FW: Proposal: SMTP service extension for Yadis discovery

[David Fuelling]

Sorry, forgot to CC the Yadis list on my reply to Dmitry.

david

> -----Original Message-----
> From: David Fuelling [mailto:sappenin at gmail.com]
> Sent: Monday, February 05, 2007 11:46 AM
> To: 'Dmitry Shechtman'
> Cc: 'general at openid.net'
> Subject: RE: Proposal: SMTP service extension for Yadis discovery
> 
> Hey Dmitry,
> 
> See my replies inline...
> 
> Thanks!
> 
> David
> 
> > -----Original Message-----
> > From: Dmitry Shechtman [mailto:damnian at gmail.com]
> > Sent: Monday, February 05, 2007 8:22 AM
> > To: 'David Fuelling'
> > Cc: yadis at lists.danga.com; general at openid.net
> > Subject: RE: Proposal: SMTP service extension for Yadis discovery
> >
> > David, although this may be an interesting idea, Yadis (unlike SMTP)
> isn't
> > meant for host-global communications. A Yadis document normally
> > corresponds to a single user record.
> 
> Is this accurate?  Yadis is relatively new to me, so I could be completely
> wrong here, but it seems like all of the services that OpenId advertises
> via Yadis *are* host-global.  For example, it doesn't look like most OP's
> allow divergent OpenId services on a per-user basis (e.g., Jane can only
> use OpenId 1.0, but Bob can use 1.0 & 2.0).
> 
> Instead, it seems like the Yadis service document that advertises OP
> openid services *is* host-global (although this may be unintentional in
> many cases).
> 
> Am I wrong here?
> 
> If I am wrong, then there's nothing wrong with transforming an email to an
> OpenId Endpoint url (using the root domain of the email).  Then, via
> OpenId Discovery, the user-specific Yadis document could be retrieved
> based on the "transformed" OpenId endpoint URL (per the protocol today).
> 
> >
> > Furthermore, a user identifier isn't necessarily a URL. It may very well
> > be an XRI. So if your iname expires, you could theoretically tie your
> > email address to your inumber.
> 
> Isn't there already an established methodology for getting an email
> address from an XRI?  Again, that's outside of my domain expertise, so
> forgive my ignorance here.
> 
> >
> >
> > > A.) I guess I'm not clear why we would involve SMTP in this at all.
> >
> > In case you didn't read my reply to Martin, there is quite a large niche
> > of
> > users who don't feel comfortable with OpenID because it requires them to
> > use
> > a URL for an identifier. An email address is a much more familiar
> concept.
> 
> Sorry for the ambiguity.  I think my question has 2 possible meanings.  I
> meant to say that I'm not clear why one would involve SMTP services in
> attempting to do the email-to-OpenId transform.  I wasn't trying to say
> that I thought the idea was not worthwhile.  I definitely think OpenId
> needs some sort of email address integration.
> 
> >
> >
> > Regards,
> > Dmitry
> > =damnian