[Standards-JIG] JEP-0045 (MUC) - IQ Stanza Semantics

Richard Dobson richard at dobson-i.net
Thu Feb 19 09:23:45 UTC 2004


>     In order to exit a multi-user chat room, an occupant sends a presence
stanza of type
>     "unavailable" to the room at service/nick it is currently using in the
room.
>
>     The service then sends presence stanzas of type "unavailable" from the
departing occupant's
>     room JID to the full JIDs of the departing occupant and of the
remaining occupants:
>
>
> When user enters the room he receives presences from hist and current
> occupants' room JIDs:
>
> <presence to="..." from='room at server/nick1'/>
> <presence to="..." from='room at server/nick2'/>
> <presence to="..." from='room at server/nick3'/>
> <presence to="..." from='room at server/him'/>
>
> When in the room he receives presences from users entering and leaving
> the room:
>
> <presence to="..." from='room at server/nick2' type='unavailable'/>
> <presence to="..." from='room at server/nick4'/>
>
> When it leaves the room he receives only his own unavailable presence:
>
> <presence to="..." from='room at server/him' type='unavailable'/>
>
> So his client or any entite in the middle may consider
'room at server/nick1',
> 'room at server/nick3', 'room at server/nick4' still online.

Why would it?, surely when leaving the room the client will forget the
presences of anyone in that room and servers shouldnt be caching that data
anyway, and if servers are caching presences it will break far more than MUC
room leaving you wouldnt be able to join the room in the first place, as if
the server is caching people will recieve the presence of the room from
their servers cache and the presence they send will never get to the MUC
server as reducing the amount of presences requests leaving the server would
surely be the primary objective of caching, so it seems to me that any form
of caching is incompatible with MUC so I wouldnt worry about it as I very
much doubt people will implement caching in this way, and if they do they
will find out more than MUC breaks I expect.

Richard




More information about the Standards mailing list