So, I think adding roles to the spec. is a good idea for HTML native
elements - especially the new ones. If we have roles that are going to shop
up in desktop and web apps I think it is a good idea to create roles for
these and it does reduce the amount of parsing needed. However, At some
point I think relying on role constants in the API is a problem as you have
to redo the IDL all the time.

I absolutley believe that exposing object attributes should be standardized
in their implementation. Those that are used today should be documented in
the IA2 spec.

I don't know that I would create anything special for landmarks because ATs
do expand what ARIA uses for landmarks. While I don't agree with that this
is something they believe their users would prefer. So, the net is, even if
we define the set of landmarks we won't be able to expose all of what ATs
expose as landmarks unless we do something like allow the AT to register
what they use for landmarks. Then we could do something like have a method
of isLandmark() or getLandmarksObjects();

Rich Schwerdtfeger
CTO Accessibility Software Group



From:   David Bolter <[email protected]>
To:     [email protected]
Date:   11/24/2010 10:28 AM
Subject:        [Accessibility-ia2] about object attributes
Sent by:        [email protected]



Hi again,

I want to raise this issue in parallel to the IA2_ROLE_NOTE thread. I
don't want to complicate that discussion since this one might take fair
bit longer to resolve.

One of the IA2 object attribute pairs that FF exposes is the aria roles
string. So for example:
<div role="navigation">
Will have the object attribute xml-roles=navigation. (I apologize for
the "xml-" prefix, but that prefix is there for purely historical
reasons as I understand it and is tangential to this discussion.)

Now, one of the things we do for html landmarks like <nav> is expose it
like we would the aria role=navigation, via an xml-roles object
attribute. This allows AT that already work with our xml-roles (based on
aria) to "just work" with <nav>.

A couple of questions:
1. Should we strive to add roles to the IA2 spec such that we don't need
to expose the role as an object attribute?
2. Should we promote the object attributes method as a first class
solution and standardize how they are implemented?
3. Are landmark roles distinct enough to warrant a distinct way of
exposing them?

Hoping to hear thoughts from everyone, including AT devs.

Cheers,
David
_______________________________________________
Accessibility-ia2 mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2

<<inline: graycol.gif>>

_______________________________________________
Accessibility-ia2 mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2

Reply via email to