[OpenID] OpenID based on email addresses... Just Works!

Martin Atkins mart at degeneration.co.uk
Thu Oct 30 16:08:25 UTC 2008


David Fuelling wrote:
> 
> This line of reasoning doesn't make sense if you're thinking of an email 
> address as a "1st-Class" OpenID.  However, if your OpenID is really an 
> XRI or URL (only), then your email address becomes a surrogate for your 
> OpenID, or a pointer.  That's why email addresses are a special case -- 
> today they're not really OpenID's, so if we're going to start using them 
> "like" openId's, then we need to add a lot of flexibility into the 
> mechanism so that (at a domain owner's discretion, and the discretion of 
> user controling a particular email address in that domain) any 
> particular email address can be a surrogate for any of that user's OpenIDs.
> 
> Besides, with OpenID 2.0, I can use a particular URL (e.g., 
> http://openid.sappenin.com/david) but it really "maps" to my acutal 
> OpenID (http://sappenin.myopenid.com <http://sappenin.myopenid.com/>).  
> It seems odd that so many people are arguing to take away this feature 
> of OpenIDs when it comes to email addresses.
>  

In the delegation case, the URL you enter is still the 
claimed_identifier. Using the email address as the claimed identifier -- 
which can optionally delegate to another identifier -- is in fact *more* 
consistent with OpenID delegation as it exists today. And that is, in 
fact, what mart at degeneration.co.uk is currently configured to do, per my 
DNS proposal.

I think it's important to avoid changing the model when we go over to 
email addresses. Email addresses are just another URL scheme.

While OpenID does have some limitations, we should solve them on a 
broader level so that they apply to all identifier types, and keep the 
core protocol and model the same regardless of what URL scheme your 
identifier happens to use.




More information about the general mailing list