Wierd I've just been fighting with LiveJournal to implement their
OpenID correctly - which got me into thinking how are OpenID and XMPP
going to co-exist.
Then I realised that it should be possible to build an OpenID server that
should be able to use XEP-0070 as a backend.
So your process flow would be:
- Enter your webpage's address as your OpenID on a website that
allows OpenID for auth,
- WebSite reads your webpage, and determines OpenID server address,
- The OpenID server sends a XEP-0070 request to your XMPP client,
- You auth the request with your client,
- and log in to the website using your OpenID!
And the OpenID server doesn't really even need any backend storage.
On 10/28/06, Magnus Henoch <[EMAIL PROTECTED]> wrote:
In theory, acting as a client should mean that responses are
automagically sent to the correct web server thread, as the user would
reply to the same full JID which sent the original stanza. In
practice, some clients (e.g. Gajim) reply to the bare JID. As my
script doesn't send any presence, it won't get such messages...
Psi doesn't like the comma in the resource name (I think it allows
multiple recipients seperated with a comma). You could use date('c')
or date('U') instead, or why not login with SASL plain and allow the
server to generate a random resource instead.
Or better yet, why not login with SASL Anonymous, that way even Gajim
will work :-)
What do you think about using type='chat'? The XEP doesn't dictate
which message type is used, and I think chat is kinder on the eyes.
--
- Norman Rasmussen
- Email: [EMAIL PROTECTED]
- Home page: http://norman.rasmussen.co.za/
