Re: Last Call: <draft-jabley-dnsext-eui48-eui64-rrtypes-03.txt> (Resource R ecords for EUI-48 and EUI-64 Addresses in the DNS) to Proposed Standard

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




--On Tuesday, May 28, 2013 15:42 +0900 Randy Bush
<randy@xxxxxxx> wrote:

>... 
>> While the RFC should not be materially misleading, I don't
>> think there is a requirement for Informational RFCs to
>> guarantee any particular level or security or privacy.
> 
> that the draft now tries to slide by as info does not change
> that it specified protocol elements and how they are to be
> used.  and the draft makes very clear that this is
> juristiction specific and a serious privacy problem.

Hmm.   I'm not happy with several aspects of the content of the
draft, including the privacy issue.  I'm also not happy with it
as an apparent example of "if one has a piece of information
that needs to be accessible sometimes, put it in the DNS whether
the characteristics of the DNS are a good match for the
information and retrieval requirements or not".   For those
reasons, and because of the principles expressed in RFC2804, I
believe it was completely unacceptable as a Standards Track
document.

Perhaps these RRTYPEs should never have been created and
registered.  If so, the balance between community review and
maximizing the number of things that are registered (and the
resulting avoidance of conflicts) is wrong for RRTYPEs.  That
would imply a need to revisit the registration policy, but it
wouldn't change the status of these two RRTYPEs.

But it seems to me that none of that is at issue at this stage.  

What is at issue, IMO, is whether the Internet is better off
having a couple of RRTYPEs around with no documentation or
having them documented.  Put differently, are these RRTYPEs
sufficiently reprehensible that we would hope to make them
non-interoperable in practice by not telling people how to make
them work.  I believe that would be a bad plan and would violate
principles far more basic than the 2804 ones -- the IETF should
not be setting itself up as the arbiter of what can be deployed
on the Internet if only because we would certainly fail.

If people find these RRTYPEs (or the document) reprehensible,
then let me suggest that they generate an I-D in Applicability
Statement form that identifies them as "Not Recommended" and
generally obscene and to do so and show sufficient consensus
quickly enough to convince the IESG to force a normative
reference to it into this document so they can be published
together.

I also don't see "slide by as info" in this.  Like you, I
opposed making this standards track and saw serious issues with
2804 in that context.  Tuning aside, only two changes have been
made to the document between -03 and -04 and both are intended
to make it clear that the _description_ of these RRTYPEs does
not constitute a recommendation to use them and that, if one has
serious security concerns that outweigh other considerations,
one should not use them (or put EUI-* information into the DNS
in any other way).   That makes the document about as close to
being information about how these RRTYPEs work without
recommending their use as I can figure out how to get it (others
can probably do better, but apparently haven't stepped forward
with text).  It also contains a pretty strong caution about
their use.  I think that is appropriate too.  It stops short of
saying "not recommended" because the latter would imply
standards track (and Joe might not agree).

>> RFC 2804 is about
> 
> i am very well aware what 2804 contains
> 
>> RFC 2804 doesn't seem to me to be particularly applicable.
> 
> i disagree.  i believe the first two bullets in section one
> are very applicable to joe's draft.
 
>    - The IETF, an international standards body, believes
> itself to be      the wrong forum for designing protocol or
>...

In authorizing the publication of an Informational document that
describes how something works for the information of the
community, the IETF is not acting as an "international standards
body" even though it is being consistent with its goal of making
the Internet better by providing documentation that facilitates
interoperability.  The idea of providing registrations for
non-standardized objects has the same goal.  If we were willing
to register only those objects that met our standards-track
criteria, then we would encourage both code point squatting
(seen in many other cases) or kludges that overload existing
code points (as we have seen with TXT RRs) and damaging
interoperability in both cases.  I don't see the issues in
simply documenting things as different and we certainly have a
long tradition of encouraging the RFC Editor to publish that
type of documentation.

best,
   john





[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Fedora Users]