Re: [apps-discuss] WebFinger Discussion Points

["Paul E. Jones" <[e-mail]>]

On 5/3/2012 4:44 AM, Melvin Carvalho wrote:
>
>
> On 2 May 2012 13:36, Paul E. Jones <[e-mail] 
> <mailto:[e-mail]>> wrote:
>
>     On 5/2/2012 7:08 AM, Melvin Carvalho wrote:
>
>         One thing that would be a kind of joint problem statement, of
>         the main thing that's trying to be solved.
>
>
>     Is this not understood?  Keep in mind that this was not fabricated
>     last week.  I've been personally interacting with people working
>     on WebFinger for a couple of years (or more) now.  The idea is all
>     about discovery.  Given a URI, what can you learn about it?  This
>     resulted in the creation of the XRD spec, RFC 5785, RFC 6415 and
>     perhaps other documents along the way.  Definitely, the web
>     linking document (RFC 5988) is also an important part of this.
>
>
>         I think this started out as a way for webmail providers to
>         give auxiliary "follow your nose" style data linkage, using
>         the "well known" pattern.
>
>
>     I don't know where it all started, but I got my start with OpenID.
>      I did not like entering https://openid.packetizer.com/paulej
>     whenever I wanted to log in.  Rather, I wanted to enter
>     [e-mail] <mailto:[e-mail]>.  I was
>     referred by folks on the OpenID list to go look at WebFinger.
>      And, here we are.
>
>
> Yes I do remember the transition from Yadis to OpenID through 1.1 and 
> to 2.0.  As you say there was a marked change.
>
> Seems to be two polarizing world views.  One is to use email style 
> identifiers to describe things, the other is to use HTTP URIs to 
> describe things.  The UI choice need not necessarily influence the 
> backend architecture.  For example, facebook open graph uses HTTP URIs 
> to define a profile, but allows login with an email style identifier.

Yeah, I agree with you entirely.  This is a topic for the OpenID list, 
perhaps, but I had no objection to having OpenID use a URI.  I just 
didn't want to enter it by hand :-)  Thus, using WebFinger as a 
discovery protocol to facilitate the mapping of a email-style address 
into an OpenID URI made a lot of sense.

>
>
>         It's sort of morphed a bit into a few different things, such
>         as a generic discovery method for the net, a new uri scheme,
>          an account identity system for the net, serialization formats.
>
>
>     I don't think it has morphed.  These were things that needed to be
>     defined to realize the vision.
>
>
>         All these things *can* be important, but there's already
>         solutions in many places, and overlap with existing
>         technology.  The recent WWW 2012 conference has shown that
>         structured data is now incoporated in somewhere between 25%
>         and 32% of the web, eg using RDFa and microdta.  Do we think
>         it's productive to push yet another data lookup format?
>
>         http://www.bbc.co.uk/blogs/researchanddevelopment/2012/04/notes-from-the-www12-conferenc.shtml
>
>         It would be awesome if we can main pain point, or problem
>         statement that these specs can solve, find out what the easy
>         wins are, solve those with minimum fuss, imho.
>
>
>     I'd argue there is nothing else in existence that can do what
>     WebFinger does, modulo the competing SWD draft. That said, I think
>     there is agreement to move forward with WebFinger to provide a
>     single solution.
>
>
> I dont have any problem with moving forward with the WF draft, perhaps 
> taking the best features from both specs, is a great idea.  I'm just 
> interested on whether the combined problem statement will change, and 
> the extent to which, there will be any focus on email based lookup.
>
> As an aside, in answer to your specific point.  Im curious, are you 
> saying that SPARQL, which has been a W3C rec, for a number of years, 
> is incapable of a query such as the ones covered by SWD or WF?

I'm not sure if it is capable or not, but I get the impression that 
SPARQL was not designed for that purpose.  I suppose if there is a way 
to query for multiple data elements and the entirety of what one might 
want to query is located at a known location, perhaps it could.  
WebFinger is intended to allow on to issue a query for any URI against 
the associated domain, including acct:, mailto:, https:, etc.  And the 
response is a simple set of link relations.  The process is so trivial I 
can implement a "server" using static files and use "curl" to fetch the 
data set.  Can SPARQL get information for any URI as WebFinger can?  For 
certain, if it can it does so with a lot more complexity.  I would need 
to see some examples, since those in the spec don't seem to be aligned 
with what WF or SWD do.

>
>     Are you arguing for a different data format?  I think it's worth
>     highlighting the simplicity and beauty of the XRD/JRD
>     representations.  These formats primarily include a set of link
>     relations.  The link relations are the same syntax and possible
>     values as one might use in the "Link" header in HTTP and <link>
>     tags in HTML.  So, WebFinger ties in very nicely with the Web.
>
>
> Not trying to form opinion here.  Just to understand.  In structured 
> data, things can change pretty quickly, and we're all trying to hit a 
> moving target.  I do think it's sometimes helpful to look at what 
> other technologies are starting to gain adoption, in order to get a 
> bigger picture of the whole landscape, or how it might look in a few 
> years.

This is one reason I think XRD/JRD are good, too.  One might represent 
data in a variety of formats (e.g., Portable Contacts vs. hCard vs. 
vCard vs. xcard).  WF does not dictate the format in which useful 
information is presented.  What it does is merely provide a pointer to 
that information.  It's that simplicity and data independence which I 
find appealing, as I can continue to use WF long after formats holding 
user data become obsolete.

Paul