27 messages in net.openid.general[OpenID] An OpenID "mobile" Hint?
FromSent OnAttachments
David RecordonJun 4, 2008 2:08 pm 
Hans GranqvistJun 4, 2008 2:34 pm 
David RecordonJun 4, 2008 4:50 pm 
Johannes ErnstJun 4, 2008 9:49 pm 
Nat SakimuraJun 4, 2008 11:51 pm 
Martin AtkinsJun 5, 2008 12:02 am 
Kick WillemseJun 5, 2008 3:49 am 
Steven Livingstone-PerezJun 5, 2008 4:06 am 
SitG AdminJun 5, 2008 8:31 am 
Johannes ErnstJun 5, 2008 9:15 am.gif, .gif
David RecordonJun 5, 2008 9:50 am 
David RecordonJun 5, 2008 9:51 am 
Martin AtkinsJun 5, 2008 10:35 am 
SitG AdminJun 5, 2008 12:42 pm 
Martin AtkinsJun 5, 2008 1:34 pm 
SitG AdminJun 5, 2008 3:58 pm 
Nat SakimuraJun 5, 2008 6:59 pm 
Nat SakimuraJun 5, 2008 7:06 pm 
Nat SakimuraJun 5, 2008 8:36 pm 
Martin AtkinsJun 6, 2008 12:06 am 
Johannes ErnstJun 6, 2008 3:08 pm 
Warren JamisonJun 6, 2008 6:05 pm 
Carsten PötterJun 6, 2008 8:47 pm 
Brandon RamirezJun 7, 2008 10:28 am 
Brandon RamirezJun 7, 2008 10:33 am 
SitG AdminJun 7, 2008 9:22 pm 
Tan, WilliamJun 16, 2008 10:57 am 
Actions with this message:
Paste this link in email or IM:
Paste this link in email or IM:
Atom feed for this thread
Paste this URL into your reader:
Subject:[OpenID] An OpenID "mobile" Hint?Actions...
From:David Recordon (drec@sixapart.com)
Date:Jun 4, 2008 4:50:01 pm
List:net.openid.general

I think authentication mechanism would actually be in the minority of the decisions for this sort of flow, was using the YubiKey OpenID Provider as a very clear example of what won't work on a mobile phone.

Today I'm more interested in how we can increase the number of OpenID Providers that have good mobile experiences and am thinking that this sort of extension may be a tactic in achieving that goal. Even just pitting two OpenID Providers which both use passwords against each other with one having a mobile experience and the other not would be a good thing in solving this problem.

--David

On Jun 4, 2008, at 2:34 PM, Hans Granqvist wrote:

It makes sense (though the YubiKey is a mechanism, not a provider), but it's a bit dangerous since security decisions on the RP based on the User-Agent's self-issued origin/type are quite tricky.

An attacker would pick the easiest mechanism if there is a choice, too.

Regardless, the XRDS file could map accepted authentication mechanism(s) to each URL as a simple attribute.

On Wed, Jun 4, 2008 at 2:08 PM, David Recordon <drecordon at sixapart.com> wrote:

In developing a mobile application that uses OpenID for logins one of the things I've become really cognizant of is how poor of the mobile experience most Providers have when it comes to OpenID. It obviously doesn't take a lot to create a streamlined Provider flow for authentication and the trust request, but so far it seems that no one has really done that. I was also thinking more about Providers such as YubiKey where authenticating with a USB device (despite how awesome it is) won't work on my iPhone.

I'm wondering if it would be useful to write a dead simple extension to provide some hints around mobile support? Allow a Provider to advertise in an XRDS file that they support a mobile login flow so that Relying Parties could discover that theoretically making it so that I could use a Provider such as YubiKey on the desktop and then MyOpenID on the phone.

Am I barking up a useful tree? If I spec'd this would any Providers actually implement a mobile friendly flow?