iframe has non-empty name - problems with examples and applicability [cae760]

[dd8]

Each browser / AT combination presents a slightly different user interface for navigating iframes:

• VoiceOver/Safari 13 - exposes iframe as container in normal reading order named by title, but iframe contents not in until user presses Control-Option-Shift-Down Arrow to enter iframe, first item inside iframe focused by tab key
• JAWS/IE11 - includes iframe contents in normal reading order, first item inside iframe focused by tab key and frame title not read
• JAWS/FF68 - includes iframe contents in normal reading order, first item inside iframe focused by tab key and frame title is read
• JAWS/Chrome - includes iframe contents in normal reading order, first item inside iframe focused by tab key and frame title is read
• NVDA/IE11 - includes iframe contents in normal reading order, first item inside iframe focused by tab key
• NVDA/FF68 - includes iframe contents in normal reading order, iframe is focusable but always silent (not named by title/acc-name)
• NVDA/Chrome - unable to test due to focus trap in browser address bar

On Windows iframe title is read when tabbing, but not in normal screen reader reading order
On Mac iframe title is read in normal screen reader reading order, but not when tabbing

Test results for the examples are

Example	VO/Safari 13	JAWS/IE11	JAWS/FF68	JAWS/Chrome	NVDA/FF68	NVDA/Chrome	NVDA/IE11
Passed 1	OK	OK	OK	OK	Fail	Untested	OK
Passed 2	Fail	OK	OK	OK	Fail	Untested	OK
Passed 3	Fail	OK	OK	OK	Fail	Untested	OK
Failed 1	OK	OK	Fail	OK	Fail	Untested	OK
Failed 2	Fail	OK	Fail	OK	Fail	Untested	OK
Failed 3	Fail	OK	Fail	OK	Fail	Untested	OK
Failed 4	Fail	OK	Fail	OK	Fail	Untested	OK
Inapplicable 1	OK	OK	OK	OK	OK	Untested	OK
Inapplicable 2	OK	OK	OK	OK	OK	Untested	OK
Inapplicable 3	Fail	OK	OK	Fail	OK	Untested	OK

On Safari untitled iframes are hard to use because there's no indication of frame contents, and users often assume they contain adverts. This is bad if the iframe contains something important - I've seen a Safari user unable to find a login form in an iframe for exactly this reason.

Inapplicable Example 3 fails in VoiceOver with Safari
The Safari issue on Inapplicable Example 3 is a consequence of the UI design (frames exposed as nested containers) but excluded from the applicability. The problem happens in normal reading order, not when tabbing between focusable items. I think the applicability needs extended to cover problems with VoiceOver/Safari by removing the following:

and that can be accessed by sequential focus navigation

https://act-rules.github.io/rules/cae760#applicability

Passed examples 1,2,3 always fail on NVDA/FF68 because tabbing to the frame always produces silence - even if a title or ARIA label is provided. This looks like a bug so I don't think the examples or applicability should change for this.

Passed examples 2,3 always fail on VoiceOver with Safari because it doesn't use acc-name calculation for iframes (but does use the iframe name which is not supposed to be human readable). Again this looks like a bug so I don't think the examples of applicability should change for this.

NVDA with Chrome is untested because there's a focus trap in address bar so you can't tab to the page to test.

tl/dr; remove "and that can be accessed by sequential focus navigation" from applicability and make Inapplicable Example 3 a fail.

[WilcoFiers]

Thanks for the thorough research. I totally agree with your conclusions. That applicability looked odd, against my better judgement I approved that... oops!

The passed example 2 / 3 Safari issue is known and was noted in accessibility support. The NVDA one is not and definitely should be.

[jeeyyy]

Thanks for the extensive research @dd8. Sequential focus navigation has been part of this rule for a long time.

I dug up through the history, and found these discussions:

• What does it actually mean to be part of sequential focus navigation? (cae760)
• Some AT test results

[WilcoFiers]

Need a volunteer to work on this rule.

[carlosapaduarte]

@WilcoFiers assigning this to you since you are already handling w3c/wcag-act#473