703198 - JAWS doesn't announce combobox navigation in collapsed combobox

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

Attachment: patch

regression from bug 673958. JAWS expects focus event for changed option. Similar to Orca. I contacted to FS to check if they can do IE way for Firefox (i.e. based on value change events), that's should be consistent across browsers and ATs (NVDA picks up value change event well). In either way I think we should add a hook for JAWS to keep it working.

Comment 1

[Marco Zehe (:MarcoZ)]

Comment on attachment 575099 [details] [diff] [review]
patch

r=me, even though I think this is rather hacky. ;) But at least it will get JAWS working with those comboboxes.

Comment 2

[alexander :surkov (:asurkov)]

(In reply to Marco Zehe (:MarcoZ) from comment #1)
> Comment on attachment 575099 [details] [diff] [review] [diff] [details] [review]
> patch
> 
> r=me, even though I think this is rather hacky. ;) But at least it will get
> JAWS working with those comboboxes.

absolutely, in bug 673958 we ended up that we shouldn't fire focus event for options when they aren't visible so I do this for JAWS only. Marco, would you mind to check how other screen readers feel about that like WE or Supernova?

Comment 3

[alexander :surkov (:asurkov)]

inbound land https://hg.mozilla.org/integration/mozilla-inbound/rev/8914d038bd09

Comment 4

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/8914d038bd09

Comment 5

[alexander :surkov (:asurkov)]

Comment on attachment 575099 [details] [diff] [review]
patch

need to be ported into Firefox 10 where regression was introduced (bug 673958). The bug prevents JAWS (major screen reader) to work with comboboxes. Patch is trivial, the code is JAWS specific (nobody else is affected).

Comment 6

[Ed Morley [:emorley]]

I believe this is responsible for various a11y MozAfterPaint regressions on dev.tree-management:
http://graphs-new.mozilla.org/graph.html#tests=[[113,63,1],[113,63,12],[113,131,1]]&sel=1321910662231.955,1322043257360&displayrange=7&datatype=running

Comment 7

[alexander :surkov (:asurkov)]

then it means ::GetModuleHandleW() call is high expensive

Comment 8

[alexander :surkov (:asurkov)]

Comment on attachment 575099 [details] [diff] [review]
patch

canceling request until perf problems are figured out

Comment 9

[alexander :surkov (:asurkov)]

Attachment: patch2

let's see if it helps

Comment 10

[Marco Zehe (:MarcoZ)]

Comment on attachment 576441 [details] [diff] [review]
patch2

r=me. Good idea to switch these checks around so the ::getModuleHandle call doesn't get executed each time.

Comment 11

[alexander :surkov (:asurkov)]

inbound land (patch2) https://hg.mozilla.org/integration/mozilla-inbound/rev/c7a66776a891

Comment 12

[alexander :surkov (:asurkov)]

(In reply to alexander surkov from comment #11)
> inbound land (patch2)
> https://hg.mozilla.org/integration/mozilla-inbound/rev/c7a66776a891

yep, it fixes regression: http://graphs-new.mozilla.org/graph.html#tests=[[113,63,1],[113,63,12],[113,131,1]]&sel=1322043767971.9514,1322067320066&displayrange=7&datatype=running

Comment 13

[Ed Morley [:emorley]]

Great, thanks! :-)

Comment 14

[Matt Brubeck (:mbrubeck)]

https://hg.mozilla.org/mozilla-central/rev/c7a66776a891

Comment 15

[alexander :surkov (:asurkov)]

Comment on attachment 576441 [details] [diff] [review]
patch2

see comment #5 (perf regression is fixed)

Comment 16

[Marco Zehe (:MarcoZ)]

Alex, you might want to put up a unified patch for Aurora that just shows the diffs against the current Aurora code. This patch2 depends on patch1, which is confusing, esp if none of us ends up transplanting the patch in case we get approval. So my recommendation is to make a patch specific to Aurora that only contains the fixed version of the code, and request approval for that.

Comment 17

[alexander :surkov (:asurkov)]

Attachment: combined patch

Comment 18

[Marco Zehe (:MarcoZ)]

Pushed to Aurora on Alexander's behalf: http://hg.mozilla.org/releases/mozilla-aurora/rev/d36cab2ac025

Comment 19

[alexander :surkov (:asurkov)]

Thanks, Marco!

Comment 20

[u279076]

Is this something QA can verify?

Comment 21

[alexander :surkov (:asurkov)]

(In reply to Anthony Hughes, Mozilla QA (irc: ashughes) from comment #20)
> Is this something QA can verify?

yeah, you need JAWS for this though. If you have JAWS running then you should tab into HTML select (combobox) and then arrow down through its options. If JAWS reads them then bug is fixed.

Comment 22

[u279076]

Unfortunately, this is not something I'm set up to test. For now I will mark this qa- in the hopes that someone from the community can help verify this fix.