[standards-jig] JEP-0033 Jabber Packet Headers in iq

Peter Saint-Andre stpeter at jabber.org
Mon May 5 22:23:38 UTC 2003

On Fri, Apr 25, 2003 at 02:44:58PM +0200, Ralph Meijer wrote:
> On Fri, Apr 25, 2003 at 05:25:38AM -0700, Justin Karneges wrote:
> > On Thursday 24 April 2003 11:49 pm, Ralph Meijer wrote:
> > > Iq stanzas can contain only one namespaced (other than iq itself) child.
> > 
> > I don't believe this is true.  Do you have a reference?  From what I 
> > understand, nearly every iq-related protocol happens to only have one child 
> > element of iq, but this is not a requirement.
> I have had this clarified in the latest -core IETF draft, section 7.5.3 [1].
> You can't have multiple childs of IQ because it opens a whole can of worms:
>  - if one of the request inside an iq fails, should actions performed because
>    of the other requests be rolled back?
>  - which request did actually fail and how do I know?
> There are probably a lot more, and I'd rather not go there.

We used to waffle on this point, but have recently made it explicit:
only one child of an IQ (plus <error/> if <iq type='error'>).


Peter Saint-Andre
Jabber Software Foundation

More information about the Standards mailing list