-
Notifications
You must be signed in to change notification settings - Fork 124
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Revisit aria-setsize="-1" authoring requirement #1759
Comments
I started looking into this. Since this is a "copious spare time" thing for me, I expect doing multiple updates. But regarding your observations of the inconsistencies:
My counter proposal to the "MAY" -- or taking it out completely -- is:
Lastly, regarding the special casing that is only on Linux: Orca doesn't actually need |
Note: The previous comment has been edited to point to the correct issue and commit in Orca. C&P fail before. Sorry for the noise. |
The Windows situation, using Chrome Canary which exposes the value of -1:
I've also confirmed that Firefox exposes the calculated count via IA2, just like it does on Linux via ATK. |
The macOS situation, using Chrome Canary, my locally-built Chromium, and Safari with the following test case: <ul>
<li aria-setsize="-1" aria-posinset="5">apples</li>
<li aria-setsize="-1" aria-posinset="6">bananas</li>
</ul>
<ul>
<li aria-setsize="10" aria-posinset="5">apples</li>
<li aria-setsize="10" aria-posinset="6">bananas</li>
</ul> Note: To examine the accessibility tree, I used Findings:
@cookiecrook: Am I doing something silly, missing a setting/command, etc. to get VoiceOver to report the author-provided posinset and setsize? |
Another finding: Unlike |
@joanmarie thanks for the really thorough testing and issue logging! That makes it really clear where the issue comes from browser mappings vs. screen reader behavior. Would it make sense to push more for screen reader vendors to commit to actually handling -1 in a meaningful way, and revisit removing it only if that doesn't happen? The feature as a whole does make sense to me, my concern is just that without screen reader implementations the |
You're quite welcome! This sort of thing (answering "why doesn't this authoring work like the spec says it should?!") strikes me as a good use of the spare time I do have. So if you / the WG needs this again, please ping me.
I wholeheartedly agree with the above. @cookiecrook do you think VoiceOver should be modified to present sets of unknown size? (See this comment above.) I don't know if commenting on the NVDA and JAWS bugs as authors / companies who need this supported would make it happen faster (i.e. due to limited developer resources on those teams). But it might be a tad more compelling than the Linux screen reader developer saying "this needs doing." 😉 |
action items:
|
The current text of
aria-setsize
has the following requirement for when the size is unknown:The issue is that while there is an author requirement to set the value to
-1
, there is no corresponding UA requirement for how to handle-1
. There is also no apparent support that I could find foraria-setsize="-1"
indicating an unknown number of items (though I did not test on Linux).In core AAM the only mapping to special-case
-1
is ATK/AT-SPI, which has the following:In tests, I found a mix of directly announcing the set size as -1, announcing the set size as 0, and ignoring the attribute entirely in favor of calculating the set size based on DOM. The result is weird or confusing UX, such as landing on "row 1 of 0" (in the case of a grid with a header row and loading body rows).
I think the author "SHOULD" should be removed in favor of "MAY", or potentially the entire section on "-1" should be taken out until there is at least minimal support for it.
related: #1323
The text was updated successfully, but these errors were encountered: