- From: David Bolter <david.bolter@utoronto.ca>
- Date: Thu, 05 Mar 2009 15:08:22 -0500
- To: Joseph Scheuhammer <clown@utoronto.ca>
- CC: "wai-xtech@w3.org" <wai-xtech@w3.org>
Hi Joseph. Great questions, some bits inline: Joseph Scheuhammer wrote: > David, > > Some pros and cons. > > Pro: if aria does become ubiquitous in the sense that it is used for > markup other than HTML, then it would be good to have a common way of > detecting changes to aria attributes. Thus, regardless of its use in > SVG, XUL, or whatever, there is a standard way to react to changes to > aria information. > > Con: it has taken a long time for user agents to support DOM mutation > events. I despair of a new event coming along anytime soon. I guess > it depends on the will to do it -- get enough people behind it and it > will happen. Well, in the case of FF, I'm sure we're willing to implement it if it makes the web more accessible. > > Con: and a very specific one. What about "native" widgets such as > <input type=checkbox>? They typically do not need aria information. > Specifically, aria-checked isn't needed for checkbox inputs. > Furthermore, there is a reason not to include aria attributes in these > cases because it leads to a synchronization issue. That is, given > <input type=checkbox checked="checked" aria-checked="true">, which of > @checked vs. @aria-checked do you trust if they are not in agreement? > But, if aria-checked is left out there won't be an AriaAttrModified > event when the checked state changes. So: how do you deal with > native widget state changes? Good catch. I'm not sure this is the best case though, since we already have a working system for native markup. FF exposes and performs a click action to/for AT. I personally don't like aria-checked applied to a native checkbox. Are people doing this now in the wild? (I'm asking everyone) > > Miscellaneous: maybe it doesn't need to be a mutation event. There > are other "high level" events in the DOM3 event model, such as > UIEvent. Since a lot of aria deals with the user interface, perhaps > that branch of the event model should be extended. That is, perhaps > aria events are a fomr of UIEvents. (This obviously needs more thought). > Interesting... please think more :) cheers, David
Received on Thursday, 5 March 2009 20:09:02 UTC