First, I think we have to take the question of whether and when to use 2119
key words into the context of the document we're writing. BCPs and
Informational documents can and do use them, but they use them in somewhat
different ways than Standards Track documents do, while still maintaining
the spirit of that text you quote from RFC 2119.
Second, I bristle when I see instructions to IANA in the form of "IANA MUST
[do this]." We should not use 2119 key words in registration information
in that way.
Third, as instructions to registrants, I think it's absolutely fine to say
"you MUST [include this information," and "you SHOULD [include that
information]," and so on. I think that respects the spirit of 2119, in
that things are more likely to interoperate if the documentation of
registered values is up to snuff.
Barry
On Tuesday, October 29, 2013, Julian Reschke wrote:
On 2013-10-29 21:29, Bradner, Scott wrote:
seems to me to be completely reasonable to say MUST include the number of
the RFC that describes
the protocol being registered (for example)
Scott
But then:
6. Guidance in the use of these Imperatives
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.
To me this indicates that we should keep them out of registrations
procedures.
(I also note that the "MUST" in the text I quoted shouldn't been used if
the text followed its own advice :-).
Best regards, Julian