Frédéric Wang (:fredw) Reporter
	Description • 9 years ago

WebKit exposes the <mmultiscripts> element with subrole "AXMathMultiscript" and two attributes "AXMathPrescripts" and "AXMathPostscripts". We already expose the subrole, however I'm not sure we can expose the attributes without creating some kinds of "fake" accessibles (WebKit's accessible tree seem to be very tied to WebKit's MathML implementation with anonymous flexbox renderers).

	Joanmarie Diggs
	Comment 1 • 9 years ago

Frédéric, I'm not seeing a bug for ATK support for mmultiscripts. Even with a tag hack, etc., Orca needs some reasonable way to look at a given element and figure out if it's a presubscript, postsubscript, presuperscript, or postsuperscript. I'd like to be able to do that without having to look for the prescripts accessible (which I would argue should not be exposed) and then having to look at child order (taking that prescripts accessible into account).

	Frédéric Wang (:fredw) Reporter
	Comment 2 • 9 years ago

(In reply to Joanmarie Diggs from comment #1)
> Frédéric, I'm not seeing a bug for ATK support for mmultiscripts. Even with
> a tag hack, etc., Orca needs some reasonable way to look at a given element
> and figure out if it's a presubscript, postsubscript, presuperscript, or
> postsuperscript. I'd like to be able to do that without having to look for
> the prescripts accessible (which I would argue should not be exposed) and
> then having to look at child order (taking that prescripts accessible into
> account).

Some work has been done for Firefox OS / Android in bug 1177765. Essentially, children will have some xml-roles "presubscript", "subscript", "presuperscript" and "superscript". You can use the following try build here:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=8cf89f20b768

Note that the same xml-roles will be used for <msub>, <msup>, <msubsup> so I believe this will avoid to have clash between roles in these cases (the famous case of fraction as script) and avoid to introduce fake accessibles.

I know that you do not like object attributes and Jamie also mentioned that xml-roles is a bit a hack. As discussed with Alex, we may eventually want to implement that as relations at the end. I'll reply to Jamie's mail and give more details there.