369082 - expose IA2 localizeXXX methods

Status: RESOLVED INCOMPLETE

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

Methods are defined at http://accessibility.freestandards.org/a11yspecs/ia2/docs/html/

Comment 1

[alexander :surkov (:asurkov)]

Does accessible module has a place where .properties files should be places? If it hasn't then should accessible have own package?

Comment 2

[Aaron Leventhal]

We have these:
http://lxr.mozilla.org/seamonkey/find?string=accessible.properties

But the truth is that most things should not be localized -- even our actions. Otherwise when AT goes to look for a specific action, such as "toggle" to find out  a tree cell checkable, then it can't find it.

Most of the localization stuff is in there for the future, when web authors will be able to define their own roles.

Comment 3

[alexander :surkov (:asurkov)]

So, should be localized values of methods localizedRoleName, localizedStateNames, localizedExtendedRole, localizedExtendedStates?

Comment 4

[Aaron Leventhal]

I don't think we'll need those for now.

In the future web authors may want to create new kinds of accessibility roles, which are basically classes. The roles will need to be able to support various properties, relations, actions, etc. All of those things will have to have author-supplied localized names.

For the built-in semantics the AT is responsible for the localization. If the API makes it look different it's a problem with the API, and we should report it.

Comment 5

[alexander :surkov (:asurkov)]

I just like to have bug 347791 fixed. For this I need to get string representation  of role, states.

Comment 6

[Aaron Leventhal]

That's different. See my last comment in that bug. We should be exposing the programmatic constant, since the information is consumed by other programmers.

Comment 7

[alexander :surkov (:asurkov)]

I thought that AT use rather constants than their localized names, for example, they would use nsIAccessible::role than localizedRoleName. Wouldn't it?

What should localizedRoleName return?

Comment 8

[Aaron Leventhal]

We won't need localizedRoleName for Firefox 3. I think returning E_NOIMPL would be correct for now.

In the future this will allow web authors to define new widgets. We'll also need to find a way to expose the inheritance hierarchy for the widget they define, as well as whatever custom properties/interfaces etc. it has.

Comment 9

[Aaron Leventhal]

> In the future this will allow web authors to define new widgets.

Let me clarify. In the future an author could perhaps define a new role somewhere, call it "myroles:calendar". It might inherit from "grid". They'll need to expose the localized version of "calendar" somewhere, so that the screen reader knows how to tell the user what it is, even if it hasn't ever seen one before.

Comment 10

[Aaron Leventhal]

Same thing with states/relations. That's for authors defining new properties, which isn't possible with ARIA yet.

Comment 11

[alexander :surkov (:asurkov)]

(In reply to comment #10)
> Same thing with states/relations. That's for authors defining new properties,
> which isn't possible with ARIA yet.
> 

Does it mean we won't provide localized versions of reserved relations/actions? And all this methods are going to be used only when ARIA gets an ability to specify this?

Comment 12

[Aaron Leventhal]

Yes. At least for now we don't need this.

Comment 13

[Axel Hecht [:Pike]]

This sounds like accessible.properties should not be looked into and just be cvs rm'ed? Which would make bug 374605 WORKSFORME or something.

Comment 14

[David Bolter [:davidb] (NeedInfo me for attention)]

AUTO-CLOSED. This bug untouched for over 2000 days. Please reopen if you can confirm the bug and help it progress.