MathML is coming, need to consider what we need to do to support it

[WilcoFiers]

MathML is new in Chrome 97, still behind a feature flag, but that means it's coming. Chrome is the last major browser without support for MathML, so native support for it is getting pretty close.

I doubt anyone's considered what impact native MathML has to axe-core rules, so we're going to need to look into that in the near future. First priority there is to make sure we don't produce false positives. Second is to figure out what new rules we can write for it.

[danielweck]

Hello, MathML markup inside headings and table headers triggers errors. See:

daisy/ace#398

daisy/ace#410

[dbjorge]

For the higher priority "not producing false positives" part, the lowest-hanging fruit is probably just setting up a new mathAlttextText: attrText.bind(null, 'alttext') method in /lib/commons/text/native-text-methods.js and adding namingMethods: ['mathAlttextText'] to the math spec in /lib/standards/html-elms.js.

The bigger chunk of work there would be validating UA/AT compat, not the actual code change. We'd want to verify for each accessibility-supported UA/AT combination that it respects the alttext attribute on a <math> element and that its behavior works as expected consistently. I'd expect it to fit into step ARIA 1.2 section 4.3.2 step 2.E; the MathML 4 spec is a bit vague about exactly what "unspecified" vs "specified as empty" means, but I'd expect it to be treated similarly to an img alt attribute.