- From: David Singer <singer@apple.com>
- Date: Thu, 13 Aug 2015 12:45:21 -0700
- To: Adrian Bateman <adrianba@microsoft.com>
- Cc: "rob@blaeu.com" <rob@blaeu.com>, Carl Cargill <cargill@adobe.com>, "public-tracking-comments@w3.org" <public-tracking-comments@w3.org>, team-tracking-chairs <team-tracking-chairs@w3.org>, John Simpson <john@consumerwatchdog.org>
> On Aug 13, 2015, at 12:27 , Adrian Bateman <adrianba@microsoft.com> wrote: > > On Thursday, August 13, 2015 3:11 AM, Rob van Eijk wrote: >>>> Based on conversation with editors, we believe these typical CR exit >>>> criteria should apply to the TPE Candidate Recommendation that we >>>> expect to publish after a Director decision: >>>> >>>> * Multiple (at least two), interoperable implementations of each >>>> feature, excepting features marked at risk >> >> If the requirement >= 2 is not a normative W3C requirement, I would >> rather suggest that >=1 implementation would suffice to keep the fetaure >> at risk in the recommendation. The main reason for this perspective is >> that implementation may be broader if all implemented edge-cases are >> taken into account, even if there is only one such a case. > > You can't prove interoperability with one implementation. There must be at > least two. Some groups require more than two to progress depending upon > complexity and risk of adoption. The point is to show that different groups > can read the spec and arrive at interoperable implementations. > > Adrian. well, if (for example) Microsoft’s implementation of a given feature works with a multitude of sites, then you have proven interoperability to some extent. However, we desire that those same sites also work with at least one other independent implementation, so we have consistent interoperability, i.e. some evidence that implementers at both ends read the spec. consistently. (The same issue might arise if only one site uses the feature, of course). David Singer Manager, Software Standards, Apple Inc.
Received on Thursday, 13 August 2015 19:46:14 UTC