Here are the landmark roles: * application * banner * complementary * contentinfo * main * navigation * search We need full coverage here. Things like "application" have interesting mappings like ROLE_APPLICATION, which might actually become hairy monsters since some AT probably still don't expect applications embedded in content. Spin off bugs as necessary. Roles like "banner" should default to ROLE_NOTHING

(In reply to comment #0) > Roles like "banner" should default to ROLE_NOTHING I am not so sure this is the right approach. If one puts a role of "banner" on an h1 element, for example, which I've seen happen, this will cause NVDA and others to no longer expose the heading info properly. I would think that these landmark roles should simply fall back to the native HTML markup and simply expose the fact that they're a landmark through object attrs or whatever.

(In reply to comment #1) > (In reply to comment #0) > > Roles like "banner" should default to ROLE_NOTHING > > I am not so sure this is the right approach. If one puts a role of "banner" on > an h1 element, for example, which I've seen happen, this will cause NVDA and > others to no longer expose the heading info properly. I would think that these > landmark roles should simply fall back to the native HTML markup and simply > expose the fact that they're a landmark through object attrs or whatever. Agree. If "banner" is landmark then it shouldn't override native role, iirc this is intendent behaviour of landmarks.

(In reply to comment #2) > (In reply to comment #1) > > (In reply to comment #0) > > > Roles like "banner" should default to ROLE_NOTHING > > > > I am not so sure this is the right approach. If one puts a role of "banner" on > > an h1 element, for example, which I've seen happen, this will cause NVDA and > > others to no longer expose the heading info properly. I would think that these > > landmark roles should simply fall back to the native HTML markup and simply > > expose the fact that they're a landmark through object attrs or whatever. > > Agree. If "banner" is landmark then it shouldn't override native role, iirc > this is intendent behaviour of landmarks. Check your first point on bug 481114 comment #18 :)

(In reply to comment #3) > > Check your first point on bug 481114 comment #18 :) Hey David there is no bug 481114 comment #18 there. But I really not sure role nothing is more useful than native one.

OK let's see what lands on bug 481114 first. I agree with you guys we don't want to blow away a useful native role with ROLE_NOTHING.

Fixed via changeset: http://hg.mozilla.org/mozilla-central/rev/f2f7d5625759