On Mon, Aug 29, 2005 at 03:09:52PM -0700, Ralph Giles wrote: > On Mon, Aug 29, 2005 at 08:12:09PM +1000, Trejkaz wrote: > > > The difficulty with that approach, of course, is that it can't possibly > > with > > with existing, standard IQ queries. I don't suppose we need some kind of > > <iq > > type='cancel'/> where you pass the same ID of the query you want to > > cancel... > > This has the same problem that you don't know if the response you get > failed because it failed, or because you cancelled it. I guess we could > create a matching xmpp error element.
Well, failing is easy, there are lots of stanza errors for that. Cancelling could yield a <undefined-condition/> stanza error, along with an application specific condition element (see 9.3.2 of RFC 3920). > No one's commented on whether my idea of sending another 'set' with the > same iq id before the initial response comes back is reasonable. Ok, here is my opinion: don't do that. Jabber assumes the pattern iq-get or iq-set followed by a iq-error or iq-result response, where the 'id' attributes match up. -- Groetjes, ralphm _______________________________________________ jdev mailing list [email protected] http://mail.jabber.org/mailman/listinfo/jdev
