Joanmarie Diggs Reporter
	Description • 9 years ago

Currently the sub and sup elements are not exposed as separate accessible objects; instead, the position is exposed as a text attribute associated with the parent element. (See bug 735645.)

That works. However, I'm wondering if it would make sense to expose sub and sup as accessible children of the parent element. The reason why is that in normal navigation (by line, by paragraph, etc.), the user is generally not interested in text attributes. So, at least in the case of Orca, text attributes are not asked for or examined unless the user specifically requests them. Orca does, however, piece together the content in that text (child links, child "clickables", child form fields, etc.) by looking at the accessible tree. Thus if sub/sup elements were exposed as accessible objects, Orca would present them as just another child. If they remain text attributes, presenting them will require Orca to check text attributes on the off chance there's a sub or sup element. Furthermore, due to limitations in ATK (and IA2??), that might mean multiple checks because one asks for the attributes at a given offset and gets a range back. If the end of the range is not the end of the text to be presented, subsequent call(s) are required.

Ccing James Teh for his thoughts.

	James Teh [:Jamie]
	Comment 1 • 9 years ago

While I do follow your reasoning here, I do think it makes sense that these are text attributes. Also, NVDA already depends on this behaviour. NVDA actually does consume text attributes in normal navigation, as the user can configure automatic reporting of requested formatting changes for both speech and braille.

So, for IA2, I'd prefer the text attributes weren't removed. That leaves two options:
1. Expose this differently for IA2 and ATK. I suspect Mozilla won't be willing to do this. :)
2. Do expose sup and sub as separate accessible objects, but keep the text attributes as well. That should keep both IA2 and ATK happy. Note that this will result in extra embedded objects (which we all know and love), which means a slight perf hit for IA2.

	Joanmarie Diggs Reporter
	Comment 2 • 9 years ago

(In reply to James Teh [:Jamie] from comment #1)

> So, for IA2, I'd prefer the text attributes weren't removed.

I don't mind the text attributes being present; I simply don't want to have to use them to find sub and sup elements.

> That leaves two options:
> 1. Expose this differently for IA2 and ATK. I suspect Mozilla won't be
> willing to do this. :)

Agreed. :)

> 2. Do expose sup and sub as separate accessible objects, but keep the text
> attributes as well. That should keep both IA2 and ATK happy.

Agreed.

> Note that this will result in extra embedded objects (which we all know and love),
> which means a slight perf hit for IA2.

I assume that the hit will only happen when the sub or sup elements are used. If that is indeed the case, I would hope the hit would go unnoticed because sub and sup aren't used all that often (right?).

	James Teh [:Jamie]
	Comment 3 • 9 years ago

(In reply to Joanmarie Diggs from comment #2)
> I don't mind the text attributes being present; I simply don't want to have
> to use them to find sub and sup elements.
Out of interest, what will you use to detect this? Are there roles for sup and sub or will you use an attribute?

> > Note that this will result in extra embedded objects (which we all know and love),
> > which means a slight perf hit for IA2.
> I assume that the hit will only happen when the sub or sup elements are
> used. If that is indeed the case, I would hope the hit would go unnoticed
> because sub and sup aren't used all that often (right?).
Hopefully, you're right.

	Joanmarie Diggs Reporter
	Comment 4 • 9 years ago

(In reply to James Teh [:Jamie] from comment #3)
> Out of interest, what will you use to detect this? Are there roles for sup
> and sub or will you use an attribute?

We have new roles for subscript and superscript.
 
> Hopefully, you're right.

Fair enough. Perhaps it could be done either as a try build or put into nightly so you could verify that. And if the change does have a noticeable impact, then we can discuss further how to proceed.

	Joanmarie Diggs Reporter
	Comment 5 • 9 years ago

Ping?