[rfc-i] citing historic internet drafts -- and related issues
ah at tr-sys.de
Fri Oct 24 02:11:06 PDT 2008
I have followed this thread for a long time without raising
my voice because I got used to obstain from contributing to
lengthy controversial debates that regularly tend to get
circular, in order to more efficiency dedicate spare time
to IETF work.
But it seem to get the proper time now that the
'silent majority' starts joining this caravana,
noting obvious observations and pointing out the needs
perceived by a passionate reader of IETF publications.
(1) Archival of I-Ds
As far as I have been told, lawyers and other IPR gurus have
adviced the IETF to properly archive I-Ds in order to be
able to effectively claim the IPR held by the IETF Trust,
in a credible manner.
And so it happens in the meantime, despite of disclaimers
in outdated boiler plates, authoring guides and RFCs.
Thus, the nature of I-D documents has changed due to
circumstances outside the control of the IETF, and not
(yet) properly reflected in the RFCs regarding the matter.
(2) Utility of proper distinction
There's a recent example showing that the RFC Editor has
been caught in it's own trap of not properly distinguishing
"work in progress" citations from "expired work in progress"
It has been mentioned on this list that the reader should
make use of the publication dates in the citation and the
RFC to make that distinction.
But that doesn't work at all if the RFC Editor has missed
to update the "work in progress" citation or does not
indicate that they did so intentionally, in order to refer
to a previous version that has been grossly outdated by
current work in progress.
In the recently published RFC 5254, the Informational Ref.
'[PWE3-OAM]' points to a predecessor of a *current*
"work in progress" dated 'March 2005'.
The I-D repositories contain a document with the same title
and a slightly reduced authors list, dated 'July 2008'
( draft-ietf-pwe3-oam-msg-map-07.txt ).
The RFC reader is left alone with the interpretation of this
citation. Was it really the intention to quote an old version?
Should it be corrected by an Errata Note? I don't know.
The information is insufficient. I'm left confused.
This single example gives an irrefutable proof of the need for
an unambiguous practice.
(3) Need for "work in progress" citations to documents that
did not become eligible for publication on any RFC stream
Currently, the IAB is developing a series of documents
at the "meta level" in order to provide better guidance
on protocol development, extensibility, etc.
In these documents, a welth of examples is quoted,
even "bad" ones, for educational reasons. The IESG would
certainly have (or indeed has) objected to the publication
of such documents after the significant issues or defects
therein had been detected.
Publishing these evolving IAB documents without clearly
qualifying the "bad" documents as expired would be rather
Thus, the statement found in this thread that I-Ds should
better have been published as Independent Submission RFCs to
make them eligible for citation in later work, is misleading.
There's a clear need to distinguish in the References between
work actually "in progress" and obsolete/abandoned/stopped
work in progress -- commonly denoted as "expired", if no
kind of (subjective?) qualification is intended.
(4) Observations on citation style in I-Ds
The page one boilerplate in I-Ds pretends to enforce the
puristic style of *only* using "work in progress" and not
giving proper bibliographic details for I-D citiations,
which authors in fact ought to be due diligence of because
of IPR concerns.
This rule is obsolete and no more followed in practice since
many years, due to the recognized need for precise citations,
and the xml2rfc tool making it easy to include the draft name
in I-D citations.
Loosing this information at RFC publication time is perceived
as questionable practice neglecting the service needs of the
user, the reader of the RFC.
(5) Personal observations
The last point immediately leads to what I see as the most
We should stop asking only to follow legacy rules; we should
pay more attention the service point of view: How can we best
serve the needs of our 'customers' the readers of RFCs ?
Which meta information has most utility for the reader --
today, in two years, in ten year, and perhaps in fifty years?
That way of thinking should positively influence our decisions.
I am used to follow up "work in progress" citations in RFCs
(and record that on the paper copies, since the files 'never
This always incurs the troublesome process of (re-)discovering
whether or not the quoted document has evolved into an RFC,
and if not, what the document name might have been.
This task would have been simplified very much if RFC
authors would have been able to make the distinction
*at the time of publication*, by including a proper keyword
signaling the difference.
There's always the possibility that work in progress at the
time of RFC publication will be abandoned later, but current
practice shows that this is a rare event.
IMO, the perspective at the time of publication of an RFC
does not help the reader studying an RFC many years later
if it is not clearly made visible.
She should be informed precisely whether a "work in progress"
quoted in an RFC was intended as a historical reference
or as a snapshot in time of what is perceived as something
that is being expected to make its way into an RFC.
Keeping the 'sacred' rule of quoting I-Ds as "work in progress"
IMO does not preclude adding more information, in particular
clearly stating "*expired* work in progress" if that is the
intended judgement of the author(s).
I share the concerns regarding URLs and their stability.
But it would be *really* a big deal for readers not in close
contact with the 'kitchen' originating the document (I apologize
for the verbiage, but it relieves me from distinguishing streams,
WG vs. non-WG, etc.) to be supplied with the *name* of the draft.
Application of common sense and available tools would thus allow
the slightky experienced reader to locate a copy of the document
on his/her favorite I-D repository, and in the media type he/she
Thus, I would really appreciate amendments to the mandatory
"work in progress" (which I don't question here!), in particular
"expired work in progress" -- iff appropriate --
and "document name: draft-xx-yy-xx"
The most useful rule I can envision would be to include the draft
version number if and only if it is intended to make reference of
a particular draft version, and to omit it otherwise, if further
progress of the draft is expected by the RFC authors.
Additionally, over some period of time, early RFCs regularly used
to contain annotated citations, helping the reader to refer to
just those References needed for his/her particular level and
kind of interest. I really appreciated this style, and I really
miss it generally in contemporary RFCs.
This is an appeal to the RFC Editor to admit and encourage the
use of annotations in the References section(s) of RFCs, and
to authors to make use of that opportunity in a prudent way.
| TR-Sys Alfred Hoenes | Alfred Hoenes Dipl.-Math., Dipl.-Phys. |
| Gerlinger Strasse 12 | Phone: (+49)7156/9635-0, Fax: -18 |
| D-71254 Ditzingen | E-Mail: ah at TR-Sys.de |
More information about the rfc-interest