Re: Webfinger

On 01/30/2015 10:39 PM, James M Snell wrote:
> On Fri, Jan 30, 2015 at 1:28 PM, Harry Halpin <hhalpin@w3.org> wrote:
>> [snip]
>>
>> WebFinger maps from URIs to things like email and accounts, not the
>> other way around.
>>
>> A new solution to the "discovery" problem is out of scope I'm pretty
>> sure. However, we can keep references to WebFinger as informative if
>> some folks prefer some other solution. I haven't seen anything deployed
>> in this space besides content negotation, which is again, basically
>> never actually used for well-known reasons regarding its set-up.
>>
> 
> Harry, this isn't helpful. I've written several iterations on social
> API implementations and Discovery has always been a key element. We
> have to be able to discuss it without having someone immediately jump
> up and say it's out of scope. The ability to say, "I have this social
> identifier and I need to find out more about it" is essential.

As author of the charter, that's my job :)  Likewise,  how does
WebFinger not solve your problem?

Again, I consider things like having multiple 'standards' for the same
problem (i.e. the microformat/RDFa/microdata issue) a standardization
failure as it results in incompatible systems. Furthermore, a successful
working group requires resisting scope creep, including the ever-present
temptation to re-inventing the wheel.

If you have another solution to the Discovery problem that is superior
to WebFinger, you are more than welcome to write it down and I'm sure
we'd all be happy to discuss. However, my preference is that the
open-ended scope of the Social IG may be better suited for experimental
work and that yet another discovery mechanism should probably not go
Rec-track. We already have 3 Rec track deliverables, two of which (API
and Federation) we should probably focus on :)

   cheers,
       harry





> 
> Sando said:
>>> The indieweb reasons against webfinger are mostly not compelling for me, but a few of them are.    If we're going to use it, I'd think we should update it to be JUST a mapping from email to profile URL.  That is, http://w3.org/.well-known/wf2?email= sandro@w3.org would http redirect to http://www.w3.org/People/Sandro.   And that would be the entirely of the standard, give or take edge cases.
>>>
> 
> I'd argue against using it (webfinger) and opting for something much
> less complicated. The basic idea of starting with an email like
> identifier is fine, as is doing a GET to some well-known endpoint, but
> the result ought to just be an HTTP redirect back to some fixed point,
> just as you describe here.
> 
>   GET /.well-known/whois?id=jasnell@gmail.com HTTP/1.1
> 
>   HTTP/1.1 302
>   Location: http://jasnell.me
> 
> - James
>

Received on Friday, 30 January 2015 21:46:53 UTC