[rfc-i] Titles for divided reference sections in non-standards track documents

Brian E Carpenter brian.e.carpenter at gmail.com
Sat May 31 13:08:34 PDT 2014


Barry,

I don't think you will easily find consensus for this unless
it is written in a way that requires a judgment call in every
case.

I tend to object to separating the references in an Informational
document, unless it's actually a specification (e.g. of a proprietary
or de facto protocol) or something that looks mighty like a BCP.
To a considerable extent, it's the same issue as whether it uses
RFC 2119 terminology - and we have a running sore, IMHO, about
whether we should use RFC 2119 in requirements documents, which
are usually published as Informational.

So, if the text says "...MUST support avian carriers [RFC1149]"
it's entirely reasonable to expect the references to be
separated. But if it says "...might be useful to consider
avian carriers [RFC1149]", not so much.

Experimental RFCs are easier, because it's usually quite obvious
whether they are actually specifications, in which case RFC 2119
and separation of references are appropriate.

To put the same thing another way, if a reviewer is likely to
ask "Why isn't this standards track or BCP?", then both
RFC 2119 and separation of references are probably appropriate.

But I don't think that legislating for this is going to help,
since it surely remains a judgment call anyway.

Regards
   Brian

On 01/06/2014 04:41, Barry Leiba wrote:
> As Elwyn knows, this comes partly from comments of mine on a recent
> document that the IESG evaluated.  The document was Informational and
> had only Informative References, and the shepherd writeup said, under
> the appropriate question, "This is an informational document, and so
> has no normative references."
> 
> I can't tell you how many documents I've reviewed as an AD that were
> just like that, often with the same or a similar note in the shepherd
> writeup.  I try to always comment on it (I don't make them DISCUSS;
> perhaps I should), and to ask the authors to tease out the most
> important references and make them normative -- especially if the
> overall list of references is long.
> 
> Elwyn's trying to address this in a way that will make authors,
> shepherds, and chairs less likely to misunderstand the situation --
> less likely to say that Informational documents don't have normative
> references.
> 
> If no one else thinks that changing the titles of the sections will
> help, perhaps someone can suggest another way.  Training might help;
> perhaps an IESG statement would help.  I really prefer not waiting
> until IESG Evaluation and a comment by an AD for the references to be
> properly split.
> 
> Barry
> 
> 
> On Fri, May 30, 2014 at 7:35 PM, Elwyn Davies <elwynd at folly.org.uk> wrote:
>> On Fri, 2014-05-30 at 18:48 -0400, Joel M. Halpern wrote:
>>> I have learned to interpret "Normative" references in all our
>>> specifications as meaning "needing to be understood to understand (or
>>> implement, or similar usage) this RFC."  That seems to apply equally
>>> well as a meaning for references in Informational or Experimental documents.
>>>
>>> Sorry, not seeing the problem,
>>> Joel
>> With my Gen-art hat on, I would say that the fact that IETF insiders
>> understand why a non-standards track document has a "Normative
>> References" doesn't help people who are picking up documents with little
>> or no prior exposure to IETF culture.
>>
>> A small change to make our documents more user friendly.
>>
>> /Elwyn
>>
>>> On 5/30/14, 6:39 PM, Elwyn Davies wrote:
>>>> A large proportion of informational documents do not describe protocols.  As I said some do and normative would be appropriate there.
>>>>
>>>> /Elwyn
>>>>
>>>>
>>>> Sent from my ASUS Pad
>>>>
>>>> Julian Reschke <julian.reschke at gmx.de> wrote:
>>>>
>>>>> On 2014-05-30 23:00, Elwyn Davies wrote:
>>>>>> Seriously?  Well, it's misleading because it has a different meaning.  Its good to have precision.
>>>>> I disagree that it has a different meaning. Informative and Experimental
>>>>> RFCs still describe protocols, even if we don't call them "IETF
>>>>> standards". These protocols have normative parts, no?
>>>>>
>>>>>> We got normative by default after we stopped requiring undivided references in informational documents.
>>>>>>
>>>>>> /Elwyn
>>>>> Best regards, Julian
>>>>>
>>>> _______________________________________________
>>>> rfc-interest mailing list
>>>> rfc-interest at rfc-editor.org
>>>> https://www.rfc-editor.org/mailman/listinfo/rfc-interest
>>>>
>> _______________________________________________
>> rfc-interest mailing list
>> rfc-interest at rfc-editor.org
>> https://www.rfc-editor.org/mailman/listinfo/rfc-interest
> _______________________________________________
> rfc-interest mailing list
> rfc-interest at rfc-editor.org
> https://www.rfc-editor.org/mailman/listinfo/rfc-interest
> 


More information about the rfc-interest mailing list