`svg` element in HTML-AAM, SVG-AAM and WAI-ARIA Graphics Module specs
This is an intended duplicated issue already raised in SVG-AAM repo 4 years ago, but I opened it to gain traction on this topic since it affects quite heavily tool vendors.
Current specs:
- ARIA in HTML maps the
<svg>element to role="graphics-document" - SVG AAM set the default platform WAI-ARIA role mappings to "graphics-document"
- WAI-ARIA Graphics Module requires role="graphics-document" to have an acc name
Concerns:
- Currently, different AT don't expose the
svgelement to the user unless it has an acc name, ignoring the browsers computed role - Currently, browsers are exposing
svgelements differently (see last WPT test) - Web authors expect
svgnot to be exposed by default; if authors have to follow current specs, they are forced to either set an accessible name, or role=none to eachsvgelement, even if thesvgdoesn't convey any information.
Proposal:
- expose
svgwith computed role=none if no acc name (similarly as img element with empty alt) - otherwise graphics-document
This is an intended duplicated issue already raised in SVG-AAM repo 4 years ago,
Which one? I see https://github.com/w3c/svg-aam/issues/38 but that's much younger.
FWIW I would remove the label "duplicate" since this is more like a meta issue. Then we can collect the various issues that are related.
Ah, maybe https://github.com/w3c/svg-aam/issues/12 ?
Yes, sorry, it was meant to be a placeholder to be replaced with the link, but I forgot to do it. Thanks @pkra
Should we deal with this just like we do section and role=region?
Discussed in today's meeting: https://www.w3.org/2025/02/13-aria-minutes.html#465f
After talking with @pkra, added deep-dive label to setup a deep dive during our next meeting this Thursday.
Schedule deep dive in mid/late March
Tentatively scheduling for March 20th.
Also to discuss these WPT test results...
and these remaining interop-accessibility task issues
@giacomo-petri or @cookiecrook or @jnurthen -- can someone summarize the conclusions of the deep dive? Thanks! :)
According to the AI summary of the meeting, the next steps are:
- Giacomo to create a pull request for the proposal to expose Svg as graphics document only when it has an accessible name, otherwise expose as generic or ignored.
- Giacomo to develop additional Web Platform Tests (WPT) for complex SVG cases not currently covered in existing tests.
- James Craig to investigate how changing the role of SVG elements might impact browser-based computer vision features for accessibility.
- James Craig to review and potentially update existing WPT tests for SVG accessibility to ensure they cover all necessary scenarios.
- Giacomo to compile a list of real-world examples demonstrating the impact of SVG accessibility issues on user experience, particularly in e-commerce scenarios.
Hello!
I'm quite busy these weeks, so I'm not sure when I'll be able to tackle points 1 and 2. In the meantime, I'd like to address point 5 by sharing some screenshots that highlight how svgs used as icons impact people (I'm currently using VO on Chrome):
This is quite common, you could pick a random website and likely find the same issue.