Steps to reproduce:

1. Compose a message and attach multiple files
2. Select one of the attachments in the list
3. Launch Accerciser, and select the list box corresponding to the attachment list
4. In Accerciser's iPython console type the following:

s = acc.querySelection()
s.nSelectedChildren

Expected result: 1 selected child would be reported.

Actual result: 0 selected children are reported.

Impact: Orca uses the accessible object:selection-changed event to present selection changes. When it gets the event, it uses the selection interface to get the number of selected children and, if it's greater than 0, present the newly-selected item. Because of this bug, Orca is failing to present the selected attachment when the user arrows among items.

Hi,

This bug seems the technical explanation for bug 1519346.

should be merged then. Thanks for this useful complement as devs so far fail to fix this.

Regards

(In reply to Jean-Philippe MENGUAL from comment #1)

Hi,

This bug seems the technical explanation for bug 1519346.

So this is the cause of bug 1519346, and we don't need to wait for bug 1526811 ?

(In reply to Wayne Mery (:wsmwk) from comment #2)

(In reply to Jean-Philippe MENGUAL from comment #1)

Hi,

This bug seems the technical explanation for bug 1519346.

So this is the cause of bug 1519346, and we don't need to wait for bug
1526811 ?

I think this bug explains both bug 1519346 and bug 1526811. Due to this bug, user cannot hear the attached file on which the focus is, and he cannot browse in the attached files list. The presentd reproduction scenario seems closer bug 1519346 but I would not be surprised the bug shown by Accerciser has consequences on the other.

I have good news and bad news for you.

The good news: The specific problem in the opening report is fixed. Thanks!

The bad news: The listbox claims to implement the table interface and gives bogus results. More specifically, try the following:

Perform the steps described in the opening report, BUT replace step 4 with the following:
t = acc.queryTable()
t.nRows
t.nColumns

I did this with three attachments, both in the compose window and in the preview window.

Results:

• The number of reported rows is 3. This makes sense in the compose window. It does NOT make sense in the preview window where all three attachments are in a single row.
• The number of reported columns is 0. This doesn't make sense in either place. If it's really a table (which it claims to be), then it should report a sane number of columns.

Given the new interface, maybe this particular widget just shouldn't implement AtkTable...

Another blind user complains about that on the Orca list, see https://mail.gnome.org/archives/orca-list/2019-May/msg00004.html

@Jorg: Is there a plan to work on that ? At the current time, a blind person couldn't known which attachment has been added.

To workaround that, we're always on TB 52 for now.

Best regards,
Alex

Sorry, I don't manage resources. Magnus, there are a few bugs affecting vision-impaired people.

We need to finish bug 1523607 first. Then we should look into if there's anything left to do in that regard.

(In reply to Magnus Melin [:mkmelin] from comment #7)

We need to finish bug 1523607 first. Then we should look into if there's
anything left to do in that regard.

FYI, we've the issue in TB 60 so blind users that would like to read attachments lists have to stay on TB 52.

Best regards,
Alex.

(In reply to Magnus Melin [:mkmelin] from comment #7)

We need to finish bug 1523607 first. Then we should look into if there's anything left to do in that regard.

That's done (modulo regression bug 1554943). What's next, do we need Alex to test?

(In reply to Wayne Mery (:wsmwk) from comment #9)

(In reply to Magnus Melin [:mkmelin] from comment #7)

We need to finish bug 1523607 first. Then we should look into if there's anything left to do in that regard.

That's done (modulo regression bug 1554943). What's next, do we need Alex
to test?

What test do you expect to? For bug 1554943 ? Or this one?

Best regards

The plan is to first also do bug 1547947 which is quite small, and with that all of the attachment widget is de-xbl'd. After that it would be useful to get feedback on what is broken (or not).

Alex, it will be helpful to have your quick feedback by retesting nightly build now that bug 1547947 is fixed. Unless alessandro can test it.

(Based on Magnus' comments I assume we won't be fixing ESR 60)

(In reply to Wayne Mery (:wsmwk) from comment #12)

Alex, it will be helpful to have your quick feedback by retesting nightly
build now that bug 1547947 is fixed. Unless alessandro can test it.

(Based on Magnus' comments I assume we won't be fixing ESR 60)

Hi,

Actually we thought it was a technical fix needed to fix bug 1519346 . Hence our few feedback.

So I tested (I go on using TB nkghtly everyday although a11y bugs in th calendar which make me switch to 60 to go on working). I am pleased with telling you that now, I can browse the attachments via the keyboard and the screen reader speaks any file when I browse them via up/down arrow keyss. So the bug seems fixed. Many thanks for this effort!

Best regards,

Bonjour Jean-Philippe, that's good news!

We'd like to know whether this was fixed by bug 1547947. In order to check that, would you try with a Daily from before and after that bug landed, so, 11th June before the bug landed and 12th June after it landed from here:
http://ftp.mozilla.org/pub/thunderbird/nightly/2019/06/2019-06-11-09-02-10-comm-central/
http://ftp.mozilla.org/pub/thunderbird/nightly/2019/06/2019-06-12-07-19-42-comm-central/

What is the status of the other accessibility-related bugs? You mentioned bug 1519346, is that fixed as well? And bug 1519330? The last one is the Calendar bug which isn't fixed, if I understood you correctly.

I see you also reported bug 1552212, so that will need some investigation.

Thanks a lot for your help!

(In reply to Jorg K (GMT+2) from comment #14)

Bonjour Jean-Philippe, that's good news!

Hi! :)

We'd like to know whether this was fixed by bug 1547947. In order to check
that, would you try with a Daily from before and after that bug landed, so,
11th June before the bug landed and 12th June after it landed from here:
http://ftp.mozilla.org/pub/thunderbird/nightly/2019/06/2019-06-11-09-02-10-
comm-central/
http://ftp.mozilla.org/pub/thunderbird/nightly/2019/06/2019-06-12-07-19-42-
comm-central/

Well it makes me surprised but it seems the problem was fixed in both builds. I cannot track when it was fixed exactly then.

What is the status of the other accessibility-related bugs? You mentioned
bug 1519346, is that fixed as well? And bug 1519330? The last one is the

I think we can consider Bug 1519346 is fixed, yes. Should I close it?

Calendar bug which isn't fixed, if I understood you correctly.

Yes, I have 2 critical bugs: Bug 1519330 and Bug 1520781. They make calendar unusable for a blind person.

I see you also reported bug 1552212, so that will need some investigation.

Indeed. It is less critical than Bug 1533291 however.

Other "normal" bug (not critical): Bug 1519348

Thanks a lot for your help!

Thanks for you work, it is wonderful to cooperate with your team

OK, so I'm closing this now as well as bug 1519346. Last question: Does TB 68 beta 1 work?
https://www.thunderbird.net/en-US/thunderbird/beta/all/

If not, I can give you a link to a beta 2 binary, you need it for Linux64, right?

FIXED as per comment #13.

(In reply to Jorg K (GMT+2) from comment #16)

OK, so I'm closing this now as well as bug 1519346. Last question: Does TB
68 beta 1 work?
https://www.thunderbird.net/en-US/thunderbird/beta/all/

Yes, it works fine now.

regards

If not, I can give you a link to a beta 2 binary, you need it for Linux64,
right?

FIXED as per comment #13.

Hmm, it works in TB 68 beta 1 already, then it was NOT fixed by bug 1547947 since this is bug is NOT in beta 1. Also, you didn't see any changes between the 11th and 12th of May when that bug landed on trunk. So it must have gotten fix some other way.

However, we uplifted that bug to TB 68 beta 2, so I'd like to know whether things still work in that version. For Linux64, please try:
https://queue.taskcluster.net/v1/task/JYPh9x8_SemlzUawdLPetQ/runs/0/artifacts/public/build/target.tar.bz2

If you need a binary for another OS, please let me know.

(In reply to Jorg K (GMT+2) from comment #18)

Hmm, it works in TB 68 beta 1 already, then it was NOT fixed by bug 1547947
since this is bug is NOT in beta 1. Also, you didn't see any changes between
the 11th and 12th of May when that bug landed on trunk. So it must have
gotten fix some other way.

ok. I am unable to say when then. I had it, did workarounds, and seeing this bug handled I tried again and it worked. Difficult to know when

However, we uplifted that bug to TB 68 beta 2, so I'd like to know whether
things still work in that version. For Linux64, please try:
https://queue.taskcluster.net/v1/task/JYPh9x8_SemlzUawdLPetQ/runs/0/
artifacts/public/build/target.tar.bz2

With this binary, it works fine.

Best regards

If you need a binary for another OS, please let me know.