[Jabber-IETF] Summary

Peter Saint-Andre stpeter at jabber.org
Mon Oct 14 17:02:57 CDT 2002

On Sun, 13 Oct 2002, Pete Chown wrote:

> 1.  It would be useful to have clarification of the permitted
>      character encodings.

Agreed. Your text looks good.
> "Software implementing XML streams MUST support the UTF-8 and UTF-16
> encodings for received data.  Software MUST NOT attempt to use any
> other encoding for transmitted data.  The encodings of the transmit
> and receive streams are independent.  Software may select either UTF-8
> or UTF-16 for the transmitted stream, and should deduce the encoding
> of the received stream as described in [8]."
> 2.  The purpose of dialback authentication should be clarified.

Agreed, the current explanation is perhaps overly terse. The purpose of
dialback is not primarily to assist organizations who would like to run
separate inbound and outbound XMPP servers, but to (as you say) "make
message forgery more difficult" or, more precisely, to make server
spoofing more difficult (which results in making message forgery more
difficult). Perhaps a comparison wiht SMTP would be in order here too.
> 3.  The status of the opening <?xml ... ?> tag should be clarified.
> "Note that the prolog to an XML document -- <?xml ... ?> -- is not a
> processing instruction.  Applications MAY send a prolog.  Applications
> MUST follow the rules in [8] concerning the circumstances in which a
> prolog is included in the document."

> I've just realised something else.  If the namespace issues are
> resolved, the second paragraph of XMPP Core section 7.2 should be
> deleted.  We probably need, instead, something that tells you what to
> do in order to interoperate with current software.

I agree that a section of Implementation Notes would be helpful.

Here are a few more:

5. Make it explicit that a node must not provide a 'from' address on XML
chunks to be routed via the host. If a node does provide a 'from' address,
the host must discard the 'from' address and must add a 'from' address
equal to the user at host/resource from which the chunk originated.

6. Do we want to address XMPP URIs at all (a la JEP-0032)? For example, we
might want to be able to address a message to a node using the following
kind of syntax:

      xmpp:node at host?message&body=Hello%20World

I don't see much else in xmpp-core that needs serious attention, other
than xml:lang attributes on streams and core data elements. I'm leaving
out the usual small cleanup items (such as further clarification of the
error codes and whatnot).

Also I need to read through xmpp-im again -- it's been four months since
I've worked on these docs! One general thought regarding xmpp-im is how
much detail to provide -- right now what is included there is all and only
the protocol necessary to address RFCs 2778/2779, not the entire scope of
the Jabber protocol. I think it's probably best to keep it that way, and
to note that there is more to the protocol than what is defined in


Peter Saint-Andre
Jabber Software Foundation

More information about the xmppwg mailing list