[standards-jig] Version 0.5 of JEP-0045

Peter Saint-Andre stpeter at jabber.org
Mon Sep 23 23:19:42 UTC 2002

Is there a specific proposal to change the protocol here?

I could see the following:

Clients are not allowed to send messages of type="groupchat". If they do
send such a message, the type is stripped off by router or the network
edge or whatever. The message is then routed to room at service. Before
reflecting the message out to the room participants, the groupchat service
adds type="groupchat" back onto the message. Now the router sees that this
message comes from a chat service and accepts the message type by routing
it on unchanged.

However, is this really necessary?

What are the potentially dangerous or confusing scenarios y'all have in
mind here? I could open an XML debug window or use telnet to send you a
message of type="groupchat" and your client might treat that as a reason
to launch a groupchat interface rather than a regular old chat interface.
Anything else?


Peter Saint-Andre
Jabber Software Foundation

On Mon, 23 Sep 2002, Iain Shigeoka wrote:

> On 9/21/02 10:22 PM, "David Sutton" <jabber at dsutton.legend.uk.com> wrote:
> > On Sat, Sep 21, 2002 at 06:37:57PM -0500, Peter Saint-Andre wrote:
> >> On Sat, 21 Sep 2002, Mike Mintz wrote:
> >> 
> >>> Sorry if these comments have already been mentioned.
> >>> 
> >>> 1. Is there any reason for a user to define the type="groupchat" or
> >>> type="chat" in outgoing messages?  The component just changes them to the
> >>> appropriate type based in if there's a resource or not, so why should it be
> >>> sent?
> >> 
> >> Postel's law: "Be liberal in what you accept, and conservative in what you
> >> send." It's not nice to force the component to change the message type.
> >>  
> > Personal feelings:
> > 
> > Its good for the sake of clarity. Although, I might be in favour of
> > the server checking the type. I can see how misdefining the type could
> > be used maliciously in one case, or confusing in the other
> I tend to agree. Mtr speaks eloquently on the problems introduced by
> following Postel's law to excess (see www.beepcore.org and the discussions
> on BEEP's design). At our current stage of development, I would prefer
> servers to be strict (conservative in both what you accept and send) in
> order to ensure protocols aren't "abused".
> -iain
> _______________________________________________
> Standards-JIG mailing list
> Standards-JIG at jabber.org
> http://mailman.jabber.org/listinfo/standards-jig

More information about the Standards mailing list