- From: Jim Jewett <jimjjewett@gmail.com>
- Date: Wed, 23 Sep 2009 14:18:03 -0400
- To: HTML WG Public List <public-html@w3.org>
Still viewing the tables at http://dev.w3.org/html5/spec/Overview.html#annotations-for-assistive-technology-products-aria (1) <meter> should have a role=status, at least as a weak semantic. (2) <td> and <th> have the aria-labelledby set to the headers attribute. That covers explicit and oddly placed headers. But shouldn't it also take the table headers algorithm into account, and get the standard headers? http://dev.w3.org/html5/spec/Overview.html#header-and-data-cell-semantics If it is assumed that this would be redundant with the rowheader and columnheader roles and automatic determination from the DOM, that should at least be made explicit. Also, for purposes of aria-labelledby I assume -- but am not sure -- that the headers attribute would replace, rather than augment any standard headers, as it does in the table header algorithm. (3) Should thead and tfoot have the (abstract) role sectionhead? (4) Should tbody have the role of grid, or is that covered by the table parent? (5) For table, role must be either grid or treegrid. Most tables in practice are presentational, and should have role=presentation While such tables are themselves non-conforming, I do believe that honesty about it will leave AT less badly off. I would therefore suggest at least mentioning role=presentation. -jJ
Received on Wednesday, 23 September 2009 18:19:08 UTC