Re: CfC: Request transition of DOM4 to Proposed Recommendation

On 10/7/14, 6:52 AM, Paul Cotton wrote:
> Test results for this specification can be found at [4-5].  An analysis
> of the test results is available at [6].

Sadly, I must object, due to lack of implementation experience for some 
parts of this specification (as in, there are 0 implementations that 
actually implement some parts of this).

Of most concern to me are these two items:

1) Are none of these tests specifically testing createElement on a 
non-HTML document?

I believe the spec text there calls for an HTML element to be created, 
which doesn't match any implementation (see 
http://jsfiddle.net/003c04ew/1/ which alerts null in Safari, IE, Chrome, 
and Firefox but per spec draft should be alerting the HTML namespace).

Note that this is kind of tested in an oblique way by 
http://www.w3c-test.org/dom/nodes/Document-constructor.html but that 
part of the test is never reached in non-Gecko browsers because of lack 
of support for the Document constructor, so part 4 of the test 
presumably got written off as "just lack of support for |new Document|" 
or something during the analysis.

I have serious concerns about whether the spec requirement here is 
web-compatible, given that it changes existing behavior for the case of 
SVG documents as far as I can tell, and that no implementations actually 
do this so far.

2) The fact that all browsers fail the whitespace-related tests for 
classList.add/remove also makes me worried about whether the behavior 
the spec calls for there is web-compatible.  I would be much happier if 
we had at least _one_ implementation that was shipping with the spec 
behavior here.  This is a smaller worry than item #1.

That said, any row in that table that tests functionality in this 
specification and has "FAIL" across the board should be at least 
somewhat worrisome.

-Boris

Received on Tuesday, 7 October 2014 15:55:16 UTC