I posted this to the specs-ab list earlier today.

Links for those that haven't looked yet.
 
https://browserid.org/
http://arstechnica.com/web/news/2011/07/mozillas-browserid-aims-to-simplify-authentication-on-the-web.ars
 
They are using asymmetrically signed JWT with an introspection endpoint.
 
There are limitations on attributes, identifiers and other serious issues with 
what Mozzila is proposing.
 
Though it is relatively close to what Nat and I were thinking with 
asymmetrically signed id_tokens, and a introspection endpoint.
 
In some ways our flow would be simpler if the id_tokens were always 
asymmetrically signed and anyone not supporting that uses the introspection 
endpoint, as they propose.
 
If the RP doesn't understand asymmetric signatures it just throws to the 
introspection endpoint.  
The big advantage is for smart clients.  They would not need to manage shared 
secrets to validate tokens.
 
For a smart client I suppose that you could let it generate it's own access 
tokens if those access tokens are JWT and they wrap a JWT containing the 
client's public key and some scope constraints etc.   In principal that could 
lower the IdP's authorization load.  It could also be a way to prevent the IdP 
from knowing who the RP is in the simple SSO case.

If the browser supports asymmetric keys securely (they are using html5 local 
storage keyed to a trusted domain) you could have the smart client provide it's 
public key to the OP and have a assertion without an audience generated and 
signed.   The client would then over-sign with an audience.  (some potential 
size issues with double base46 encoding)
 
Just some things to think about.
 
John B.


On 2011-07-16, at 9:25 AM, David Recordon wrote:

> Thoughts?
> 
> http://identity.mozilla.com/post/7669886219/how-browserid-differs-from-openid
> _______________________________________________
> specs mailing list
> [email protected]
> http://lists.openid.net/mailman/listinfo/openid-specs

_______________________________________________
specs mailing list
[email protected]
http://lists.openid.net/mailman/listinfo/openid-specs

Reply via email to