W3C home > Mailing lists > Public > public-w3process@w3.org > November 2014

Issue-100 - proceeding with open exclusion periods

From: <chaals@yandex-team.ru>
Date: Wed, 26 Nov 2014 14:49:23 +0300
To: Revising W3C Process Community Group <public-w3process@w3.org>
Message-Id: <63831417002563@webcorp01h.yandex-team.ru>

TL;DR: I don't think we need a change, but I won't be fighting hard against one. 

Currently, it is *theoretically* possible to publish a Recommendation while the CR exclusion period is still open:


On entering CR, a 60-day exclusion period begins.
The CR period must be four weeks (i.e. 28 days) long.
The Proposed Recommendation review period (which can theoretically begin the day after the CR period ends) must be 28 days.
If there is no dissent, the director's decision could be made and the document published 56 days after the CR period begins - i.e. while there are 4 days left in the exclusion period.

The current process explicitly states that the PR review *should* end at least 10 days after the final exclusion period, and in practice entering PR the day after finishing CR (equivalent in the old Process to a zero-length CR period immediately after the period for last call comments closes) is rare and indeed very difficult to organise. Note that under the current process, AC comments may come in at the *beginning of *CR**, not just during the PR period. One reason for this is to give a sense of whether the spec under review is urgently desired by industry, or possibly only relevant to the editor and a friend, which in turn should help W3C work out how much they need to prioritise getting the spec through the process.

The "threat model" is as follows: 
FooAPI 1.0 is developed by the MadScience WG, and enters CR at 00:00 on 1 october. 
All members of the WG explicitly declare that they have no patents to exclude by 3 October, there are no comments (after all, it is a great spec), and at 00:00 on 29 October a PR is published. 
The responses are entirely positive, and on 26 November (thanksgiving blackouts being a thing of the past) a Recommendation is published. 
On 27 November, the "Cool Science In Real Operations" anarcho-syndicalist collective joins the working group, since they are very interested in FooAPI 2, and announce that they have a patent they plan to exclude. Which, it turns out, is as essential for FooAPI 1.0 as FooAPI 2.0. 

(Note, this skates over some issues where the Patent Policy itself is apparently not clear, but in any event a Recommendation which is known to be covered by an patent that isn't offered RF is a problem, and depending on how you interpret the policy the scenario can be changed to fit - e.g. the collective never joins the WG but announces their patent and outrageously unreasonable dsicriminatory terms, or some other variation).

So, what to do?
That this is at all possible has been the case forever. In practice it assumes no comments requiring much time to process, a desire on the part of the director to get everything done as fast as possible, and no dissent in the last 10 days of PR review - under the current process any dissent must be answered in public 14 days before the director can make a decision, which would blow the ability to produce a *Recommendation*.

There are some recommendations (if it becomes one, longdesc springs to mind) which are entirely based on work done 15, 20 or more years ago, for which it is truly unlikely that there are any patent concerns. Nonetheless, members may feel happier with a process that does not allow the freedom of judgement to W3C. The issue proposes that a constraint be added so a PR cannot be published while an exclusion period is open.

This is certainly fairly trivial to achieve in the text. There several possible approaches, including extending the CR review period to match the exclusion period, or stating that for a document to become a PR there must be no open exclusion opportunity.

My personal preference is to leave the situation as it stands, making the situation where a PR goes out before the exclusion period closes unlikely and getting to Rec *almost* impossible, but allowing it in principle. This matches the "moar agile" aspirations of the Process Task Force, but I also don't think the question is particularly important - adding an extra delay of 5 weeks at the very end of the Process, by which time we expect the decision to be a formality when everyone has done their job well, is unlikely to be a huge problem in the real world.



Charles McCathie Nevile - web standards - CTO Office, Yandex
chaals@yandex-team.ru - - - Find more at http://yandex.com
Received on Wednesday, 26 November 2014 11:49:55 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:51:23 UTC