IETF mission (RE: pausable explanation for the
Document Series)
John C Klensin
john-ietf at jck.com
Thu Jun 5 09:01:11 CEST 2003
--On Thursday, 05 June, 2003 12:33 +0200 Harald Tveit Alvestrand
<harald at alvestrand.no> wrote:
>> Depends upon how you look at things. I would say that the
>> more-core problem is that our quality control may be less
>> than ideal. As the IETF is not a protocol enforcement
>> agency, what the industry does with what we make is beyond
>> our control, in my opinion.
>
> actually this comes back to the IETF mission statement
> thing....
>
> if the mission of the IETF is to "make the Internet work",
> with our particular task in pursuit of that mission being to
> "make high quality, timely standards for the Internet", then
> flaws in the standards that the industry runs on are signs
> that we haven't achieved our mission.
>...
> RFC 2026 invented the term "applicability statements" - that's
> a term that seems to have fallen by the wayside......
Actually, unless my memory has gone _very_ bad, we have had
applicability statements as tools in the collection for a long
time, certainly before 2026. I think they were even called
that. We haven't used them very much because it has been
extremely difficult to get energy and focus for those efforts --
they are hard to do, and rarely as rewarding as developing a
technology standard. Note that RFC 1122 and 1123 are definitely
AS documents, as is RFC 1812. If you go through the RFC list
looking for the word "requirements", you will find it in several
other places as the term is used by an AS in the 2026 sense.
The term "applicability statement" appears in RFC 2000 (and
probably earlier, I haven't had time to go back through that
thread).
<tirade>
The fact that these are hard to get done doesn't imply that we
should have written them off, which we essentially have done...
except by accident, when we call them something else.
We have created a lot of confusion by also using the term
"requirements" to describe "things we really think this protocol
ought to have in it". We also devalued ASs by claiming, of
late, that attempts at ASs must be published only as
Informational, and then telling anyone who will listen that
Informational documents don't count, despite the fact that 2026
explicitly identifies them with standards track maturity levels.
That doctrine has been imposed, and dropped, more or less at the
convenience (I'm tempted to say "whim") of the IESG. For
example, using the definitions of 2026, RFC 3454 meets the
criteria for an AS: it is a profile of an external standard that
specifies how to use parts of that standard in the Internet and
in other IETF protocols. And it is definitely standards-track,
but the term AS was never, as far as I know, used during its
development or approval process.
Speaking of tools we have stopped using, section 3.3 of 2026
describes requirement levels (formerly "status", see below) for
documents -- required/ recommended/ elective/ limited use/ not
recommended. Those statements, or ones similar to them, should
be (or should have been) much more important tools for vendors
and others in determining what documents to use than the
orthogonal standards-track maturity levels. Before we dropped
them --about the same time, if I recall, that we dropped the
notion of conformity clauses in standards-track documents --
they appeared regularly and were attached to _every_
protocol-specifying RFC. See 1083 for examples.
</tirade>
Incidentally, reading RFC 1083, especially section 2, can be
very illuminating with regard to where some of 2026 ultimately
came from. Interestingly, it uses "proposed protocol", and not
"proposed standard". The term "standard" is introduced only at
"draft", as in "draft protocol standard". And it is very clear
about what "draft" means in the context of the discussions
recently on this list: "[This] draft standard state is
essentially a warning to the community that unless an objection
is raised or a flaw is found this protocol will become a
'standard'". I wonder whether we might have reduced some
confusion had we preserved that particular distinction --not
using the term "standard" at all to refer to proposed-level
documents --in wording.
regards,
john
More information about the Problem-statement
mailing list