On 17 Mar 2024, at 11:41, John C Klensin wrote:
--On Saturday, March 16, 2024 19:17 -0400 Keith Moore
<moore@xxxxxxxxxxxxxxxxxxxx> wrote:
On 3/16/24 17:20, Carsten Bormann wrote:
The current times (close 2 weeks before, open 1 day before)
are exactly what is needed in a good number of cases, and
their consistency helps people who want to do work across WGs.
Nothing's perfect, but IMO the current 2 week deadlines are a
good compromise most of the time.
I agree. Two weeks is about right...
So, as I said earlier, many things have changed, and in this case I
disagree that the 2-week moratorium on posting is needed anymore. As
Keith pointed, there were two reasons we set up the moratorium in the
first place: (1) The secretariat had to hand-process I-Ds back in the
day and was swamped before the meeting; and (2) In order to be prepared
for the meeting, you wanted to have people to read the same version of
the document so that everyone was on the same page. Only later, as a
side effect, did we get reason (3): People act well in the face of
deadlines, and this one was conveniently imposed due to other
considerations.
However, several things have changed. Justification (1) has completely
gone away; the tools take care of things.. But even in the case of (2),
I think the justification has mostly gone away. It used to be that
posting a changed version would really be a problem: If some edits went
in late, you would have to re-read the whole document and try to figure
out what changed, and you'd had better kept the old version, because
once the new version was out the old one was deleted from the server.
But nowadays, we keep the old versions on the server, and we have a
perfectly reasonable diff tool, where you can choose any version and get
the diffs from the current version in an easy to read format. It's no
longer such a big deal to read whatever version is available when you
are preparing fore the meeting, and then look at the diffs just before
the meeting. (And again, remember that the moratorium ends now at the
beginning of the week, so it is not even effective for purpose as it
stands.)
That leaves the "deadlines are useful" reason. And that's a perfectly
good reason, on a WG-by-WG basis, to allow chairs to close the
submission for their WGs if they want to impose a deadline. But it
doesn't justify shutting the queue for all WGs (some of which might not
be meeting f2f) or all documents (some of which might not be associated
with any WG or RG), or all chairs (some of whom might have good reasons
for a late add).
and, at least last I checked,
we had procedures for ADs to make exceptions. I assume that, if
we needed exception procedures for other streams, we could
probably make them up in a hurry (if they don't already exist
without my noticing).
But this seems to me too high a burden. If a chair wants to make an
exception, they should be empowered to do so and not make this depend on
an AD OK, particularly right before a meeting where ADs have lots of
other things to deal with. And if a chair or an AD is not directly
involved, there is no reason an author shouldn't be able to submit a
document that has nothing to do with a WG.
We are using the accident of an old set of circumstances to drive
procedures rather that discussing what we really want out of the
tooling. Please let's stop doing that.
(During a chat last night, Barry reminded me that when a change was
proposed several years ago, some chairs objected to the change because
they did not want the responsibility to allow exceptions and instead
wanted it to be an AD override so they could claim powerlessness to
insistent authors. I find such an argument a sign of complete
dysfunction.)
The question that started the thread is whether other mechanism
of getting documents posted --other than, e.g., mailing list
discussions-- frustrate the intent of that two week limit and,
if so whether they are reasonable. I had intended to open up,
and ask for community consideration of, the much broader set of
questions and was not asking about localized patches. Pete
clarified that point and said what I should have said more
explicitly -- that we need to look at the whole collection of
interrelated issues rather than applying isolated small patches.
I would add that we should not get distracted by possible
patches to the point of losing sight of that collection of
issues. I don't know that it is what either Keith or Carsten
intended, but a discussion of any of whether the cutoff should
be two weeks before the meetings start, what "starts" means for
that purpose, whether it should be 14 days, 20 or 10 days, one
day, some other number, or abolished entirely would be, IMO,
just the sort of distraction from the larger issues that I think
Pete and I are concerned about.
Indeed, even talking in terms of a "posting cutoff" is a mistake. If
mailing list discussion within the moratorium period comes to consensus
(with text) on a particular issue in a document, that is just as
problematic (or not) in keeping everyone on the same page preparing for
f2f discussions. Does that mean that we should have an IETF-wide 2-week
moratorium on mailing list discussions before f2f meetings? Of course
not (I hope). But some chairs may wish to say, "On issue X, let's close
discussion on the list until the f2f meeting, as I think a more
interactive discussion (perhaps with people with
Transport/Security/I18N/whatever expertise present) would be more
helpful." And that would be fine. Again, let's discuss what we actually
want to have happen, and then decide if we need some grand principle or
rule imposed, at an IETF/Area/WG level, for that and what tooling (if
any) we need to make it happen.
So, btw and IMO are discussions on the tools list about ways to
make Github more or less efficient for document review and
whether it is time to prohibit some I-D preparation tools or
formats. They are probably all part of the larger set of
questions but, like others of those questions, they raise
important policy questions including, again IMO, just what they
IETF means when we talk about being open, welcoming,
transparent, etc., and what we all mean when we say that
something represents IETF community consensus.
Yep. See above.
I'm less sure about opening submission to new drafts one day
before a meeting. I might prefer that they be opened again the
day after the meeting closes.
Again, part of the broader problem (with interactions with other
parts), as are the questions of when WGs are expected (or
required) to firm up agendas and organize and post meeting
materials.
And questions about who/what is imposing the expectation/requirement
(e.g., the chair, the AD, the IETF as a whole, the tooling, etc...)
pr
--
Pete Resnick https://www.episteme.net/
All connections to the world are tenuous at best
