Google's OpenID and discovery protocol

diff --git a/doc/forum/google_openid_broken__63__.mdwn b/doc/forum/google_openid_broken__63__.mdwn
index 68b44f2c1e94252a040ebd6f0b5f4b0836d87e63..0e41d4cede753f9108c0c743f6ddaa9532830e4d 100644 (file)
--- a/doc/forum/google_openid_broken__63__.mdwn
+++ b/doc/forum/google_openid_broken__63__.mdwn
@@ -50,3 +50,11 @@ The openid is
 <https://www.google.com/accounts/o8/id?id=AItOawltlTwUCL_Fr1siQn94GV65-XwQH5XSku4>
 (what a mouthfull!), and I don't know who that is or how to use it since it
 points to a fairly useless xml document, rather than a web page. --[[Joey]]
+
+> That string is what's received via the discovery protocol. The user logging in with a Google account is not supposed to write that when logging in, but rather <https://www.google.com/accounts/o8/id>. The OpenID client library will accept that and redirect the user to a sign in page, which will return that string as the OpenID. It's not really usable as an identifier for edits and whatnots, but an alternative would be to use the attribute exchange extension to get the email address and display that. See <http://code.google.com/apis/accounts/docs/OpenID.html#Parameters>.
+
+> Yahoo's OpenID implementation works alike, but I haven't looked at it as much. It uses <https://me.yahoo.com/> to receive the endpoint.
+
+> I've added buttons that submit the two above URLs for logging in with a Google and Yahoo OpenID, respectively, to my locally changed OpenID login plugin.
+
+> Using the Google profile page as the OpenID is really orthogonal to the above. --[[kaol]]