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

New roles for MathML fraction, roots and scripts landed last week in WebKit/ATK/AT-SPI:

http://trac.webkit.org/changeset/179253
https://git.gnome.org/browse/atk/commit/?id=c7195e78c
https://git.gnome.org/browse/pyatspi2/commit/?id=c08fdaa099b8b0410d4dabf36325897bea73ba27

This bug is for adding them to Gecko.

	Frédéric Wang (:fredw) Assignee
	Comment 1 • 9 years ago

+ Expose mfrac as ATK_ROLE_MATH_FRACTION, and both mroot and msqrt as
+ ATK_ROLE_MATH_ROOT. In the case of script elements, expose the container
+ (msub, msup, msubsup, and mmultiscripts) using the generic text block role
+ ATK_ROLE_SECTION. Then expose the actual script child elements using role
+ ATK_ROLE_SUBSCRIPT/ATK_ROLE_SUPERSCRIPT. In the case of mmultiscripts,
+ expose whether it is a prescript or postscript via AtkObject attribute.

Joanie: can you explain what is the plan here? Is it another interim solution until we have math subroles? (I don't remember whether we discussed that and that was not in your notes).

	Joanmarie Diggs
	Comment 2 • 9 years ago

It became sadly apparent that the subroles would not happen in the immediate future. Therefore, this is indeed an interim solution. Or part of it, anyway.

	Frédéric Wang (:fredw) Assignee
	Comment 3 • 9 years ago

I'm removing the "scripts" part here as I don't think ATK_ROLE_SUBSCRIPT/ATK_ROLE_SUPERSCRIPT will work for things like

<msup><mn>1</mn><mfrac><mn>2</mn><mn>3</mn></mfrac></msup>

Let's just handle fractions and roots for now.

	alexander :surkov (:asurkov)
	Comment 6 • 9 years ago

Frederic, is it ready to land?

	Frédéric Wang (:fredw) Assignee
	Comment 7 • 9 years ago

Comment on attachment 8575479 [details] [diff] [review]
Patch

OK, this and bug 1109022 can be taken now that bug 1001634 is merged. I checked that they apply and build correctly. However, I have some problem with accerciser right now, so could not check whether the ATK tree is correctly exposed...

	alexander :surkov (:asurkov)
	Comment 8 • 9 years ago

Comment on attachment 8575479 [details] [diff] [review]
Patch

Review of attachment 8575479 [details] [diff] [review]:
-----------------------------------------------------------------

::: accessible/atk/AccessibleWrap.cpp
@@ +712,5 @@
>    else if (aAtkObj->role == ATK_ROLE_STATIC && !IsAtkVersionAtLeast(2, 16))
>      aAtkObj->role = ATK_ROLE_TEXT;
> +  else if ((aAtkObj->role == ATK_ROLE_MATH_FRACTION ||
> +            aAtkObj->role == ATK_ROLE_MATH_ROOT) && !IsAtkVersionAtLeast(2, 16))
> +    aAtkObj->role = ATK_ROLE_UNKNOWN;

some generic role would be nicer but perhaps it doesn't really matter since no semantics anyway

	Frédéric Wang (:fredw) Assignee
	Comment 9 • 9 years ago

(In reply to alexander :surkov from comment #8)
> ::: accessible/atk/AccessibleWrap.cpp
> @@ +712,5 @@
> >    else if (aAtkObj->role == ATK_ROLE_STATIC && !IsAtkVersionAtLeast(2, 16))
> >      aAtkObj->role = ATK_ROLE_TEXT;
> > +  else if ((aAtkObj->role == ATK_ROLE_MATH_FRACTION ||
> > +            aAtkObj->role == ATK_ROLE_MATH_ROOT) && !IsAtkVersionAtLeast(2, 16))
> > +    aAtkObj->role = ATK_ROLE_UNKNOWN;
> 
> some generic role would be nicer but perhaps it doesn't really matter since
> no semantics anyway

Right. WebKit fallbacks to ATK_ROLE_UNKNOWN, so let's do that too.

http://trac.webkit.org/browser/trunk/Source/WebCore/accessibility/atk/WebKitAccessibleWrapperAtk.cpp#L677

	alexander :surkov (:asurkov)
	Comment 10 • 9 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/1b0004f24d3f

	Ryan VanderMeulen [:RyanVM]
	Comment 11 • 9 years ago

https://hg.mozilla.org/mozilla-central/rev/1b0004f24d3f