[Standards-JIG] Refactor: Converting a One-to-One Chat into aConference (JEP-0045)

JD Conley jconley at winfessor.com
Sun Dec 19 01:46:28 UTC 2004

> On Sun, 19 Dec 2004 10:35, JD Conley wrote:
> > The RECOMMENDED process for creating the room name is as follows:
> > "[first person JID node]_[current local unix time]_[creation
> > For example, a room created by "crone1 at shakespeare.lit/desktop" on
> > 1/1/2005 at 00:00:00 SHOULD first attempt to create a room named
> > "crone1_1104537600_1". The first person's client MUST generate a new
> > unique room name if the name they have chosen is already in use and
> > continue this process until the room is created, the process is
> > by the user, or a timeout period elapses.
> Perhaps it would make sense to use the full JID in the room name to
> it
> even more unique.  That way, the only way it can fail is if the same
> attempts to create two rooms within one second.
> I know that it's pretty unlikely for two users with the same node and
> different domains, to create a chat room on a server at the same time,
> perhaps there would be other benefits to using the full JID in there,
> users who get invited into the room could see exactly who started it.
> Or maybe seeing who started it is a bad thing (privacy?)  Perhaps the
> string should be SHA-1'd for paranoia compliance.

Well, we could recommend clients use a UUID (GUID) instead or, as you
mention a SHA1 hash of a combination of the initial participant full

Ideally the client doesn't have to deal with the name at all.  We could
add an iq get/result that creates a unique name at the service so the
client doesn't have to worry about it.  The service could then reserve
that room for a period of time until the requesting resource creates it.
It would most likely be a GUID, or simply an incrementing number.

<iq to="conference.example.com" type="get" id="1">
	<name xmlns="http://jabber.org/protocol/muc"/>

<iq from="conference.example.com" type="result" id="1">
	<name xmlns="http://jabber.org/protocol/muc">asdfghjkl</name>


More information about the Standards mailing list