It's interesting that a current AD and a recently retired AD have such different views on the best procedure, and that suggests to me that there is no fully automated approach that will apply to all cases.
Also, would review-team reviews that are initiated before an IETF Last Call be reported to the last call list? If not, they are just another part of the normal WG process and we already have had them (early reviews) for years, as Mary noted.
So I don't think that fiddling with the automation will help here at all. If the review team members are too busy to respond in time, it's probably just another consequence of the size and breadth of the IETF and we have to live with it. As far as the formal process goes, the team reviews are very useful, but not formally required.
Regards
Brian
On 17-Oct-24 07:43, Eric Vyncke (evyncke) wrote:
Every AD has a different view it seems as I prefer to do my AD review first then wait for a revised I-D (if required) before starting the IETF Last Call and associated requests for directorate reviews. My reasoning is to prefer having a review on the latest I-D.
But your mileage may vary ;-)
-éric
*From: *Rob Wilton (rwilton) <rwilton=40cisco.com@xxxxxxxxxxxxxx>
*Date: *Wednesday, 16 October 2024 at 20:15
*To: *Barry Leiba <barryleiba@xxxxxxxxxxxx>
*Cc: *Murray S. Kucherawy <superuser@xxxxxxxxx>, John C Klensin <john-ietf@xxxxxxx>, iesg@xxxxxxxx <iesg@xxxxxxxx>, ietf@xxxxxxxx <ietf@xxxxxxxx>
*Subject: *Re: Telechat reviews [Re: Tooling glitch in Last Call announcements and records]
My suggestion is to automatically kick off the reviews as soon as it reaches WG “Submitted to IESG for Publication” state, rather than Last Call, which should be immediately after the shepherd review and writeup has been completed. I.e., before the AD has done anything with the document.
Regards,
Rob
*From: *Barry Leiba <barryleiba@xxxxxxxxxxxx>
*Date: *Wednesday, 16 October 2024 at 17:42
*To: *Rob Wilton (rwilton) <rwilton@xxxxxxxxx>
*Cc: *Murray S. Kucherawy <superuser@xxxxxxxxx>, John C Klensin <john-ietf@xxxxxxx>, iesg@xxxxxxxx <iesg@xxxxxxxx>, ietf@xxxxxxxx <ietf@xxxxxxxx>
*Subject: *Re: Telechat reviews [Re: Tooling glitch in Last Call announcements and records]
Working group chairs can always request early reviews at any time — they seldom do, but they sometimes do, particularly when they know a document is complex or needs someone with specific expertise to have a look.
Reviews are automatically requested by the tooling when the document enters the Last Call state (and the IESG Evaluation state with a telechat date set). That’s where we get most of the review requests, exactly because it’s automatic. No one has to think about it and (remember to) ask.
We could certainly use other state transitions (such as “In WG Last Call”) to trigger an automatic request, but I don’t think that would work well in general. Some working groups have multiple “last calls” for various reasons, and I think chairs would rather not have side effects such as this. Perhaps a separate check-box on state changes for “request directorate and review team reviews” would work.
In general, anything that requires a specific request will mostly not be used.
Barry
On Wed, Oct 16, 2024 at 12:32 PM Rob Wilton (rwilton) <rwilton@xxxxxxxxx <mailto:rwilton@xxxxxxxxx>> wrote:
Hi Barry,
Another choice, that perhaps could be considered, would be to initiate the directorate reviews slightly earlier in the cycle. E.g., at the point that the WG has said that is ready before publication but before the AD has reviewed and agreed to publish. In fact, input from the directorate reviews might be very helpful input to decide whether the document is really ready to progress, or if there are significant issues outstanding.
Of course, this might mean that a second follow up lighter directorate review is needed to cover any changes that occurred between the initial review and the version going before the IESG ballot, but if that second review was focussed on the differences and issues raised previously then I would have thought that the increase in workload on the directorate would probably be fairly small, and hopefully manageable. I.e., I am assuming that the second review would be assigned back to the originate directorate reviewer.
Generally, I think that it is better to get as many reviews as early as possible in the process when the folks working on the document and still very fresh and vested in getting the document published. Perhaps bigger changes to the process could also be considered …
Anyway, just a thought.
Regards,
Rob
*From: *Barry Leiba <barryleiba@xxxxxxxxxxxx <mailto:barryleiba@xxxxxxxxxxxx>>
*Date: *Wednesday, 16 October 2024 at 13:47
*To: *Murray S. Kucherawy <superuser@xxxxxxxxx <mailto:superuser@xxxxxxxxx>>
*Cc: *John C Klensin <john-ietf@xxxxxxx <mailto:john-ietf@xxxxxxx>>, iesg@xxxxxxxx <mailto:iesg@xxxxxxxx> <iesg@xxxxxxxx <mailto:iesg@xxxxxxxx>>, ietf@xxxxxxxx <mailto:ietf@xxxxxxxx> <ietf@xxxxxxxx <mailto:ietf@xxxxxxxx>>
*Subject: *Re: Telechat reviews [Re: Tooling glitch in Last Call announcements and records]
We should make it a general policy to add two weeks to the last call
period when a document is long, for some value of "long" (I might say
over 60 pages of substance (not counting change logs and such)). I
try to get to assigning ART-ART reviews a couple of times a week, but
that still means that, depending upon the timing, with a two-week last
call I might be giving a reviewer only a 7- or 8-day deadline for a
100+-page document, and I always blanch when I have to do that. While
ADs regularly have to review long documents with a week or two notice,
I think it's unreasonable to expect last-call reviews from
directorates/review-teams on that notice for long documents.
We decided on the two-week last call period at a different time, when
the IETF was a different organization. Maybe we should re-think it
now, and keep in mind that an extra two weeks of last-call review is
*not* going to be the most significant delay in a document's life
cycle.
Barry, ART-ART manager
On Fri, Oct 11, 2024 at 5:53 PM Murray S. Kucherawy <superuser@xxxxxxxxx <mailto:superuser@xxxxxxxxx>> wrote:
>
> Hi John,
>
> On Fri, Oct 11, 2024 at 2:01 PM John C Klensin <john-ietf@xxxxxxx <mailto:john-ietf@xxxxxxx>> wrote:
>>
>> Thanks for the clarification. Seems entirely reasonable with one or
>> two qualifications. First, if you (and/or other areas) are doing
>> things that way, the review needs to be posted to the Last Call list
>> well before the Last Call closes out so there is time for people from
>> the Area and the broader community to comment on it. Second, if the
>> posted end of Last Call date is unreasonable or unattainable for some
>> reason, I'd hope the responsible AD could be notified of that early
>> in the Last Call window -- at least no later than a week before it is
>> closed -- rather than, e.g., after the close date. That would permit
>> actions, if needed, to be taken without things looking like a game of
>> "Gotcha" with the AD and WG and/or author(s) responsible for the
>> document.
>
>
> For what it's worth, in my time on the IESG, I haven't found the need to manage this vigorously. If there's a directorate review I'd really like to have, I have the discretion to wait for it before scheduling the document onto a telechat even though Last Call has ended. If the review has come in but it provokes discussion, I have the discretion to wait for that discussion to resolve before moving forward. If we're talking about a document that isn't one of mine and a review comes in from my area review team raising something on which I'd like to dive deeper, I can use DISCUSS for that (so long as I am diligent about clearing it once the discussion is had, of course). That's been my strategy for a while now and it's never raised a complaint, which (so far, at least) includes the document you're talking about here.
>
> The thing I used to determine if the review has come in is the datatracker. I will check the last-call list too, but the datatracker provides a nice snapshot of which reviews have been requested and which have come in, and is usually where I start when checking on a document's status.
>
> Just to keep this all public: For this particular document, I have pinged the assigned directorate reviewers to ask them to upload their reviews ASAP on this document. As I said elsewhere, I might be fine advancing a document missing a couple of directorate reviews, but not all of them. If they don't come in soon, I'll reach out to the review team chairs to ask for reassignments.
>
> Lastly, I would definitely appreciate a notification (automated or otherwise) when a directorate review is going to be late. Right now all the tracker tells me is "not done", which could mean "not done yet" or could mean "don't hold your breath".
>
> -MSK
