Re: Proposed Resolution to PROT I-D Issues List

["Tom Petch" <nwnetworks@dial.pipex.com>]

This issue surfaced just recently on another list and I am racking my brains to
remember where (I thought IPv6 but cannot see it there).  As I recall, it was a
discrepancy between ABNF and text with substantial disagreement as to which had
precedence; I think the ABNF won.

Tom Petch

----- Original Message -----
From: "Joel M. Halpern" <joel@stevecrocker.com>
To: "McDonald, Ira" <imcdonald@sharplabs.com>; "'Wes Hardaker'"
<wjhns1@hardakers.net>; <sberl@cisco.com>
Cc: "'Andy Bierman'" <abierman@cisco.com>; <netconf@ops.ietf.org>
Sent: Tuesday, March 22, 2005 6:18 PM
Subject: RE: Proposed Resolution to PROT I-D Issues List


> The other point is to echo Andy's comment.
> Except in how we use captialized MUST, MAY, SHOULD wording, the IETF does
> not make a strong distinction in our specifications between normative and
> informative text.  (This is largely because we are not in the conformance
> testing business.)
>
> As such, there is likely no reason to make an issue in the RFC of the
> question of what meaning applies in the event of conflict between text and
> the XSD.  There is a very good reason to make sure to the best of our
> ability that there is no such contradiction.
>
> Yours,
> Joel M. Halpern
>
> At 12:11 PM 3/22/2005, Joel M. Halpern wrote:
> >The simplest answer is to look at other cases.
> >The IEEE 802.1d bridging spec was quite clear that the English was
> >normative.  Nonetheless it include code for all the procedures, and that
> >code was quite useful to people.
> >
> >I think that it is more helpful to look at the question differently.
> >One expects even informative language in a specification to be
> >helpful.  There is often a lot of informative language to help the reader
> >understand the intent of the specification, and to help get effective
> >results in the real world.
> >
> >The only question under discussion is, if there is a disagreement between
> >the English and the formal specification, which one should the reader
> >assume was intended.   Such a disagreement is a bug in the spec, even if
> >the part that is wrong is not normative.
> >
> >In the absence of a disagreement, having the XSD will enable implementors
> >to more reliably implement the intent of the specification.  This is helpful.
> >
> >It can be argued that we should elevate it further.
> >We could go down the path of insisting that all normative behavior must be
> >described either in the formal portions of the XSD or in comments
> >explicitly part of the XSD.  That would make reading the specification and
> >understanding it hard, but it would mean that we would be clear about what
> >was intended to be binding.
> >We could say that the English is binding for semantics, but the XSD is
> >binding for those things which it specifies.  This would make things very
> >confusing.
> >There is admittedly also confusion in having the XSD but not making it
> >normative.
> >
> >Feel free to pick your confusion.
> >But please do not assert that having a differing viewpoint is nonsensical.
> >
> >Yours,
> >Joel M. Halpern
> >
> >At 08:27 AM 3/22/2005, McDonald, Ira wrote:
> >>Joel, Randy, Wes, et al - could you please explain
> >>to this list how XSD is useful in NetConf if it's
> >>not Normative?
> >
> >
> >--
> >to unsubscribe send a message to netconf-request@ops.ietf.org with
> >the word 'unsubscribe' in a single line as the message text body.
> >archive: <http://ops.ietf.org/lists/netconf/>
>
>
> --
> to unsubscribe send a message to netconf-request@ops.ietf.org with
> the word 'unsubscribe' in a single line as the message text body.
> archive: <http://ops.ietf.org/lists/netconf/>


--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>