Re: References and Modularity

[ Bcc: public-w3process, chairs and w3c-ac-forum; please reply to 
www-tag@w3.org ; Robin's e-mail to www-tag is archived at 
<http://lists.w3.org/Archives/Public/www-tag/2013May/0049.html> ]

On 5/29/13 6:02 AM, ext Robin Berjon wrote:
> The common lore on this issue is that you can only reference stable
> documents, or documents that are at most one degree of maturity behind
> your own. In truth, that requirement is not a solid one: Process allows
> the Director to decide whether (s)he feels the way a specification
> handles references is satisfactory or not. That said, "satisfactory" is
> a fuzzy concept and while it remains undefined the natural tendency of
> cautious stakeholders will be to reach for the strict interpretation in
> order to be on the safe side.

Thanks for starting this thread Robin. (I started a related Draft 
several months ago and I am delighted you beat me to it ...)

I agree the current out-of-band normative reference policy is suboptimal 
and in some cases I argue it is actually harmful. For example, if/when a 
Proposed Recommendation (PR)  is blocked solely because of this policy, 
implementers, developers, etc. are not protected by the IP commitments 
that start when the PR is published as a W3C Recommendation. This 
scenario isn't fiction - it is true for three PR in WGs I chair and I 
would not be surprised if the reference policy is also blocking other 
PRs (Geolocation?, others?).

I don't have a strong preference on how this issue is addressed other 
than I would, naturally, prefer a very lightweight approach. To that 
end, I'd like to see WGs have the final "say" on the reference decision. 
After all, it is the WG members that typically have the most skin in the 
game re implementations and thus they are best suited to determine the 
related risks (e.g. if interop problems will occur if/when a reference 
changes). If a WG agrees to publish a Candidate Recommendation (CR) with 
references that are CR or less, then I think that decision should carry 
a lot of weight.

If the policy permitted more WG autonomy as described above, there could 
be a requirement that all W3C normative references must be to a dated 
spec and that all such refs must be at least a FPWD. For Candidate 
Recommendations, the Status of the Document statement could explicitly 
state that if implementers consider the "maturity" of a normative 
reference a substantial issue, they should formally raise an issue 
during the implementation phase. In the absence of such issues, the 
group can then (accurately, IMHO) conclude the CR's references are 
sufficient to advance the spec to REC and the Director should honor the 
group's decision.

-AB

Received on Wednesday, 29 May 2013 12:02:31 UTC