On Sat, Apr 11, 2020 at 01:33:29PM -0400, Joel M. Halpern wrote:
> Looking at the draft, at a sample IANA registry (I picked one I knew), and
> Nico's comment, I think I see a potential for confusion in the current text.
Perhaps I mistook "IETF documents" for "RFCs". A simple fix to that is
to clarify "IETF documents".
OLD:
Abstract
IETF documents have been inconsistent in what they specify as the
registrant (or contact, or change controller) in IANA registrations
they make. This document provides a consistent specification
("IETF") to be used, and allows for exceptions with IESG approval.
NEW:
Abstract
IETF documents have been inconsistent in what they specify as the
registrant (or contact, or change controller) in IANA registrations
they make. This document provides a consistent specification
("IETF") to be used as the registrant when the registration arises
from publication of a Request For Comment (RFC) that has received
IETF Review. Exceptions are allowed with IESG approval.
When/why would the IESG approve exceptions, and who might ask for them?
Is an exception mechanism just a safety switch in case of unforeseen
circumstances?
> The text is clear that it is talking about IETF drafts. That part is fine
> in my view. But it is unclear within that what registries the "change
> control" marker is to be applied to. On the one hand, there is no point in
> a Change Control designation for an IETF review registration, as the
> definition for change control is the IETF. On the other hand, for expert
> review, first-come / first -served, or experimental use, change control does
> not reside with the IETF or the working group.
Note that you can do both, publish an I-D targeting publication on the
Standards Tack, and within it direct IANA to make registrations in
Expert Review of FCFS registries. Because the authors could have
requested registration independently (and possibly did), what is the
best registrant in that case? The registration might have been accepted
and effected before publication -- should it be updated upon
publication?
Note too that by the time an RFC is published, the original registrants
of any related FCFS registrations might no longer be authors. Losing
this information from the IANA registration record would be bad. One
way to prevent this would be to _add_ "IETF" as a registrant rather than
have it be the only registrant.
Perhaps the IESG exception is about this?
> For some registries, the "IETF review" really does mean the working group.
> (Routing Protocol code points go through the relevant working group.) For
> others, there can be and often is overlapping responsibility, which is why
> the policy is "IETF review" not "Working Group Review".
Another inconsistency. But again, the registration record should record
all relevant information. Whether we supplant "IETF" for the authors of
the RFC makes little actual difference to me.
> And contact information, as is discussed in the draft, depends upon contact
> for what purpose. For background on a registry, the working group is the
> contact. But for expert review, IANA is the contact, who then goes to the
> experts.
For a registration that has been made, the contact information of future
use can be all of: the registrants', the Expert Reviewers' (if
applicable), the WG list (if applicable), a directorate (if applicable).
The first of those (the registrant's) is redundantly available as the
RFCs' authors in the case of a registration arising from RFC
publication, so replacing it with "IETF" makes some sense, but it
neither adds nor subtracts information. The IANA already links to
relevant documents (such as RFCs) in registrations.
The fact that this proposal neither adds nor removes information makes
me thinkg it's at best just a matter of style. Couldn't the IANA
implement this change entirely of their own volition? Why hasn't it?
Couldn't the IESG simply request that editors apply this change to all
RFCs it reviews? Does the IESG require IETF approval to do that?
(Ironic, I know, because doing something like that might be perceived as
a power grab, but to me this is purely a matter style.)
Nico
--
