Re: Deliverable for Action 72 @headers

On Aug 24, 2008, at 11:14 PM, Laura Carlson wrote:

>
> James Graham wrote:
>> The
>> strongest way to make the case that these issues need to be  
>> addressed is
>> to go away and implement the current spec and show how it doesn't  
>> work
>
>> Findings from headers/id Testing (Bug 5822):
> http://esw.w3.org/topic/HTML/TableHeadersTestingBug5822#head-4dd98a1e6b2646ef8c2be83dd7b6c93622e25f4b
>
>> Headers/id allowed the assistive technology combinations tested to
>> successfully announce relationships 5 out of 6 times.
>>
>> Scope failed 6 out of 6 times. The failure of support for scope means
>> that scope is currently not an effective option.
>>
>> Although it is widely known that scope isn't well supported by
>> assistive technology [1 2 3 4 5], its use is strongly recommend,
>> because it's easy to author, works with simpler data tables, and
>> support is likely to improve.

I will also note that in addition to being poorly supported by  
existing UAs, the scope attribute has many ambiguities as its  
specified in the HTML 4.01 recommendation, as its specified in the  
HTML5 draft and I think in some sense its ambiguities are inherent to  
its very design. It may be these ambiguities that have retarded the  
support for scope in AT UAs.

I've outlined some of the problems on the wiki[1]. In particular the  
'colgroup' and 'rowgroup' keywords are a problem. It would be far  
better to add support in HTML5 for headers to reference other headers  
with the headers attribute and to otherwise treat the applicability of  
a header as ending when the first continuous range of data cells ends  
(when the next header cell is reached after a range of data cells).

Take care,
Rob


[1]: <http://esw.w3.org/topic/HTML/TableHeaderCellScope>

Received on Sunday, 24 August 2008 20:42:34 UTC