Frédéric Wang (:fredw) Assignee
	Description • 10 years ago

Bug 1001634 applies ATK_ROLE_UNKNOWN to all MathML elements. Until we add proper (sub)roles into ATK and AT-SPI, we can already change ATK_ROLE_UNKNOWN to something better. Here is what we discussed with Joanie:

* <mtable> - ATK_ROLE_TABLE (permanently)
* <mtr> - ATK_ROLE_TABLE_ROW (permanently)
* <mtd> - ATK_ROLE_TABLE_CELL (permanently)
* <mlabeledtr> - ATK_ROLE_TABLE_ROW (permanently)
* <mglyph> - ATK_ROLE_IMAGE (permanently)
* group-related elements (<mrow> etc): ATK_ROLE_PANEL (temporary)
* text-related: ATK_ROLE_STATIC (temporary)

ATK_ROLE_STATIC is currently only available in the unstable version of ATK, so we'll need to update the one used in Gecko.

	Frédéric Wang (:fredw) Assignee
	Comment 2 • 10 years ago

Linux builds will be available at http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/fred.wang@free.fr-d6db7b46c550

	Frédéric Wang (:fredw) Assignee
	Comment 4 • 10 years ago

This new version adds the new ATK_ROLE_STATIC and uses it for token MathML element. We're still not sure about whether ATK_ROLE_PANEL or ATK_ROLE_SECTION should be used for MathML grouping elements, so I'm just asking for feedback for now.

(this applies on top of Jonathan's patches)

	Trevor Saunders (:tbsaunde)
	Comment 5 • 10 years ago

Comment on attachment 8534331 [details] [diff] [review]
Patch V2

>   if (aAtkObj->role == ATK_ROLE_LIST_BOX && !IsAtkVersionAtLeast(2, 1))
>     aAtkObj->role = ATK_ROLE_LIST;
>   else if (aAtkObj->role == ATK_ROLE_TABLE_ROW && !IsAtkVersionAtLeast(2, 1))
>     aAtkObj->role = ATK_ROLE_LIST_ITEM;
> 
>+  if (aAtkObj->role == ATK_ROLE_STATIC && !IsAtkVersionAtLeast(2, 16))
>+    aAtkObj->role = ATK_ROLE_TEXT;

else if

> ROLE(MATHML_MATH,
>      "math",
>-     ATK_ROLE_UNKNOWN,
>+     ATK_ROLE_MATH,

you need a runtime check that role is available.

> ROLE(MATHML_SPACE,
>      "mathml space",
>      ATK_ROLE_UNKNOWN,

I think there's an argument to be made for this being a text role though really a accessible just for whitespace is kind of strange.

	Frédéric Wang (:fredw) Assignee
	Comment 6 • 10 years ago

Thanks for the feedback Trevor. Indeed, I had forgotten to remove mspace in bug 1001634.

	alexander :surkov (:asurkov)
	Comment 7 • 10 years ago

Comment on attachment 8534902 [details] [diff] [review]
Patch V3

Review of attachment 8534902 [details] [diff] [review]:
-----------------------------------------------------------------

::: accessible/base/RoleMap.h
@@ +1121,5 @@
>       eNoNameRule)
>  
>  ROLE(MATHML_FRACTION,
>       "mathml fraction",
>       ATK_ROLE_UNKNOWN,

do we wait for new role for this?

	Frédéric Wang (:fredw) Assignee
	Comment 8 • 10 years ago

(In reply to alexander :surkov from comment #7)
> do we wait for new role for this?

The idea was just to use the ATK roles that already exist, like in the corresponding WebKit bug I'm adding to the "see also" list. We started to talk about more (sub)roles with Joanie but that will be done in follow-up bugs.

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

Unbitrot

	Frédéric Wang (:fredw) Assignee
	Comment 10 • 9 years ago

Unbitrot and make it apply before bug 1001641.

	Frédéric Wang (:fredw) Assignee
	Comment 11 • 9 years ago

So the ATK patches can be taken before Jonathan's other patches.

	alexander :surkov (:asurkov)
	Comment 12 • 9 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/fb6c18e47df5

	Carsten Book [:Tomcat]
	Comment 13 • 9 years ago

https://hg.mozilla.org/mozilla-central/rev/fb6c18e47df5