ietf
[Top] [All Lists]

The LTRU initialization document (was: Re: [Ltru] Re: Last call comments on LTRU registry and initialization documents (issue #878))

2005-09-10 17:15:46


--On Friday, 09 September, 2005 22:38 -0700 Randy Presuhn
<randy_presuhn(_at_)mindspring(_dot_)com> wrote:

...
(2) The document is passed to the RFC Editor for
publication,

It's not clear to me what the point of publishing it as an
RFC is, and in fact if an explicit decision to that effect
has been made, it went past me entirely.

IMHO after initial-registry has served its purpose, it should
be allowed to time out and die.
...

(As ltru co-chair)

Those who follow the ltru WG mailing list closely will recall
that there were two distinct issues recorded in the issue
tracker: (at https://rt.psg.com/ user and password are "ietf")
#875 should initial registry contents be made into an
internet-draft? #878 should the initial registry contents i-d
be published as an RFC?

There was a rough consensus in support of #875, but the
discussion of #878 was inconclusive.   We asked our AD whether
he had any preference with respect to #878, and he indicated a
preference for publication as an (informational) RFC.

That is how we got to where we are today.

The decision on #875 (initial registry as i-d) has in
retrospect proven to be a good one, in my personal opinion.

Regarding issue #878, I think we've all looked at it as just a
matter of how to work the process to get the right stuff into
the registry.  The working group has been quite clear in
wanting to ensure that the initial registry is not mistaken
for the registry itself, and was never enthusiastic about
publishing it as an RFC.

Consequently, I do not think that it would be a big problem if
the IESG were to tell the ltru WG that we don't need to
publish the initial registry i-d as an RFC, or if the WG,
reconsidering issue #878 as a result of the IETF last call
comments were to reach a consensus to withdraw the publication
request, as long as we have some assurance that the registry
will be end up with the right stuff.

Sure, we would have wasted a lot of energy getting the initial
registry i-d RFC-ready, but the important thing is to
initialize the registry.  Process is a tool, not an end in
itself, and the result we are interested in is getting the
updated registry going, not publishing a history of its
initialization, as interesting as that might be.

Extremely violent agreement.

The only problem I see is that the registry draft,
http://www.ietf.org/internet-drafts/draft-ietf-ltru-registry-1
2.txt, currently references the initial registry i-d in
multiple places; we'd need someone to provide the correct
incantation to make the right thing happen.  I can hardly
imagine that a document (whether BCP or some kind of
Standard-to-be) would be permitted to reference (even
informatively) an i-d never intended for publication. Please
send suggested text.

Although my personal preference would be to publish the i-d
http://www.ietf.org/internet-drafts/draft-ietf-ltru-initial-04
.txt as an Informational RFC, ideally marked "Historic" from
the first day it appears, I do not object to non-RFC
alternatives that let us accomplish the initialization, as
long as we keep the initialization data out of any container
that would cause it to be mistaken for the registry itself.

Well, that pair of issues is exactly what I was concerned about
and suggesting fixing, despite Doug's strong reaction to a
suggestion I didn't intend to make.

Specifically: If a document is published in the RFC series that
lists elements of a registry, it will, inevitably, be used as a
reference instead of the registry.  Disclaimers in the document
(as now exist) will help with that a bit.  The word "historic"
in the RFC Index will help with that a bit.  But it will still
be referenced by someone, somewhere, as an "official" RFC and an
easy-to-obtain list.   At the same time, there is useful, albeit
mostly-historical, information in the initialization document
that I would personally prefer not get lost.  And that
referencing issue is a pain in the neck, although not a huge one.

So, very specific suggestion:

        (1) Get the registry created and initialized, exactly as
        specified in the "initial" document at -02.  I think
        that is a no-brainer and we all seem to be agreeing
        about it.
        
        (2) Issue, and send to the RFC Editor for Informational
        publication, a version of draft-ietf-ltru-initial that
        replaces the contents of Section 3 ("Initial Registry
        Contents") with a pointer to the registry itself and, if
        desired, a hint as to how the initial elements can be
        identified, for historical purposes, in the registry.
        Leave the rest of it alone.  It contains useful text and
        some information that should not be lost (such as the
        omitted code element list).  Just to be orderly, remove
        the "don't pretend this is the registry"
        statements/warnings from the abstract and the end of
        Section 1.

This prevents anyone from referencing the RFC form of the
initialization document as the registry far more effectively
than any warnings or labels -- there will be nothing there to
reference.  It saves time patching references to the
initialization document because the initialization document will
exist as an RFC and even have the same section numbering.  It
saves worrying about whether there is important historical
information in the initialization document that would need to be
moved somewhere else if that document were dropped, and where to
move it. 

The WG and IESG should do what they think best, but this really
seems pretty obvious to me.

Of course, none of the above has anything to do, one way or the
other, with my concerns about BCP versus Standards Track,
replacement of 3066 without doing something specific about the
matching rule question, applications-specific profiles, and so
on.  Whatever the right answers to those concerns are, they are
not as obvious, at least to me.

    john


_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf

<Prev in Thread] Current Thread [Next in Thread>