Re: Aria-labelledby for form controls- examples

On 9/11/2013 7:52 AM, Sailesh Panchang wrote:
>       On 9/10/2013
>   7:54 AM, Sailesh Panchang
>         wrote:
> Please review:
>   Aria-labelledby for form controls
>       JN: I believe this is no different from
>       Is there anything I am missing that the example in the
>   current      technique doesn't have?
>   Sailesh 09/11:
> The survey page for 'aria-labelledby for controls' had no techniques. It was marked 'TDB'.
> That's why I sent these examples.
> If there  were examples already available, they should have been on that survey page, no?

JN 9/11:
This title of this technique is misleading and needs fixing. This is an 
aria-labelledby technique intended at 1.1.1
As such labels for form fields are not appropriate. We have other 
techniques for labeling form fields.

>       JN: We certainly need a role=group technique to cover
>   this use case.
>       Would someone like to volunteer to write this?
> Sailesh 09/11:
> There's one already  submitted in May 2013.
> I had submitted a few techniques including one for the role=group as an alternative for fieldset-legend
> There's a problem with the technique submission form which I had conveyed to Andrew and code had not got submitted alright.
> So a Word doc was sent to Andrew with my recent submissions / comments  a few weeks ago. See attached.
> A few times, I requested that these be reviewed.
JN 9/11: I haven't seen this and it certainly hasn't been passed to the 
task force for consideration. When I get time I will try to add this to 
the wiki so it can be worked a little more by the TF (or you can do this 
yourself if you want)

>       JN: I think this example needs modification and then
>   could fit      nicely into the Using        aria-labelledby to concatencate multiple text nodes
>   case.
>   As it stands there is no indication to AT that checking
>   the checkbox      does a compare operation so the labelledby needs to
>   point to the header and the checkbox label and not just the header.
> Sailesh 09/11:
> Well the aria-labelledby works fine for the checkbox and conveys its purpose clearly.
JN 9/11: The label on its own could be to buy the item or indeed any 
other usage. The label needs to include this. I do not think the 
functionality of the checkbox is clear without this.
> There is also an explicit label 'check to compare' which is read when one arrows down (out of forms mode).
JN 9/11: I'm not sure we can rely on this - of we could our lives would 
be much easier. How about a compromise where the check to compare is the 
aria-descibedby reference?
> Certainly instruction that you can compare 2 or 4 items will be conveyed somewhere at the start / end of the form.
JN 9/11: If this is the case then this should be in the example.
> That does not  have to be conveyed by every checkbox.
> Based on your argument, every form control's label needs to convey which form it belongs to and what will be accomplished by submitting the form.
JN 9/11: this is not an equivalent argument. The visually associated 
label for the checkbox is "Check to compare". Not providing this to AT 
is a clear violation of 1.3.1.
> i.e. on a login form the label for username will need to convey that it will be submitted as part of the login process.
> If every checkbox were to convey that once checked it will be used for comparison,  that will be a lot of verbosity and a lot of problem.
> When you have a list of emails displayed  on a Web page, one can check them and delete or move or mark as spam etc. The checkbox's label does not convey this.
JN 9/11: Here there is not a visually associated label. If it is 
important for a sighted user to have the "check to compare" text then 
this should be provided for a screen reader user too. If this is not 
important then the text need not be provided to anyone.


> Thanks and regards,
> Sailesh
>       Regards,
>       James
>   Secondly in the matter of aria-labelledby for non text
>   content:
>   - The first example has no alt
>   - The second one: seldom  is there text taht says '4 out
>   of 5' for *rating it is only the image.
>   But, yes, the  example is a good one if it is indeed marked
>   up that way with multiple star images and text alongside.
>   About use of role=headings:
>   It appears it is more of a remediation technique to be used
>   only when HTML h<n> cannot be used like for h7 as
>   illustrated?
>   Need to think of more used cases I guess.
>   Thanks,
>   Sailesh

Received on Wednesday, 11 September 2013 17:17:44 UTC