Re: [apps-discuss] WebFinger payload as array or object

["Markus Lanthaler" <[e-mail]>]

Melvin brought this discussion to my attention. So let me jump in here.


On Friday, November 30, 2012 12:07 PM, Mike Kelly wrote:
> 
> This isn't bikeshedding. It's a fundamental part of the interface
> you're presenting to clients. Having picked JSON.. "we want to copy
> html" is a weak reason to not make the most of JSON's object model
> here; particularly when doing so equally expressive, carries no
> apparent drawbacks, and is significantly more intuitive to deal with
> as the links can be addressed using standard JSON traversal.

I completely agree with Mike on this. Developers definitely want to use
standard JSON traversal instead of having to use filters for this basic use
case -- at least that was a common feedback we got when designing JSON-LD.

To simplify client development even further I would also require that the
value of such a rel property is always an array, even if there's just a
single item; just as James proposed [1] in "Plan C":

"links" : {
  "rel1": [
     { "href" : "http://example/1", "type" : "text/plain" },
     { "href" : "http://example/2", "type" : "text/plain" }
  ],
  "rel2": [
    { "href" : "http://example/3", "type" : "application/json" }
  ]
}

This gives you a nice, uniform data structure and if you want, you can rely
on ordering at the array level.


Btw. Have there been discussions about the media type of such a response? I
haven't been able to find anything and both the current draft and RFC6415
seem to be using application/json. I think it would be much better to create
a specific media type such as application/webfinger+json for such documents.


Cheers,
Markus


[1] http://www.ietf.org/mail-archive/web/apps-discuss/current/msg08170.html
[2] https://tools.ietf.org/html/draft-ietf-appsawg-webfinger-06
[3] http://tools.ietf.org/html/rfc6415
[4] http://tools.ietf.org/html/draft-wilde-profile-link-04



--
Markus Lanthaler
@markuslanthaler