VO speaks over itself when navigating videos at twitter.com
Categories
(Core :: Disability Access APIs, defect)
Tracking
()
People
(Reporter: morgan, Assigned: morgan)
References
Details
Attachments
(1 file)
STR:
- Enable VO
- Visit https://twitter.com/TwitterA11y/status/1395442951338160133
- Navigate to the video, hit pause/play or unmute
Expected: VO describes the focused control and then is silent while the video plays
Actual: VO begins to read the tweet from the beginning every 1sec, including (at least in the caption window -- it never gets to voice it before it speaks over itself) the time remaining on the slider.
| Assignee | ||
Comment 1•3 years ago
|
||
Comment 3•3 years ago
|
||
| bugherder | ||
Comment 4•3 years ago
|
||
Unfortunately, this breaks the test scenario I introduced the AXTitleChanged event for in Bug 1682247. Which means that unlike on Windows, the expanded or collapsed post is no longer automatically being read as it is shown/hidden in the Pinafore Mastodon client. Perhaps instead of stripping the event and test case, could it be special-cased that the AXTitleChanged event is only fired if the element in question also has focus? The way this bug fix stands now, it introduces a regression.
| Assignee | ||
Comment 5•3 years ago
|
||
(In reply to Marco Zehe (:MarcoZ) [out sick] from comment #4)
Unfortunately, this breaks the test scenario I introduced the AXTitleChanged event for in Bug 1682247. Which means that unlike on Windows, the expanded or collapsed post is no longer automatically being read as it is shown/hidden in the Pinafore Mastodon client. Perhaps instead of stripping the event and test case, could it be special-cased that the AXTitleChanged event is only fired if the element in question also has focus? The way this bug fix stands now, it introduces a regression.
Thanks for pointing this out, Marco! Does the test case you talked about on that bug work in Safari? I don't have a mastadon account so I can't check unfortunately :(
When I was investigating this bug, we found Safari only provides labels for <article> elements, not labels and titles as we were previously doing. If this works there, I'm wondering if they're using title-changed despite the element not having a title, or if they're using another event.
Comment 6•3 years ago
|
||
No, it neither works in Safari nor Chrome. I added this to a) make it behave like Firefox (and Chrome) behave on Windows) and b) give us an edge over the others on MacOS and improve accessibility.
I just checked the Twitter test case on Windows with both Chrome and Firefox, and the video playing/being muted and the article label changing definitely don't interfere as badly as you experienced on MacOS. Hence my suggestion to restore the previous behavior, but do it only if the element actually has focus and the title change is of immediate interest to the user.
Updated•3 years ago
|
Comment 7•3 years ago
|
||
I was able to reproduce the issue on Mac 10.13 using build 97.0a1(20220107213135).
Verified as fixed on Mac 10.13 using build 97.0b9 and 98.0a1.
Comment 8•3 years ago
|
||
FYI, this problem again came up today, not specifically for Firefox. With the recent uptake of Mastodon, would it make sense to file a bug to reintroduce AXTitle for articles for very specific cases that would allow Firefox on Mac to work the same as on Windows in Pinafore? I know this would be different from Safari and Chromium, but VoiceOver obviously reacted positively to that when bug 1682247 was still working. Perhaps if the firing of the AXTitleChangedEvent was restricted to only articles with the focused state?
| Assignee | ||
Comment 9•3 years ago
|
||
(In reply to Marco Zehe (:MarcoZ) from comment #8)
FYI, this problem again came up today, not specifically for Firefox. With the recent uptake of Mastodon, would it make sense to file a bug to reintroduce AXTitle for articles for very specific cases that would allow Firefox on Mac to work the same as on Windows in Pinafore? I know this would be different from Safari and Chromium, but VoiceOver obviously reacted positively to that when bug 1682247 was still working. Perhaps if the firing of the AXTitleChangedEvent was restricted to only articles with the focused state?
I think this is another reason to consider bug 1781153.
If we solidify when to use title vs. description, we can fire AXTitleChanged appropriately, and maybe we'll have less fall out :)
In the short term, I'd need to see if checking the focus state before firing the event helps us avoid the issue in #c0
Description
•