- From: Joshue O Connor <joshue.oconnor@cfit.ie>
- Date: Fri, 27 Jun 2008 09:19:07 +0100
- To: Leif Halvard Silli <lhs@malform.no>
- Cc: Dan Connolly <connolly@w3.org>, Robert J Burns <rob@robburns.com>, "public-html@w3.org WG" <public-html@w3.org>, public-comments-wcag20@w3.org
Leif Halvard Silli wrote: > Currently, under HTML 4, the screen reader follows the "algorithm to > find heading information" (Section 11.4.3 in HTML 4). The relevant > section -- for H43 -- says: "search upwards to find column header cells". > > H43 says that @headers "allows screen readers to speak the headers > associated with each data cell when the relationships are too complex to > be identified using the th element alone or the th element with the > scope attribute". > > Unfortunately, the table example of H43 does not fit the tag, as it is a > very simple table. (However, to the defencence of H43, the focus of H43 > is what *screen readers* are capable of -- and not what the HTML spec > has specified.) Note that the table algorithms from HTML4 are not correctly implemented by screen readers. Ben Millard and James Grahams work on tables uncovered this, which was very interested and myself and Steve F talked with them about this in detail at the F2F last November. Having said that being aware of how these algorithms are implemented is *very* important as it obviously effects the user experience and the quality of user interaction - and they do by and large work - just not as per specification. However this point could be moot. > There are complex/untypical tables which do need the @headers attribute > in data cells. But I agree with Rob in that it is more important to > allow @headers on TH cells than to allow @headers on data cells. Considering how current UAs deal with tabular data, then yes. Cheers Josh
Received on Friday, 27 June 2008 08:20:28 UTC