Re: ISSUE-10 Raising awareness before CR

Reply inline

On Jun 14, 2013 4:44 PM, "Wayne Carr" <wayne.carr@linux.intel.com> wrote:
>
> On 6/9/2013 10:24 PM, Larry Masinter wrote:
>>
>> Karl, thanks for doing this. But I would like to see if there's more
clarity about what the problems are, somewhat independent of the proposed
solution (if you
>>
>> https://www.w3.org/community/w3process/track/issues/10  just asks
>> " Is there a need for a "Call" whether called "Last Call" or not that
precedes CR and indicates that the WG believes it is done (no open issues)
and a last review should be undertaken? This may be different than Last
Call in the current process."
>> but I'm not sure what problem this would solve? What's wrong with "last
call" now? Are there things linked to Last Call that you don't want to
invoke, that you need another call?
>
>
> Last Call mixes several important but distinct things.  1. formal patent
exclusion period that where you don't want to keep setting that off
needlessly; 2. notice to community that you would really like review; 3)
notice that you think you're completely done other than implementation
feedback and testing feedback.
>

I think there are reasons for this granularity. For 1. companies dislike
excluding patents unnecessarily, so they want the spec to be as stable as
possible, so they want to wait till things are implementation ready (ie fit
criteria 3.)

For 2. people are naturally going to come out in greater numbers when they
think this is their last chance to make a difference and/or their
difference is unlikely to be overridden later in the process (again thus
waiting for 3.)

Getting the "community" to do anything is like herding cats. It is easier
to either go where the cats are or to lay out some cat food. What that
means I am not exactly sure.

Though let's try this. Announce each review phase at high volume community
group mailing lists. Also, offer to give credit in the spec to official
reviewers who must comment on every spec phase. Maybe also promise them
pizza (developer food).

> The granularity of a lot of that is wrong.  Different parts of a spec
stabilize at different times.  They're also implemented at different times.
 Doing that on the whole document waits until too late.   Having Last Call
for the first purpose is part of the patent policy, but no other review has
to be called Last Call (so combining with CR like Chaals did is good).
>
> I think it would be useful to have an announcement mail list associated
with each WG for important status announcements and calls for review (in
lower case, not a formal step).  This would be a low bandwidth read only
announcement list, not for discussion.  It could have things like that the
group has asked to go to last call - or thinks it is done with a particular
capability and would like feedback on that.  Sections could be marked
"stable", "under development", etc. and important transitions could be
announced with requests for feedback (feedback on the public list, not the
announcement list)
>

I think this is a great idea. This way if a spec is being delayed because
of one or two features (or only one or two features are ready) the
community will have a sense of which parts of the spec they can use (since
implementors are likely to start implementing delayed specs whether or not
they should do so)

Ideally for long delayed specs the stable sections could spun off. Limiting
the impact of any disputes.

> I like the way Chaals combined LC and CR and also ensured the AC gets
final say again when things change.  It also retains the stable CR part for
things like specs for ebooks to reference.
>
>>
>> Larry
>> --
>> http://larry.masinter.net
>>
>>
>>
>
>