Re: Effect of role=presentation on img elements with svg

(Moving discussion back to PFWG)

On Mar 11, 2014, at 3:42 PM, Richard Schwerdtfeger <schwer@us.ibm.com> wrote:

> I am simply saying that we do not have an implementation guide for SVG - yet. Nothing more. We don't have a spec. that anyone has committed to or is publicly viewable. When we kick off the implementation guide work we will know more.
> 
> That said, what I have written in the SVG2 specification for native host language semantics is consistent with a role of image for an SVG image:
> 
> https://svgwg.org/svg2-draft/struct.html#implicit-aria-semantics
> 
> Most of the other elements, when the situation is right, will default to a role of group unless no accessibility when there is enough content/attributes that would warrant producing a group role.

This is incorrect, too. Most other elements (including svg:g, html:div, etc) default to some “generic” or rendering engine internal role that ARIA has yet to map. It happens to be exposed in many places on OS X as the AXGroup (AX API Role) but that platform role is not the same as or a 1:1 mapping with the ARIA “group” role.

> I am looking at expanding on the group role determination to include other ARIA global attribute references. We do not want to create an accessible object for every SVG element unless it is warranted - as SVG elements are expensive and it adds bloat. 

I think what you’re wanting instead of “group” is something like a generic container. 

ISSUE-638: Generic container roles for things like div/span…
https://www.w3.org/WAI/PF/Group/track/issues/638

This is tangentially related to:

ISSUE-348: Due to author confusion, consider renaming role="presentation"
https://www.w3.org/WAI/PF/Group/track/issues/348


> I have been synching up vocabulary/mappings with Steve Faulkner in HTML. 

It’s a problematic to map all HTML tags to ARIA 1.0 roles, since there aren’t currently enough ARIA roles to cover all of the semantics in HTML. For example, HTML 5 has various new types of inputs that aren’t available in ARIA, so those should not be mapped to existing ARIA roles… The mapping should remain as an implementation detail until an appropriate ARIA role is added. Unfortunately we already punted some of this work to ARIA 2.0, such as:

ISSUE-409: ARIA 2.0 should consider providing roles that match some or all of the HTML 5 new input types (e.g., color picker, date picker, etc.)
https://www.w3.org/WAI/PF/Group/track/issues/409

James

> Rich
> 
> 
> Rich Schwerdtfeger
> 
> James Craig ---02/03/2014 01:24:24 PM---On Feb 2, 2014, at 8:22 AM, Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no> wrote: > The claim
> 
> From:	James Craig <jcraig@apple.com>
> To:	Richard Schwerdtfeger/Austin/IBM@IBMUS
> Cc:	Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>, Steve Faulkner <faulkner.steve@gmail.com>, Joanmarie Diggs <jdiggs@igalia.com>, Cynthia Shelly <cyns@microsoft.com>, Bryan Garaventa <bryan.garaventa@whatsock.com>, "T.V Raman" <raman@google.com>, "jongund@illinois.edu" <jongund@illinois.edu>, "jason@jasonjgw.net" <jason@jasonjgw.net>, "wai-xtech@w3.org" <wai-xtech@w3.org>, "w3c-wai-pf@w3.org WAI-PFWG" <w3c-wai-pf@w3.org>
> Date:	02/03/2014 01:24 PM
> Subject:	Re: Effect of role=presentation on img elements with svg
> 
> 
> 
> On Feb 2, 2014, at 8:22 AM, Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no> wrote:
> 
> > The claim, at that page, is that the <img> defaults to img role. 
> > However, it actually is given group role [in Safari on OS X].
> 
> 
> On Feb 2, 2014, at 12:01 PM, Richard Schwerdtfeger <schwer@us.ibm.com> wrote:
> 
> > While I understand what Apple has implemented (defaults to a group role) we have not decided on a normative mapping for SVG across browsers.  
> 
> 
> Leif and Rich, your assumptions are incorrect about how these roles are treated and what is mapped. 
> 
> I think you’re conflating the platform role mapping with the ARIA mapping. The computed ARIA role for <img> is “img” regardless if it is an SVG or raster image. The internal WebKit mapping for <img> is “ImageRole”, again regardless of the type. However, the is output differently to different platforms. In the case of accessible SVG images on OS X, the image has a rendered subtree, and the current version of AXImage (the AX API mapping) does not support descendant contents, so it’s mapped, on that platform, to AXGroup (a platform-specific role that is not the same as the ARIA “group” role).
> 
> The content is rendered and the subtree is not hidden, so it should be accessible to all.
> 
> James
> 
> 
>