[Standards-JIG] Re: from attribute on initial stream header

Peter Saint-Andre stpeter at jabber.org
Mon Oct 25 23:16:50 UTC 2004


In article <20040927200349.GA17981 at hermes.muc.charente.de>,
 Matthias Wimmer <m at tthias.net> wrote:

> Hi!
> 
> On http://www.jabber.org/wiki/index.php/xmppimp there is the following
> suggestion:
> "The originating server really really should include a 'to' address on
> the initial stream header that it sends to another server (this makes
> life happier for all concerned, especially in a world with lots of
> virtual hosts and subdomains). In addition, it is probably a good idea
> for the originating server to include the 'from' address as well on the
> initial stream header."
> 
> ... while XMPP core says:
> 
> "from -- The 'from' attribute SHOULD be used only in the XML stream
> header from the receiving entity to the initiating entity, and MUST be
> set to a hostname serviced by the receiving entity that is granting
> access to the initiating entity. There SHOULD NOT be a 'from' attribute
> on the XML stream header sent from the initiating entity to the
> receiving entity; however, if a 'from' attribute is included, it SHOULD
> be silently ignored by the receiving entity."
> 
> What should really be done?
> 
> Sending the "from" attribute might be good in case, that the receiving
> server knows that some other server has problems with one of the
> features the receiving server implements and it might not advertize
> support for this feature then. But I feed bad implementing something for
> that an internet standards says that I SHOULD NOT implement it.

This is something that I'm sure we will discuss when we work on 
rfc3920bis. :-) Really, the 'from' address is just an assertion, but can 
be useful when a deployment is serving multiple virtual hosts. In any 
case, the other text in RFC 3920 specifies that the receiving entity 
must verify the identity of the initiating entity, so it is quite 
possible that rfc3920bis will not be so strict about the 'to' and 'from' 
attributes in the stream headers. I will add this to my notes for 
rfc3920bis.

/psa




More information about the Standards mailing list