Re: [dispatch] [rfc-i] Advice when converting W3C ED to I-D

[tom petch <[e-mail]>]

From: rfc-interest <[e-mail]> on behalf of Rahul Gupta <cxres=[e-mail]>

Thank you Eric, John, Tom, Carsten for your advice:

Some specific comments:

On 14. Sep 2023, at 13:34, tom petch <[e-mail]> wrote:

Now my browser and operating system have been upgraded and I no longer can.

Is there anything I can do to make the page available? You can also find the markdown source (still fairly readable at) <https://github.com/CxRes/per-resource-events/>. I-D version is still pending, unfortunately.


Re Non-normative sections: To counterpose Carsten's point - RFC7322 Section 4.8 states:

Each RFC must include an Introduction section that (among other
   things) explains the motivation for the RFC and (if appropriate)
   describes the applicability of the document, e.g., whether it
   specifies a protocol, provides a discussion of some problem, is
   simply of interest to the Internet community, or provides a status
   report on some activity.

In my epistemic understanding, it is unclear how this can ever be normative. To use simpler language, only "What" a user of the spec is supposed to do should be normative. “How” and “Why” the specification came about, i.e. the introduction, cannot be normative.

<tp>
Well it depends what you mean by Normative.  AFAICT the only IETF definition applies to I-D references which must be divided into two and the Normative ones are those needed to understand the rest of the I-D.  This gets gamed with authors trying to sneak through as Informative references that are clearly Normative but that is another issue.

For the I-Ds I am familiar with, Julian's definition does not work.  There are legion sections in I-Ds which define structures, values, names, messages and so on with not a MUST or such like in sight but an implementation that ignores such data will not interoperate so while the presence of a MUST indicates Normative language, the absence thereof conveys no meaning.

Going back to your question, how to indicate what is Normative, would say that that is not what the IETF does, apart from References.  As I said, a common convention is to put material such as examples, implementation reports, ideas for the future, in Appendices which will convey the message (to many) that this is Informative (whatever that word means!)

The idea of dividing the I-D into what matters and what does not comes from RFC8174/RFC2119.  To quote RFC2119

  Imperatives of the type defined in this memo must be used with care
   and sparingly.  In particular, they MUST only be used where it is
   actually required for interoperation or to limit behavior which has
   potential for causing harm (e.g., limiting retransmisssions)  For
   example, they must not be used to try to impose a particular method
   on implementors where the method is not required for
   interoperability.

so you could see Normative as 'REQUIRED for interoperability' but that does not mean you have to label the Introduction as 'NOT REQUIRED for interoperability'.  The IETF is not that formal, to me, other SDO may vary.

I do not know the history of RFC2119 - it was before my time so I have always taken it as gospel.

Tom Petch  

If the entirety of numbered sections are to be considered normative, this requirement is contradictory!


Re License: Thank you, Eric, John on your advice. I believe that should work!

While not a priority, it is something I shall bring up with IETF Trust at a later date.

BR/Rahul

------- Original Message -------
On Thursday, September 14th, 2023 at 11:25 PM, Carsten Bormann <[e-mail]> wrote:



Sent from mobile, sorry for terse

On 14. Sep 2023, at 13:34, tom petch <[e-mail]> wrote:

Appendices are Informative unless there is a statement to the contrary

Julian is right. Everything that uses normative language is normative, unless specified otherwise. Position or heading style has no bearing (of course you can declare it otherwise in each document).

The habit of marking crucial parts of a document as non-normative had irritated me before. Try to use any of these documents without using the “non-normative” parts. Bizarre.