- From: John Foliot <john@foliot.ca>
- Date: Thu, 8 May 2014 16:50:22 -0700
- To: "'James Craig'" <jcraig@apple.com>, "'James Nurthen'" <james.nurthen@oracle.com>
- Cc: "'Steve Faulkner'" <faulkner.steve@gmail.com>, "'Gunderson, Jon R'" <jongund@illinois.edu>, <w3c-wai-pf@w3.org>, "'WAI XTech'" <wai-xtech@w3.org>
James Craig wrote: > > On May 8, 2014, at 4:06 PM, James Nurthen > <james.nurthen@oracle.com> wrote: > > > > There is a huge backwards-compatibility argument to allowing <input type="text" > > aria-required="true"> to act as a required field to the AT APIs. > > The same argument applies to <input type="text" required aria-required="true"> and > no one is opposed to allowing authors this redundancy for the sake of > backwards-compatibility. No argument. What James N. notes, and I agree with, is that there is already a fairly large corpus of content out there that exhibits <input type="XXX" aria-required="true">; and that you are now stating is not valid due to HTML5's new implied boolean of @required. You are suggesting that browsers "break" that based upon an apparent dichotomy. We cannot undo hixie's mess, but we also cannot create a new mess based upon "code purity". You stated: "We can't write a spec that has a one-way dependency on a boolean attribute. It doesn't make sense. You're effectively saying "True is true, except in some cases where it's not." " Why not? We can certainly state that under a specific condition (when applied to an <input> element) the ARIA semantic takes precedence: this is testable, scalable and while from a purity perspective upside down, it can still be made valid and conformant. I will further assert (again) that for non-experts <input type="XXX" aria-required="true"> makes perfect sense. JF
Received on Thursday, 8 May 2014 23:50:54 UTC