Closed Bug 628668 Opened 14 years ago Closed 14 years ago

Accessible tree not always created completely, regression from part2 of bug 606924

Categories

(Core :: Disability Access APIs, defect)

x86
Windows 7
defect
Not set
major

Tracking

()

RESOLVED WORKSFORME
Tracking Status
blocking2.0 --- final+

People

(Reporter: MarcoZ, Unassigned)

References

Details

(Keywords: regression, Whiteboard: [hardblocker])

STR: 1. With NVDA running, open any bug from an external application like Thunderbird. This seems to have an impact. Expected: Whole bug page should be rendered. Actual: The bug page is rendered to a11y APIs until the separator that separates the last table of the blocking flags and the table that contains the "Add attachment" link, attachments to the bug etc. From this point downward, EVERYTHING is missing. 2. Press CTRL+R to refresh. Result: Now, the page usually loads fully into NVDA's virtual buffer. This can be attributed to the checkin of bug 606924 part2 that happened on the 23rd of January. This was the only a11y checkin between the January 21 and January 24 nightly builds (there were no builds on the 22nd and 23rd). This must therefore be a regression from that checkin.
blocking2.0: --- → ?
(In reply to comment #0) > Expected: Whole bug page should be rendered. > Actual: The bug page is rendered to a11y APIs until the separator that > separates the last table of the blocking flags and the table that contains the > "Add attachment" link, attachments to the bug etc. From this point downward, > EVERYTHING is missing. Did you have a chance to dump a tree?
(In reply to comment #1) > (In reply to comment #0) > > > Expected: Whole bug page should be rendered. > > Actual: The bug page is rendered to a11y APIs until the separator that > > separates the last table of the blocking flags and the table that contains the > > "Add attachment" link, attachments to the bug etc. From this point downward, > > EVERYTHING is missing. > > Did you have a chance to dump a tree? No, AccProbe is currently broken on my machine, haven't had time to investigate. This was found out by investigating the tree with NVDA's object navigation.
(In reply to comment #0) > STR: > 1. With NVDA running, open any bug from an external application like > Thunderbird. This seems to have an impact. Any idea why? I guess when NVDA recognizes an application switch it starts building its VB immediately? Perhaps this happens asynchronous to its event listening?
blocking2.0: ? → final+
Whiteboard: [hardblocker]
Marco, does this bug happen when the bug 606924 patches are all applied?
(In reply to comment #3) > Any idea why? I guess when NVDA recognizes an application switch it starts > building its VB immediately? Perhaps this happens asynchronous to its event > listening? Nope. We only render a buffer if the document has focus and does not have the busy state set.
This just bit me again when opening this bug from Thunderbird with Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b11pre) Gecko/20110126 Firefox/4.0b11pre So the next checkin for bug 606924 didn't fix it yet. I do not remember ever seeing this in the try-server build for bug 606924 (the latest one).
Marco, please fix your accprobe or use any other a11y tool to check the tree. It's very important to know whether this is incomplete accessible tree or broken accessible tree (like empty text leaf accessibles are presented).
Marco, did you observer this bug again on latest nightlies? Can this bug be closed now?
This bug can be closed WFM, since I didn't see this reappear since Thursday January 27 nightly builds. So this was only a problem in the January 25 and 26 nightly builds.
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.