activity statements, producing
> implementations would be expected to produce consistent AS2.0
> serializations that consuming implementations would be capable of
> consistently handling.
> 
> The downside is that this approach can become fairly complicated but
> there's really no other way we can do it.
> 
> Thoughts?
thanks a lot James for putting out this proposal, i find it very
concrete and helpful to get started!

Cheers


> 
> - James
>