In Firefox we rely much on layout. If something was rendered as a table then it's exposed as table on accessibility layer too. There are strong voices that CSS display shouldn't affect on content semantics because the author has ARIA for that, and I find this argument reasonable. On the other hand this approach may harm web apps that don't use much ARIA and relies on the browser to get some a11y for free. I would love to have consistency here, I'm not sure though how it can be achieved. Thanks. Alexander. On Thu, Apr 30, 2015 at 7:23 AM, Steve Faulkner <faulkner.steve@gmail.com> wrote: > Something that has been bugging me is the effect of CSS display properties > on elements that are normally not represented in the acc tree > > http://s.codepen.io/stevef/debug/mJJOKq? > > in the example above, depending on the browser (checked only > IE/Firefox/chrome on windows) changing the CSS display value of a span, or > adding a tabindex=0 may change its role(s) reported via APIs, add it to the > acc tree or do nothing, > (note: chrome adds in seemingly random Ia2_section/divs to output). > > There is no uniformity of behaviour across browsers. > > Part of the issue (i think) is that expected behaviour is not defined > anywhere. > and think it should be defined: > effect of CSS display properties > effect of tabindex. > > > I think that the disparities would cause issues for AT using the acc APIs, > but am happy to be told otherwise. > > comments? > > > > > > -- > > Regards > > SteveF > HTML 5.1 <http://www.w3.org/html/wg/drafts/html/master/> >
Received on Tuesday, 12 May 2015 18:57:37 UTC