854796 - HTML a and area having no @href shouldn't have link role

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

Jason noticed in https://www.w3.org/Bugs/Public/show_bug.cgi?id=16629 that it would be more correct in MSAA terms to not map html:a and html:area having no @href attribute into link role.

CVS history doesn't have records about this and we do that from beginning of times.

IE maps html:a and html:area having no @href into ROLE_SYSTEM_TEXT.

Is there objections to not follow this?

Comment 1

[alexander :surkov (:asurkov)]

w3c HTML area related bug: https://www.w3.org/Bugs/Public/show_bug.cgi?id=16761

Comment 2

[James Teh [:Jamie]]

(In reply to alexander :surkov from comment #0)
> CVS history doesn't have records about this and we do that from beginning of
> times.
I suspect it was for IE compatibility; see below.

> IE maps html:a and html:area having no @href into ROLE_SYSTEM_TEXT.
Are you sure? This certainly isn't the case for me with IE 9. <a>foo</a> gives me ROLE_SYSTEM_LINK. It even exposes STATE_SYSTEM_LINKED, which is really annoying.

I certainly think this is a good idea in theory; we've always had to hack around it in NVDA. Two things to be aware of:
1. Some ATs might depend on <a name="foo">bar</a> having a role of link. This would need to be checked. NVDA doesn't.
2. <a name="foo">blah</a> should still produce an accessible, even if it has a generic role such as section or text frame, so that it can be the target of a scrolling start event.

Comment 3

[alexander :surkov (:asurkov)]

(In reply to James Teh [:Jamie] from comment #2)
> (In reply to alexander :surkov from comment #0)
> > CVS history doesn't have records about this and we do that from beginning of
> > times.
> I suspect it was for IE compatibility; see below.
> 
> > IE maps html:a and html:area having no @href into ROLE_SYSTEM_TEXT.
> Are you sure? This certainly isn't the case for me with IE 9. <a>foo</a>
> gives me ROLE_SYSTEM_LINK. It even exposes STATE_SYSTEM_LINKED, which is
> really annoying.

Oh, I didn't check. I supposed they suggested that wording in the spec.

> I certainly think this is a good idea in theory; we've always had to hack
> around it in NVDA. Two things to be aware of:
> 1. Some ATs might depend on <a name="foo">bar</a> having a role of link.
> This would need to be checked. NVDA doesn't.
> 2. <a name="foo">blah</a> should still produce an accessible, even if it has
> a generic role such as section or text frame, so that it can be the target
> of a scrolling start event.

sure

Comment 4

[alexander :surkov (:asurkov)]

So we shouldn't use HTMLLinkAccessible for html:a having no @href (see HTMLLinkAccessible::IsLinked(), use HyperTextAccessible instead. Don't use HTMLAreaAccessible for area (see HTMLImageMapAccessible), use LeafAccessible class instead.

Comment 5

[alexander :surkov (:asurkov)]

Jamie, is it normal/correct that ROLE_SYSTEM_TEXT have nested ROLE_SYSTEM_TEXT?

Comment 6

[James Teh [:Jamie]]

(In reply to alexander :surkov from comment #5)
> Jamie, is it normal/correct that ROLE_SYSTEM_TEXT have nested
> ROLE_SYSTEM_TEXT?
I'd say this is abnormal. I think it'd be better to use text frame for the parent role.

Comment 7

[Jason Kiss]

Apologies. Jamie, of course, is totally correct. IE uses ROLE_SYSTEM_LINK and STATE_SYSTEM_LINKED for both <a> and <area>, whether or not there's an @href.

Regardless of IE's behaviour, are you suggesting a better mapping for <a> and <area> without @href would be IA2_ROLE_TEXT_FRAME? Or is it better for AT if the implementation is the same across browsers and so stick with ROLE_SYSTEM_LINK but not expose STATE_SYSTEM_LINK?

Comment 8

[James Teh [:Jamie]]

(In reply to Jason Kiss from comment #7)
> Regardless of IE's behaviour, are you suggesting a better mapping for <a>
> and <area> without @href would be IA2_ROLE_TEXT_FRAME?
It is a more correct mapping, certainly.
> Or is it better for
> AT if the implementation is the same across browsers and so stick with
> ROLE_SYSTEM_LINK but not expose STATE_SYSTEM_LINK?
Better is a very subjective term here. It could break backwards compatibility, so it'll require a lot of testing. I'm happy for it to be made - NVDA will be just fine with it - but other AT vendors may disagree. In other words, with the nicest possible intent, I don't really care either way. :)

Comment 9

[alexander :surkov (:asurkov)]

Update, Jason suggested to use IA2_ROLE_SHAPE for areas having no @href

Comment 10

[Jason Kiss]

(In reply to alexander :surkov from comment #9)
> Update, Jason suggested to use IA2_ROLE_SHAPE for areas having no @href

It was really James who provided the suggestion :)

https://twitter.com/jcsteh/status/362340665222377473

Comment 11

[alexander :surkov (:asurkov)]

(In reply to Jason Kiss from comment #10)
> (In reply to alexander :surkov from comment #9)
> > Update, Jason suggested to use IA2_ROLE_SHAPE for areas having no @href
> 
> It was really James who provided the suggestion :)
> 
> https://twitter.com/jcsteh/status/362340665222377473

Sorry, I misread it, I think I was reading the first 'J' letter only :)

Comment 12

[James Teh [:Jamie]]

Time has marched on. The HTML-AAM spec is clear that <a> without an href shouldn't be exposed with the link role. Chrome has implemented it this way for quite some time (perhaps since the beginning), so we can assume JAWS and other ATs don't have a problem with this. As noted above, NVDA is fine with it. We're currently failing testing/web-platform/tests/html-aam/roles.html [el-a-no-href] because of this.

Comment 13

[Nathan LaPré]

Attachment: Bug 854796: Anchor, area elements without href shouldn't have link role, r?Jamie

Per the HTML-AAM spec, a and area elements without href attributes should have
generic roles. This revision implements this preference in the link accessible's
NativeRole function by checking for the attribute there. This revision also changes
the relevant WPT test expectations.

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/7b4ce93feaf8
Anchor, area elements without href shouldn't have link role, r=Jamie,devtools-reviewers

Comment 15

[Cristina Horotan [:chorotan]]

https://hg.mozilla.org/mozilla-central/rev/7b4ce93feaf8

Comment 16

[Cristina Horotan [:chorotan]]

https://hg.mozilla.org/mozilla-central/rev/7b4ce93feaf8