[standards-jig] Essence of Jabber

David Waite mass at akuma.org
Fri Mar 8 19:40:05 UTC 2002


Fabrice DESRE wrote:

>  You make a good point whith embedded devices (even if an XML 1.0 + 
> namespaces fits in less 200k bytes), but this is not what I want to 
> show. There are some design mistakes IMHO in jabber related to 
> namespaces, and the server MUST be conformant. If you really want to 
> impose a subset of XML in jabber, then don't pretend to use 
> namespaces, because you don't currently. Of course if you do so you're 
> doomed when we came to interoperability with other standards (i'll 
> show you how the actual jabber implementation is unable to carry some 
> SOAP messages)
>  Let me some time to write it cleanly, please...

As a side note, it is my understanding that Jabber (and any XML-based 
transport) is unable to carry SOAP messages as XML, by the 1.1 
recommendation. See http://www.w3.org/TR/SOAP/#_Toc478383494, 
particularly the first line - "A SOAP message is an XML document that 
consists of a mandatory SOAP envelope, an optional SOAP header, and a 
mandatory SOAP body."

Without this restriction, it may still be impossible to validate SOAP 
messages within a multiplexed XML transport, because the 'id' attributes 
are actually IDs. This means that they should be unique within the XML 
document; since the document is a stream, requests and responses on the 
incoming stream would have to have IDs which are unique across the 
session. This means they would have to be unique on all incoming traffic.

I have heard some things in SOAP 1.2 remedy this, but haven't found 
them, and never got a response to my inquiry about the line in section 4 
of the 1.1  recommendation.

All we can do is transport something which can be transformed to and 
from a SOAP message with a minimal level of complexity (such as, sending 
the node within some Jabber message, then importing this transport 
message node into a new document.)

-David Waite




More information about the Standards mailing list