1268677 - embedded object index might be no updated

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

Attachment: patch

Comment 1

[alexander :surkov (:asurkov)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=31ede417520a

Comment 2

[Marco Zehe (:MarcoZ)]

[Tracking Requested - why for this release]: Regression fix from reorg bug 1255009 and bug 1261425. Requesting tracking to make sure this lands on Aurora once fixed on 49.

Comment 3

[Marco Zehe (:MarcoZ)]

Observation with the try server build on this bug:
Until recently, there was a bug in Firefox that affected the table containing the tracking flags on a Bugzilla bug like this one. Initially, that table is hidden. Once you click the Edit link next to the "Tracking Flags:" text, the table gets inserted into the accessibility tree that contains all of those tracking flags. A table with about 20 rows or so. When this happened, NVDA's virtual buffer always needed a refresh to see that table. Other dynamic insertions were no problem, just this particular one. So the steps were, with NVDA running:

1. Find the Edit link next to the "Tracking Flags:" text, and press Enter on it.
2. Notice that the table wasn't there that actually contained these flags.
3. Press NVDA+F5 to refresh the virtual buffer.

Result: Table below the "Tracking Flags:" text was now present.

After bug 1255009 and bug 1261425 landed, this bug went away. The table would now always appear right below the text without NVDA needing a refresh of the virtual buffer.

With the patch on this bug now, the bug returns, but even stronger, because not even a virtual buffer refresh will now make that particular table visible. Again, it is the table that gets inserted when you click the Edit link next to the "Tracking Flags:" text when initially bringing up this bug.

Perhaps this patch can be expanded to address this particular bug, since it seems very very closely related?

Comment 4

[alexander :surkov (:asurkov)]

I filed a new bug 1268916 for Marco's issue, I don't find anything suspicious in code.

Comment 5

[Marco Zehe (:MarcoZ)]

(In reply to alexander :surkov from comment #4)
> I filed a new bug 1268916 for Marco's issue, I don't find anything
> suspicious in code.

That's totally fine with me. So Yura, feel free to review, and Surkov, to land, since this definitely improves things in other places, and it needs to get back-ported to Aurora, too.

Comment 6

[Yura Zenevich [:yzen] (please use "needinfo")]

Comment on attachment 8746758 [details] [diff] [review]
patch

Review of attachment 8746758 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good, one nit.

::: accessible/tests/mochitest/hypertext/test_update.html
@@ +133,3 @@
>  
> +      this.invoke = function removeFirstChild_invoke()
> +      {        

nit: whitespace

Comment 7

[alexander :surkov (:asurkov)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/0bc628de412e5d63aae5ddb17d08e04b4d783197
Bug 1268677 - embedded object index might be not update, r=yzen

Comment 8

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/0bc628de412e

Comment 9

[Marco Zehe (:MarcoZ)]

Comment on attachment 8746758 [details] [diff] [review]
patch

Approval Request Comment
[Feature/regressing bug #]: Bug 1255009 and bug 1261425
[User impact if declined]: Missing content in screen reader virtual buffers on Windows.
[Describe test coverage new/current, TreeHerder]: Manual testing, has mochitest
[Risks and why]: Low risk, corrects indexing.
[String/UUID change made/needed]: None.

Comment 10

[Liz Henry (:lizzard)]

Regression from 48

Comment 11

[Liz Henry (:lizzard)]

Comment on attachment 8746758 [details] [diff] [review]
patch

Fix for recent a11y regression

Comment 12

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/releases/mozilla-aurora/rev/e6ca64925a10