Re: Last Call: CR-LDP Extensions for ASON to Informational

[Harald Tveit Alvestrand <harald@alvestrand.no>]

--On tirsdag, januar 14, 2003 23:00:51 +0000 Bob Braden <braden@ISI.EDU> 
wrote:

>   *> If code points are to be allocated from the space to be allocated by
>   *> IETF Consensus, I strongly suggest that a *Standards Track* document
> be   *> written, with more detail on the messages, especially their
> processing.
>
> Indeed, that is what "IETF consensus" means, isn't it?
Bob,
as I said when we discussed this a bit earlier:

RFC 2434 defines this term, and it says, in extenso:

     IETF Consensus - New values are assigned through the IETF
          consensus process. Specifically, new assignments are made via
          RFCs approved by the IESG. Typically, the IESG will seek
          input on prospective assignments from appropriate persons
          (e.g., a relevant Working Group if one exists).

          Examples: SMTP extensions [SMTP-EXT], BGP Subsequent Address
          Family Identifiers [BGP4-EXT].

Quoting from the references: [SMTP-EXT] is RFC 1869:

  The IANA maintains a registry of SMTP service extensions.  Associated
  with each such extension is a corresponding EHLO keyword value. Each
  service extension registered with the IANA must be defined in an RFC.
  Such RFCs must either be on the standards-track or must define an
  IESG-approved experimental protocol.

[BGP4-EXT] is RFC 2283:

  Subsequent Address Family Identifiers (other than those reserved for
  vendor specific use) are assigned only by the IETF consensus process
  and IESG approval.

There are circumstances (IS-IS standardization is one; experiments are 
another; other non-IETF-managed extensions are a third) where it makes 
sense to have non-standards-track documents allocate values in registries 
where the IETF intends to keep some kind of control.

The IESG has made some use of the Last Call mechanism for these documents; 
that seems to be one way to implement "the IESG will seek input".

RFC 2434 reserves the term "standards action" for the case where a 
standards-track RFC is required; there is a reason why it defines two 
terms, not one.

                        Harald