So you don't care if a <create> actually creates anything, right?
That's a policy decision. :-P I'm not a fan of the PIR/Afilias rules. They do seem to follow the beat of a different drummer. I think they will argue exactly how I described that there is nothing in the spec that says a <host:chg> operation MUST succeed if they decide the operation should not be allowed BY POLICY. PIR's current implementation already ignores the <poll> requirements from draft 7 of the base specification as I described on the orgissues.net web site: http://www.orgissues.net/orgissues/index.cgi/2003/06/30 I don't expect them to implement <poll> properly under the official 1.0 spec either. It will give me the opportunity to revive the orgissues site... Going back to the original issue, I think it is MOST important that the gTLD Registries be consistent in their EPP implementations. The sTLDs should also be consistent within the confines of their sponsors' business rules. The ccTLDs (other than those who market themselves as virtual gTLDs) are probably outside of our influence. If the Registries don't agree with the existing specs, fine. But they better submit the revisions to the IETF and all MUST implement EPP consistently. That's the message I'd like to see us send to the gTLD Constituency. Just my opinion... _Mike