Given that even the proposed amends tag has two different meanings, and
how nuanced language gets, I think tryng to fix the tag is the wrong
answer. Instead, lets clarify the expectation that a draft is explicit
and clear about what it is changing relative to RFCs it updates.
Otherwise, this seems to be a recipe for more argumentation about
which if any tag a specific draft needs.
That also leaves to the IESG whether the specific effort to make
something retroactively mandatory is appropriate to the specific case,
without getting into defining tags that appear to endorse such treatment.
Yours,
Joel
On 7/7/2026 6:31 AM, Mirja Kuehlewind (IETF) wrote:
The fact that you said below multiple “to me” makes it very obvious
that everybody is applying their own set of rules to “updates” and the
problem is these don't match. Also again, the draft doesn’t require
the use of these tags, however, it requires that if you use them, they
need to align with the provided definition. I guess we could also
discuss this more in the draft, where it makes sense to use them, e.g.
if there is a registry you might no need the tag or give more examples
about other kind of known extension points. I will open an issue.
On 7. Jul 2026, at 10:43, Eliot Lear <[email protected]> wrote:
Hi Brian,
On 07.07.2026 08:13, Brian Rosen wrote:
In my draft, I added a new interface. It’s optional. It doesn’t use anything
that was previously unused. It’s an entirely new interface that provides some
useful information the existing interface doesn’t provide.
To me, the question boils down to whether implementation of your
draft would violate the base specification. If it does, then Updates
seems appropriate (at a bare minimum). If it doesn't, then no. If
the WG can't answer that question crisply, then the underlying
protocol may indeed have some issues, requiring an Update (or more),
in order to handle its evolution; although I take your point about
the pain of Obsoletes. We're about to go through that with TEAPv2.
Michael's use of Amends in 9914 to me is a classic case of Updates,
where a field that was required to be set to zero is now the P flag.
Updates needed (again, at a minimum).
As to Extends, or more precisely Extended-by, were it implemented, it
could not be in the front matter because protocols like SMTP* and TLS
would become unreadable. I'm still not clear on whether it would add
more value or confusion to the reader.
Eliot
* To answer Michael's question, there has only ever been a single MTI
on port 25 since RFC 1123, and that is EHLO, which is the only
extension that is mentioned in 5321bis.
<OpenPGP_0x87B66B46D9D27A33.asc>--
rswg mailing list -- [email protected]
To unsubscribe send an email to [email protected]
--
rswg mailing list -- [email protected]
To unsubscribe send an email to [email protected]
