This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
Date: Wed, 25 Jun 2003 21:11:37 -0700 From: Tantek Çelik <tantek@cs.stanford.edu> To: <www-dom-ts@w3.org> Message-ID: <BB1FC048.29C54%tantek@cs.stanford.edu> Subject: Re: Concerns regarding the W3 Document Object Model (DOM) Conformance Test Suites .... I am concerned about the following DOM2 core tests which appear to require that the implementation use a validating parser since the tests assume that the reading of external DTDs are supported: getAttributeNS01.html getElementByID01.html hasAttribute02.html hasAttributeNS04.html namespaceURI01.html removeAttributeNS02.html In my opinion these tests should be made optional and marked as such, since they effectively require a validating parser as described above. Even if some implementations have hacked default attribute values without a validating parser, that is no reason to require such hacks of other implementations. I tried searching the www-dom-ts archives regarding the issue of requiring a validating parser and/or default attribute values from external DTDs, and this message (that I am replying to) appeared to be the latest on the subject. http://lists.w3.org/Archives/Public/www-dom-ts/2002Feb/0170.html Was the issue brought up during the 2002 DOM WG Meeting in Cannes as [dd] indicates, and if so, what was the resolution? Or has this been discussed in a more recent thread? Thanks, Tantek ------------ Curt Arnold writes: The resolution was communicated in http://lists.w3.org/Archives/Public/www-dom-ts/2002Mar/0002.html and several DOM L1 Core tests were marked as only appropriate for validating parsers. Since the resolution came as part of the release cycle for the DOM L1 test suite, it is very likely that the corresponding fixes were not done for the L2 Core tests. My response to the WG resolution was http://lists.w3.org/Archives/Public/www-dom-ts/2002Mar/0021.html which did not mention any L2 tests. That thread started a lengthy discussion, but I don't believe it changed anything. I would suggest a similar approach to that undertaken with the L1 tests. Find a processor that passes all the tests as currently written (probably Xerces-J), remove the default attribute values from the DTD's and mark the tests that fail are requiring validation.
Created attachment 54 [details] Patch file, unfortunately a line end change causes everything to appear changed
Added <implementationAttribute name="validating" value="true"/> to getAttributeNS01, hasAttribute02, hasAttribute04, importNode07, namespaceURI01, and removeAtrributeNS02. getElementByID is a different issue and has been spun off as bug 383 See http://lists.w3.org/Archives/Public/www-dom-ts/2002Mar/0050.html