[html4all] is SCOPE sufficient - a test based on WCAG2 techniques example

[Joshue O Connor]

Hi Gregory and Gez,

Are you guys talking about using the scope attribute as a way of marking
up tables instead of using header/id combinations, rather like:

<table class="Scope_only_table" summary="This is a table testing use of
the scope attribute instead of header/id combinations>
		<tr>
			<th scope="col" abbr="First Heading">Rather long first heading</th>
			<th scope="col" abbr="Second Heading">Rather long second heading</th>
			<th scope="col" abbr="Third Heading">Rather long third heading</th>
		</tr>
		<tr>
<td>Stuff related to first heading</td>
<td>Stuff related to second heading</td>
<td>Stuff related to third heading</td>
</tr>
<tr>
[...]

and so on. This kind of table works really well in JAWS 8. I know that
because lack of support for the scope attribute in earlier screen
readers (< JAWS 6 IIRC) and this is of course a concern. Or do you guys
think this is not an ideal way to mark up tables in principle?

> In the absence of a headers attribute, JAWS and Window-Eyes guess at
> headers for a table cell, regardless of the scope attribute [...] If scope
> was well-supported by screen readers, it would still be problematic to
> depend on it, as that type of information is generally only revealed
> when the user is in browse mode. 

So to be clear, when the headers attribute is not present within the
table then the screen reader defaults to using heuristics to figure out
what the header and data cell associations are, and hopefully the
headers make sense so the user gets some common sense output from the
screen reader and can therefore figure out what cell relates to what
header, right?

How about the (simple partial table) example I gave which uses a
combination of the header and @scope?
Gez/Gregory do you think this kind of method of marking up an accessible
table is ok?

Cheers

Josh