- From: <bugzilla@jessica.w3.org>
- Date: Fri, 27 Sep 2013 17:26:02 +0000
- To: public-html-a11y@w3.org
https://www.w3.org/Bugs/Public/show_bug.cgi?id=23377 --- Comment #4 from James Craig <jcraig@apple.com> --- (In reply to steve faulkner from comment #2) > I think there are legitimate use cases for allowing aria-required on > elements without the required attribute. > > It is common practice to have a a control that is marked as required in text > for example using an asterisk: > > <label>name * <input></label> > > It is not a requirement in HTML5 that this pattern must use the required > attribute to mark the control as required, and there are reasons why an > author may not want to use the required attribute such as they do not want > the associated UI and behaviour that is implemented by some user agents for > the required attribute. Thus in the example above providing the author with > the means to convey the required state to acc APIs via aria-required will > improve the accessibility of the pattern. I could accept the behavior you want, but as it's written in the spec, it conflicts with itself because @required is a Boolean attribute. You need to make either one of these changes: 1. Either define the "false" state of @required in the Strong Native Semantics table, where you currently have only the "true" state. (This would effectively mean @aria-required had no effect on form elements that accept @required.) 2. Move all the mentioned of @required and @aria-required from the Strong Native Semantics table to the implicit ARIA semantics. (This would allow the behavior you appear to desire and have outlined above.) -- You are receiving this mail because: You are on the CC list for the bug.
Received on Friday, 27 September 2013 17:26:07 UTC