Re: General comments for Proposed Chapter 7

On Thu, 28 Nov 2013 03:34:00 +1100, Sylvain Galineau <galineau@adobe.com>  
wrote:

> On 11/27/13 5:28 AM, "Arthur Barstow" <art.barstow@nokia.com> wrote:
>
>> Here are some general comments re the proposed Chapter 7 ...
>>
>> * The proposal includes useful editorial cleanup and simplifications but
>> suffers from some organizational issues as captured in Issue-59.

Yep. As noted in the response to that.

>> * I like the elimination of Proposed Recommendation. [If AC reps are
>> interested in a spec, they should be engaged much earlier in the process
>> than PR. Additionally, PRs are mostly opaque to WG and require quite a
>> bit of `make work` for Editors.]

Glad to have got something right.

>> * In practice, I don't think the elimination of LC or combining LC and
>> CR [depending on how one spins the gist of the proposal]
>> provides a significant improvement,


I think of what we have done as:
move some of Last Call (in particular the Patent Review required to decide  
whether a member wants to make a Patent Exclusion) to CR; and
push Working Groups to get general review from the public done closer to  
when parts of a spec are "stabilised" - or changes are made - rather than  
waiting until there are preliminary implementations that have been around  
for a year or two which increase the cost associated with even changes  
that we would like to make.

>> and as I stated last June, it appears to just create a bunch of new
>> issues. F.ex. it appears a new process will be needed re "what is Wide
>> Review, how are Reviews done, who is responsible for doing what" (which,
>> IMHO is precisely the point of LC as defined in Process-20051022).

I don't think this is a new issue. There is a requirement at CR to show  
wide review, and what we have done in the current draft is give some  
clearer guidance than the document did about what the Director will  
consider in evaluating whether that requirement has been met.

> I've rarely seen a LC produce wide review, or what I think of as a 'wide
> review'. 4-6 weeks may be enough to achieve wide-enough review within the
> folks involved in W3C activities on a daily basis. For the wider  
> community of web experts and interested parties this is just far too
> short.

Indeed.

>> * As far as I can tell, the gist of the LC+CR proposal can be achieved
>> within the context of Process-20051022.

Sure. In fact, if the group has used the existing process effectively to  
do so, Publishing a pre-LC and then an LC (as many groups currently do) is  
just another bit of busywork, which will lead to almost no comments and  
almost no significant changes.

The Longdesc Extension, and Custom Elements, are two specifications whose  
Last Call seems to have required bureacratic work that slows down the work  
for no apparent benefit.

>> The underlying issue the proposal appears to try to address is "how to
>> prevent a spec from entering the dreaded LC->CR->LC->CR->... cycle".

Yep. For Working Groups, the benefit is skipping a Last Call that is  
unnecessary. For the Public, the benefit is seeing a clearer sense of  
progress (or at least spinning at a higher lever) if a spec is really not  
going backwards in any meaningful sense.

>> Of course the proposal doesn't eliminate the cycle problem (a spec can
>> still have a LCCR->LCCR->... cycle), it just appears to "shift" the
>> problem.
>
> To some extent, yes, the new CR/LCCR is now preceded by an implicit and
> undefined Review stage that somehow makes it far more likely the former
> will 'stick'.

Exactly.

> So one could argue we didn't eliminate LC as much as renamed it 'wide
> review' and left its implementation up to WGs?

To a large extent. There is a change of focus regarding "heartbeat" drafts  
(Revised Working Drafts), from 'every X months' to 'when there have been  
significant changes to the document that would benefit from review from  
beyond the Working Group' (the current text). This is intended to  
facilitate earlier review, based on what bits need review at any given  
time, without trying to define a requirement in formal terms.

While the earlier review is important IMHO, I don't think trying to define  
a formal requirement would bring enough extra benefit to be worth the  
constraints it would impose. So yes, we ultimately leave implementation to  
the Working Group.

cheers

Chaals

-- 
Charles McCathie Nevile - Consultant (web standards) CTO Office, Yandex
       chaals@yandex-team.ru         Find more at http://yandex.com

Received on Thursday, 28 November 2013 13:53:23 UTC