To:
"'Sheer El-Showk'" <sheer@saraf.com>
Cc:
ietf-provreg@cafax.se
From:
"Hollenbeck, Scott" <shollenbeck@verisign.com>
Date:
Mon, 26 Nov 2001 15:03:16 -0500
Sender:
owner-ietf-provreg@cafax.se
Subject:
RE: Provreg WG last call announcement
> -----Original Message----- > From: Sheer El-Showk [mailto:sheer@saraf.com] > Sent: Monday, November 26, 2001 3:34 PM > To: Daniel Manley > Cc: Hollenbeck, Scott; ietf-provreg@cafax.se > Subject: Re: Provreg WG last call announcement > > > Well there is another concern with the message being specified directly > in the protocol itself and that is internationalization. While status > values, xml elements, etc have to be specified in some language to make a > meaningful protocol I think we should avoid specifying error string > values. > > An alternative suggestion, and one that might be more meaningful to > automated registrar systems, would be to include a second, more detailed > response code. This would involve adding a second <minor-code> tag that > would display a numerical subcode to any of the existing EPP codes. Thus > a response object would now include two different codes: > > main Epp Code: 2003 "Required parameter missing" > Epp subcode: 2003-4 "ID required for contact create" There's a problem with the above suggestion: you've just mixed an object-specific subcode into the object-agnostic base protocol. Plus, isn't the idea of trying to specify policy-specific error situations is something we covered (and rejected) quite some time ago? I really don't think it possible to identify a subcode for every possible error situation. Even if we tried to create object-specific subcodes to put the object-specific semantics where they belong, I can't see a way to come up with a completely unambiguous set of codes. All we'd be doing is adding an additional layer of codes, and we'd _still_ have a list that is either over-specified for someone or under-specified for someone else. > I think this would make for much cleaner error handling on the part of the > registrar. > > Of course this could be done in the <value> tags and in either case it > would need to be extensible to reflect different registry rules, but I > think a basic framework for a minor code in the base protocol would go a > long way towards making client-side epp implementations more standardized > (ie provide consistant error handling in the protocol rather than rely on > parsing registry-dependent data out of <value> tags). No, it can't be done in the <value> element. That element exists to identify the data that caused the error situation [1], not to carry customized error responses. If someone is passing registry-dependent data in the <value> element, the element is being misused. -Scott- [1] From section 2.5 of epp-05: "- Zero or more OPTIONAL <value> elements that echo client-provided values that caused server error conditions."