Applicability Statement: DNS Security (DNSSEC) DNSKEY Algorithm Implementation Status
RFC 6944

Note: This ballot was opened for revision 03 and is now closed.

(Ralph Droms) Yes

(Ron Bonica) No Objection

(Stewart Bryant) No Objection

(Gonzalo Camarillo) No Objection

(Benoît Claise) No Objection

(Wesley Eddy) No Objection

(Stephen Farrell) No Objection

Comment (2012-07-14 for -03)
No email
send info
I can't parse this: "It is believed that RSA/SHA-256 or
RSA/SHA-512 algorithms will replace older algorithms (e.g.
RSA/SHA-1) that have a perceived weakness or these
recommended algorithms are seen in major deployments." I
guess something's wrong but hard to know what, maybe
s/or/or as/

(Brian Haberman) No Objection

Comment (2012-07-13 for -03)
No email
send info
I agree with Barry that a discussion is needed as to why an IANA registry cannot capture the necessary information.

(Russ Housley) No Objection

Barry Leiba (was Discuss) No Objection

Comment (2012-07-13 for -03)
No email
send info
Former DISCUSS, leaving it here for the record.

In principle, I like the idea of preventing fragmentation of the documentation by keeping the Section 2.2 table in one place, and by always replacing the document that has it with another, in its entirety.

Only... isn't that what the IANA registry is supposed to be for?  Why isn't that table just added to the IANA registry, and then maintained there?  Doesn't this document now require us to OBSOLETE this document and replace it every time we want to change the registry?

Update 1
I had no idea of the history of this, until Andrew pointed me at this:

I need to talk this over with Pete, Robert, and Russ, at least, on the telechat, and understand why we think we can't use the IANA registry the way registries are meant to be used.

Update 2
After more discussion with the chairs, here's what I understand...

1. The conformance table needs to be normative, stable, and with maintained history.  It needs to be clear, when you say you conform as of [date X], what it is that you claimed to conform to.  You can conform to an RFC, because once it's published, you have a stable, immutable reference, and you can give the RFC number.  But if I asked you whether something conformed to the RRTYPEs registry, you'd have to ask "at what date?"  And IANA doesn't maintain a history.

2. The conformance table is necessary in the first place because as things stand, you can say you "implement DNSSEC" when (for instance) you don't do NSEC3.  That's not going to be very useful, given that .com, .net, and .org are all using NSEC3.  We need a single document to which people can point and say, "We do it the way that says to."

3. If it takes an RFC to update the registry, then any change to the table needs an RFC anyway.  Thus, I suppose there's no *harm* in just publishing the table in the RFC and not in the registry.  I do like that the basic table is still in the registry, and that this just extracts the conformance criteria into an RFC-maintained table.

There's no point now, but it would have helped if, especially given the history of the earlier document, some of this information had been in the shepherd writeup.  But now that I understand the issues, I'm OK with this method of handling it.

(Pete Resnick) (was Discuss) No Objection

Comment (2013-03-11)
No email
send info
Thanks for addressing all of my comments.

(Robert Sparks) No Objection

Comment (2012-07-18 for -03)
No email
send info
I support Pete's suggestion to make this standards track.

(Martin Stiemerling) No Objection

Comment (2012-07-16 for -03)
No email
send info
A nit:
Section 2.1., paragraph 2:

>    Likewise, ECDSA with the two identified curves (ECDSAP256SHA256 and
>    ECDSAP384SHA384) are algorithms that may see widespread use due to
>    the precieved similar level of security offered with smaller key size
>    compared to the key sizes of algorithms such as RSA.


(Sean Turner) (was Discuss) No Objection

(Adrian Farrel) Abstain

Comment (2012-07-18 for -03)
No email
send info
I am Abstaining on this document. I have no specific objection to its
publication, but it  is so much not the document I would have written,
and I struggle to see its value as currently presented.

I was left wondering whether it is trying to do the wrong thing. Instead
of tying to mandate what must or must not be implemented it would be
more helpful to me to state the usage profile of each algorithm. This
could then be traced to "dangerous to deploy" or "must be implemented
if your implementation is to play a part in the deployment of this

Anyway, here are some thoughts and comments that arose...


Citing conformance to this document is confusing because nearly all of
the information in the table in Section 2.2 indicates a degree of 
choice. Conformance can only be interpretted with respect to the "must"
and "must not".


To say that the "IANA registry for these algorithms ... lacks the
implementation status of each algorithm"implies that this information
should be recorded in the registry. It should not since that is not
the purpose of the registry.                                                                               

How about...                                             

   There is currently an IANA registry for these algorithms, but there
   is no record of the recommended implementation status of each 


I think Section 1 should observe that it is the intention to issue
replacements to this document when the implementation status of any 
algorithm changes and when new algorithms are introduced.


   This implementation status indication is only to be considered for
   implementation, not deployment or operations.

I can't extract meaning from this! If something is marked "MUST NOT"
implement, how can it possibly be deployed?


Since this document makes no requests for IANA allocations and makes no
changes to the existing registries, I see no reason for it to be listed
in the registry. The IANA registry tracks code points, it is not a 
repository for documentation references.


I can't square the statement in the Security Considerations section with
the MUST NOT implement status assigned to RSAMD5. Maybe a reference is
needed for the classification of RSAMD5 and that would keep Section 4