Re: User stories for Social API

On 2015-02-04 03:41 PM, Erik Wilde wrote:
> ... might be an interesting read in the context of this thread.
It is! Thanks for that.
> back at the TPAC F2F, we decided that the "Social API" in the charter 
> should be interpreted to be a "RESTful HTTP API". my assumption was 
> that then it clearly should be one matching mark's "Many Clients, Many 
> Servers" category.
I agree!
> are we now backtracking to say that the API actually is something 
> different?
I don't think so.

For many of these user stories, like "user posts a note 
<https://www.w3.org/wiki/Socialwg/Social_API/User_stories#User_posts_a_note>", 
a single client would only make requests to a single server, whether in 
a monolithic social network or a federated social network.

There are also user stories which, in a monolithic model, would only 
require connections to a single server. In a federated model, to two or 
many servers. So a federated "follow a person 
<https://www.w3.org/wiki/Socialwg/Social_API/User_stories#Following_a_person>" 
story would probably have about 2 servers involved. The "social network 
analysis 
<https://www.w3.org/wiki/Socialwg/Social_API/User_stories#Social_network_analysis>" 
story might have hundreds or thousands.

If we surface the plumbing of a federated model, it makes the user 
stories more complex and harder to evaluate. We haven't agreed yet that 
we should have things like inbox streams, groups, friends lists or 
following. I hope we can wait until after we settle down on these 
fundamentals before complicating them with cross-domain security concerns.

-Evan