[rfc-i] draft-iab-styleguide-02 on referencing STDs

John C Klensin john-ietf at jck.com
Sat Apr 12 13:17:57 PDT 2014



--On Saturday, April 12, 2014 08:28 +1200 Brian E Carpenter
<brian.e.carpenter at gmail.com> wrote:

> This is indeed all very problematic.

Yes.

One additional observation, based on trying to fix the
underlying (citations supplied by Brian and a few more) and
having written a number of RFCs that fall into the relevant grey
areas:

(1) There are circumstances in which a reference to an STD as
the "current and most up to date version of things that make up
that potentially-compound thing" makes sense.   I think such
references can are only be informative because of the "STDs
change" discussion in the earlier notes in this thread.  And, to
avoid silly states, STD numbers pretty much have to be assigned
at PS -- either always or as soon as a new PS updates something
that is already part of a STD (hmm... the latter proposal hasn't
been made yet although I think it would be the wrong solution).

(2) What I have proposed several times now is that the style
guide and xml2rfc allow nested references, allowing 

 [STDxx]   ...
      [RFCnnnn] bloggs, J., ....
      [RFCmmmm] Smith, X., ....

(with the RFCxxxx anchors being optional).  That would permit a
reference to the STD when needed but enumerates the particular
RFCs involved, eliminating the volatility problem.  Julian's
"consisting of" form is consistent with that general idea.
Something along those lines is, IMO, the only form that ought to
be allowed in a normative reference, with the alternative being
no normative STD references at all.  The approach does have a
significant problem however, which is that STDs (and BCPs) don't
have either titles or authors (Julian caught that too).  The
last time we tried to fix that (but not the first) and some
related pieces of the problem was
https://datatracker.ietf.org/doc/draft-ietf-newtrk-repurposing-isd/

I'm actually not sure whether this is an IETF matter or an RFC
Editor one.  I am sure that the RFC Editor can't fix it without
a change in procedures that would permit assigning STD numbers
to proposed standards, at least those that update parts of full
standards.   So it is either an IETF matter or it is blocked
until the IETF takes action.

The problem is that, IIR, there are published, standards-track,
RFCs that reference STDs.  Certainly there are published RFCs
that reference multi-document BCPs that raise almost the same
issues (other than the "what to do about Proposed Standards"
one).   So, unless the RFC Editor simply wants to ban such
references going forward, the Style Manual should probably say
something.  IMO, it should be very, very, conservative,
especially wrt normative references (even normative references
to a BCP or STD that currently consists of only a single
document).

best,
   john

p.s. the main reason I've largely withdrawn from these
discussions and am unlikely to see anything on which I'm not
explicitly copied is that I have decided to treat "make
suggestions, get no response" as equivalent to "no one is
interested in your ideas, perspective, or experience; please go
away".  That personal policy considerably improves the S/N
ratio, reduces the workload, and reduces the aggravation load
even more.




More information about the rfc-interest mailing list