[Standards-JIG] RFC 3921 Better User Presence Experience (Implementation Detail)
mridul at sun.com
Thu Oct 26 10:46:51 UTC 2006
I think I did not clarify a point well enough I guess : mentioned it ,
Remote server will be maintaining the list of directed presence updates
for its user : which will not be available at the local server.
Mirroring this would on the local server will have significant cost and
reverse engineering effort (roster sync , etc).
As in , remote user r1 sends directed presence to local user lu1 - only
remote server will know if this is a directed presence or just part of
So , if you have to mirror this at the local server , you will need to
cache all n x m presence updates from that server and invalidate all of
them with unavailable presence to the n local users as soon as the
remote server goes mia - very expensive imo.
Matthias Wimmer wrote:
> Mridul schrieb:
>> You have s2s connections being dropped frequently - cos of inactivity ,
>> configured timeouts , etc.
>> So losing of an s2s connection does not imply that the remote server is
>> gone - just that the socket has been closed.
> I think JD only proposed, that presences should set to unavailable only
> if stanzas cannot be delivered (i.e. NOT on a closed connection).
> That is if the server tries to connect to another server but is unable
> to do so.
> Tot kijk
> BTW: Not sure yet on what I think about that.
More information about the Standards