Re: [math-on-web] a11y TF meeting minutes, 2017/04/10

Hi Daniel,

> Volker says that “some people are opposed to aria-labels”. T

IIRC, Volker was referrring to the deep-labels approach. There seems to be
a significant group in the a11y world (especially among AT developers) that
argue against the use of aria-labels (in favor the development of semantic
notation via roles and eventually tags).

> I do not understand the roledescription solution. Could someone rephrase
it?

This is the spec: https://www.w3.org/TR/wai-aria-1.1/#aria-roledescription

It allows authors to customize the roles representation within AT (if AT
supports it). Hence my comment that the argument that role="math" might be
misleading (for STE content) could be addressed by suggesting to use
role=math + aria-roledescription="physics".

Best,
Peter.

On Thu, Apr 13, 2017 at 11:18 AM, Daniel Marques <dani@wiris.com> wrote:

> Hi everybody,
>
> After reading the minutes, let me comment on some of the subjects.
>
> Volker says that “some people are opposed to aria-labels”. This is
> consistent with the fact that AT reads the text and ignores the
> aria-labels. It is also connected with the idea that some people don’t like
> accessible specific content.
>
> The role=math has its faults but does some of the job. I guess that it was
> a compromise reached time ago in order to use the aria-label (which seems
> to confirm what Volker said). If deprecated, an alternative should be
> provided. Maybe another role?
>
> I do not understand the roledescription solution. Could someone rephrase
> it?
>
> Dani
>
> On 12 Apr 2017, at 18:58, Peter Krautzberger <peter.krautzberger@mathjax.
> org> wrote:
>
> Hi everyone,
>
> below are the minutes from the a11y task force meeting this week.
>
> Best,
> Peter.
>
> # math on web CG -- a11y TF
>
> * Updates.
>   * Peter: continued work on deep labels
>     * more examples
>     * more focused on SVG but should
>     * more recordings by Thursday, expect to confirm the validity
>     * one problem: tables mathJax svg vs commonthml
>       * ARIA 1.1 table-related roles should help
>   * Peter: also got an update from Jason Meryll
>     * where he described their work
>     * ACTION: ask Jason to make it public
>       * then reference on website
>   * Volker: from last CG meeting: hacked around with speech API
>     * prosidy support via SSML in speech synthesis API
>     * several tests
>     * no SSML support
>     * though VoiceOer had some specialized markup (see minutes)
>     * working on next few examples
> * Peter: role=math
>   * not much feedback from CG outside of a11y TF - which may have been
> obvious
>   * Volker: at w4a a11y hackathon, nobody even knew the role existed
>   * Peter: fun fact: James Craig mentioned that the role was derived from
> image role (which makes sense given the time)
>   * Peter: from MathJax end, we could delay until we get to working on
> turning deep labels into something for production
>   * Peter: given Daniel's comments from last TF meeting, maybe the groups
> recommendation should be to deprecate role=math.
>   * ACTION: ask CG for a recommendation
>     * point out that there are features that would help but that role-math
> serves no purpose
> * Volker: some people are opposed to aria-labels
> * Peter: new roles in ARIA 1.1
>   * e.g., table related roles
>   * e.g., roledescription
>     * Kevin: could be used to specify role='Math'
>       * Peter: right, argument to say that role is no longer necessary
>         * but also reversely: role='math' with roledescription='physisc'
> could argue "against" Daniel's concern.
> * Volker: will post some examples
>
>
>