Uniquely Identifying User

[Chris Messina]

This is probably why the Social Graph Protocol normalizes everything to
sgn:// URIs for canonicalization:

The Social Graph Node Mapper is a community project to build a portable
library to map social networking sites' URLs to and from a new canonical
form (sgn:// URLs).

For example, the following URLs are all the same person:


   - http://brad.livejournal.com/
      - http://brad.livejournal.com/data/foaf
      - http://brad.livejournal.com/data/rss
      - http://brad.livejournal.com/data/atom
      - http://brad.livejournal.com/profile
      - http://www.livejournal.com/userinfo.bml?user=brad
      - etc

With sgnodemapper, all of those would map to *sgn://
livejournal.com/?ident=brad* . And then, from the sgn URL, you can map back
to http:// URLs for any known type: content page, profile page, RSS, Atom,
etc.

http://code.google.com/apis/socialgraph/docs/canonical.html
http://code.google.com/p/google-sgnodemapper/

It's still domain-based identity, but it allows multiple representations of
identifiers to be normalized more easily... this requires maintaining a huge
mapping table, but attempts to be more pragmatic about the form of the
identifier...

On Wed, Nov 3, 2010 at 8:32 AM, David Recordon <recordond at gmail.com> wrote:

> Spec draft is attached. As a side note, we should make sure that anything
> contributed to the Foundation's GitHub account is treated in the same manner
> as the Foundation's SVN server.
>
> Agreed with John about this being bound to the domain.
>
> I'm not convinced that splitting the id and domain is that much better than
> using an acct URI.
>
> --David
>
>
> On Wed, Nov 3, 2010 at 6:38 AM, Nat Sakimura <sakimura at gmail.com> wrote:
>
>> At the OpenID Summit, there were some discussion about how to uniquely
>> identify the user.
>>
>> There were some argument that it should user 'user_id' and 'domain'.
>> Upon some contemplation, I think we should use something like 'server_id'
>> which is a unique identifier (perhaps domain, but maybe UUID etc.) instead
>> of 'domain' as 'domain' may actually change.
>>
>> What do you think?
>>
>> P.S., David, could you just save your html connect proposal to a file and
>> send it to this list? Then it will constitute the contribution and  we can
>> start discussion on that formally. Otherwise, we cannot from the IPR
>> management point of view.
>>
>> --
>> Nat Sakimura (=nat)
>> http://www.sakimura.org/en/
>> http://twitter.com/_nat_en
>>
>> _______________________________________________
>> openid-specs-connect mailing list
>> openid-specs-connect at lists.openid.net
>> http://lists.openid.net/mailman/listinfo/openid-specs-connect
>>
>>
>
> _______________________________________________
> openid-specs-connect mailing list
> openid-specs-connect at lists.openid.net
> http://lists.openid.net/mailman/listinfo/openid-specs-connect
>
>


-- 
Chris Messina
Open Web Advocate, Google

Web: http://factoryjoe.com
Follow me on Buzz: http://buzz.google.com/chrismessina
...or Twitter: http://twitter.com/chrismessina

This email is: [ ] shareable [X] ask first [ ] private
I estimate the importance of this email to be: [ ] above average  [X]
average
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-connect/attachments/20101103/e8ebb70d/attachment-0001.html>