[Standards-JIG] Re: MUC presence issues

Magnus Henoch mange at freemail.hu
Fri Oct 6 13:51:33 UTC 2006


Ian Paterson <ian.paterson at clientside.co.uk> writes:

> MUC keeps the participant forever because, although the client's
> server will notice the client is offline, it will not inform MUC or
> respond to any presence probes that MUC or MUC's server send
> (because neither is either trusted by the client's server or
> subscribing to the client's presence).

How about saying that a server must bounce messages of type
"groupchat" instead of putting them in offline storage?  In that way,
the MUC service will know almost as much as the server about dead
clients.

It seems to me that this doesn't cause any privacy problems that were
not already present; you could try to use this to try to find out
whether a particular resource is online, but you can use iq:version
for the same purpose, and the client would need to defend itself in
the same way.

This would also make it possible for the MUC service to ping
participants, as clients are supposed to ignore messages containing no
understandable parts:

<message to='hag66 at shakespeare.lit/pda'
         from='darkcave at macbeth.shakespeare.lit'
         type='groupchat'>
  <ping xmlns='http://jabber.org/protocol/muc#ping'/>
</message>

-- 
Magnus
JID: legoscia at jabber.cd.chalmers.se




More information about the Standards mailing list