- From: Devarshi Pant <devarshipant@gmail.com>
- Date: Tue, 3 Mar 2015 17:11:50 -0500
- To: Sailesh Panchang <sailesh.panchang@deque.com>
- Cc: WAI Interest Group <w3c-wai-ig@w3.org>
- Message-ID: <CAJGQbjtGCMfzcFBaQgF-L_f3_0FLs=PwFJzWo4kqYXwfT0q9DQ@mail.gmail.com>
Sailesh, Thanks for sharing the link. You are correct, aria-controls is not required for something like this. Interesting to note that role = "grid" throws jaws into the pc cursor mode. aria-label and aria-sort do however get updated upon activating the column headers. thanks, Devarshi On Tue, Mar 3, 2015 at 11:41 AM, Sailesh Panchang < sailesh.panchang@deque.com> wrote: > Devarshi, > > Is aria-controls really useful here? > The JS needs to be fixed for the aria-label and aria-sort values. > Besides, the focus indicator is not visible; with CSS off, the up/down > arrows disappear. HTML images are better for the arrows I think and > can be assigned alt. > I realize this is an example of static headers with sort feature. So > consider adapting from the technique here > http://juicystudio.com/article/accessible_data_tables_static_headers.php > Exposing 2 tables to SR users and using a summary to explain the > disconnect is a no-no for me. > Thanks, > Sailesh > > > On 3/3/15, Devarshi Pant <devarshipant@gmail.com> wrote: > > Sailesh, > > The second grid seems to associate the headers in the first grid, and > that > > could be due to the use of aria-controls - is that a correct observation? > > Some questions - > >>It's not clear whether this is an example of incorrect aria use, or maybe > > aria overuse. Assuming aria (code) is correctly used, and screen reader > is > > incorrectly interpreting the code, is it a violation / failure? > >> Visually it's a single data table, not structurally. Would it be okay to > > suggest in the table summary of the first grid something like, "There are > > two tables; this is the first table that hosts header cells. Table data > > follows in the second table." Should it then be considered navigable by > > screen reader users? > >>Is it reasonable to code to specs and not worry about how assistive > > technologies would react (in the interim, determine workarounds), knowing > > that someday these will eventually merge? > > -Devarshi > > > > > > On Mon, Mar 2, 2015 at 4:49 PM, Sailesh Panchang > > <sailesh.panchang@deque.com > >> wrote: > > > >> Devarshi, > >> I am testing with JAWS 16, NVDA 2015.1: > >> > >> There are two grids and column header row is repeated in both; the > >> first has no data rows and the 2nd grid has an empty first row. > >> This is confusing in itself. > >> Then: > >> On loading the page, on grid#1: > >> aria-controls=example > >> aria-label=Name: activate to sort column descending > >> aria-sort=ascending > >> Then I sorted the table using age: > >> aria-controls=example > >> aria-label=Name: activate to sort column ascending > >> aria-sort=ascending > >> In grid#2:col headers are in row#2 and I see > >> aria-controls=example > >> aria-label=Age: activate to sort column ascending > >> aria-sort=descending > >> > >> An example of incorrect ARIA use? > >> Thanks, > >> Sailesh > >> > >> > >> On 3/2/15, Devarshi Pant <devarshipant@gmail.com> wrote: > >> > HI All, > >> > Can someone test with a screen reader and determine if it announces > the > >> > sort state of the active column header: > >> > http://www.datatables.net/examples/basic_init/scroll_y.html# > >> > > >> > I get the following results: > >> >>JAWS 13 and IE9 convey that headers are sortable but do not announce > >> >> the > >> > states (ascending / descending) once invoked. Note that the active > >> > state gets announced on tabbing and then shift tabbing back. > >> >>NVDA does not announce the headers. > >> > Would be nice to see additional test results. > >> > Thanks, > >> > Devarshi > >> > > >> > > >
Received on Tuesday, 3 March 2015 22:12:37 UTC