- From: Thierry MICHEL <tmichel@w3.org>
- Date: Thu, 27 Jul 2006 09:52:24 +0200
- To: Lofton Henderson <lofton@rockynet.com>
- CC: WebCGM WG <public-webcgm-wg@w3.org>
Lofton Henderson wrote: > Hi Thierry, > > Thanks for getting this started. > > I have one comment for now, embedded... > > At 10:02 AM 7/26/2006 +0200, Thierry MICHEL wrote: > >> I have drafted a WebCGM 2.0 CR version cover page. >> http://www.w3.org/Graphics/WebCGM/drafts/thierry-editor/overview.html >> >> >> Apart from the mandatory W3C info in the "Status of this Document" >> section, I have written the following exit criteria: >> >> >> The WebCGM Working Group expects to request that the Director advance >> this document to Proposed Recommendation when the following exit >> criteria have been met: >> >> 1. Sufficient reports of implementation experience have been gathered >> to demonstrate that the WebCGM 2.0 features are implementable and are >> interpreted in a consistent manner. To do so, the Working Group will >> insure that all features in the WebCGM 2.0 specification have been >> implemented at least twice in an interoperable way. This defines this >> as : >> >> * the implementations have been developed independently, >> * each test in the WebCGM 2.0 test suite has at least two passing >> implementations. > > We should pay attention to a subtlety of wording here. WebCGM 2.0 > includes the functionality of WebCGM 1.0, plus new stuff for 2.0. The > WebCGM 2.0 TS includes the WebCGM 1.0 TS, plus new 2.0 tests. > > WebCGM 2.0 is not a "delta" specification, and neither is the WebCGM 2.0 > Test Suite a "delta" (in my view of things). I agree. And btw W3C does not favor delta specs. On the other hand I think > this exit criterion does need to be "delta". I.e., there should be two > passing implementations for every *new* 2.0 feature. Yes this is exactly what I had in mind,and this is why I wrote: "each test in the WebCGM 2.0 test suite has at least two passing implementations." and "The WebCGM 2.0 test suite will provide at least one test case for any new feature introduced in WebCGM 2.0, covering the new DOM-related and XCF features, and the new static and "intelligence" feature." Assuming the new features do not impact old WebCGM features, which I believe is the case. FYI, the SMIL 2.1 spec we had the same use case [1]: - A spec 2.1 adding features to a 2.0 spec (but not a delta spec in terms of edition) - A 2.1 testsuite listing only the new features of SMIL 2.1. - The implementation report listed only the new features, as previous 2.0 features were of course well implemented. [1] http://www.w3.org/TR/2005/CR-SMIL2-20050513/ > > Why? Because 1.0 is already a Recommendation, since 1999. That was > before the present CR/exit-criteria stuff was part of W3C > process/convention (correct?). Although I fully expect that there are > at least two "pass" for every one of the existing 254 1.0 tests, on the > other hand it is somewhat academic, because we can't rescind features of > the 1.0 Recommendation if not. sure. > > Viewed another way, we shouldn't attempt "ex post facto" to apply more > recent W3C Process conventions to content of existing Recommendations. > > (All) Does this make sense? It sure does and will save a lot of time. > > >> 2. The Working Group releases a public test suite for WebCGM 2.0 along >> with an implementation report. >> >> The WebCGM 2.0 test suite will provide at least one test case for any >> new feature introduced in WebCGM 2.0, covering the new DOM-related and >> XCF features, and the new static and "intelligence" featuresDraftdraft. >> >> The Working Group expect that no feature has been identified as at >> risk at this point. >> >> If we have no incoming Last Call comments tomorrow to deal with, we >> may discuss these exit criteria: > > Yes, it is on the just-sent agenda. > > Thanks, > -Lofton. > > -- Thierry Michel W3C
Received on Thursday, 27 July 2006 08:24:04 UTC