[OpenID] What about Logout?

Luke Shepard lshepard at facebook.com
Wed Apr 8 23:53:52 UTC 2009

I'm not familiar with "Session Sync", but it sounds like you're asking if Facebook contacts each of the RPs to tell them that the user is now logged out?

The answer is no - on logout, Facebook just deletes its Facebook cookies. It's up the RPs (applications, in our terminology) to check with Facebook at their next convenience. In practice this means that there's usually a lag of about one page load. This isn't great, but it's a reasonable compromise for the user experience.

On 4/8/09 12:14 PM, "Martin Atkins" <mart at degeneration.co.uk> wrote:

Am I right in thinking that in Facebook Connect's case this works with
the "Session Sync" functionality so that once you log out of Facebook
you're immediately logged out of sites that make use of session sync?

In order to do this effectively with OpenID we'd need an equivalent
Session Sync system.

Luke Shepard wrote:
> I agree, logout seems to be more and more important for a full single
> sign-on / sign-out experience. We found with Facebook Connect that we
> had to offer RPs the ability to log the user out of Facebook, for
> consistency.
> Consider this: the user goes to the RP, clicks the "login" button, and
> then a popup comes up onto their OP. The user happily enters their
> credentials, popup closes, and they're in. Great! Then they hit "logout"
> on the site they're on, and go on their way. But if this is a shared
> terminal, then they still have a cookie onto their OP, which leaves them
> exposed. A better solution would be to let the RP log them out of their
> provider.
> There are workarounds, some of which were suggested by Allen in previous
> threads - for instance, having a short cookie timeout, trying to detect
> recent activity, etc, but none are quite as clean as a solid logout trick.
> I think it would be relatively easy to add to the next spec. We could
> add an additional mode or two - say, "logout_setup" or
> "logout_immediate". They would be behave the same as checkid_immediate
> and checkid_setup, except in reverse - the RP must supply the correct
> user credentials, and the OP can then log them out and return only
> "success" or "failure".
> On 4/8/09 7:05 AM, "Santosh Rajan" <santrajan at gmail.com> wrote:
>     If an RP wants to logout the user not only from his site, but also
>     from the
>     OP, there is no easy way for him to do it. Currently it is a pain.
>     He needs
>     to figure how to log out from each OP himself, while most OP's
>     havent even
>     documented this.
>     Eg. This is the Google Logout URL.
>     https://www.google.com/accounts/Logout
>     This is Yahoo's undocumented Logout URL.
>     https://login.yahoo.com/config/login?logout=1
>     Maybe we need to address this in 2.1? Like the OP may provide the
>     Logout URL
>     in the discovery itself along with the endpoint URL?
>     --
>     View this message in context:
>     http://www.nabble.com/What-about-Logout--tp22951181p22951181.html
>     Sent from the OpenID - General mailing list archive at Nabble.com.
>     _______________________________________________
>     general mailing list
>     general at openid.net
>     http://openid.net/mailman/listinfo/general
> ------------------------------------------------------------------------
> _______________________________________________
> general mailing list
> general at openid.net
> http://openid.net/mailman/listinfo/general

general mailing list
general at openid.net

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-general/attachments/20090408/755cb574/attachment.htm>

More information about the general mailing list