749719 - Activate accessibility mode when device a11y is enabled

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, defect)

Description

[Eitan Isaacson [:eeejay]]

Utilize the AccessFu js module. It will check if platform accessibility is enabled, if it is it will override the d-pad keys and superimpose the virtual cursor position.

Comment 1

[Eitan Isaacson [:eeejay]]

Attachment: Attach AccessFu in Android chrome script.

I'll flag this for review soon.

Comment 2

[Eitan Isaacson [:eeejay]]

Attachment: Attach AccessFu in Android chrome script.

This sets the default accessibility.accessfu to 2 (auto).

And.. I think this is ready for review! I think what should be important to the reviewer is that this does not affect users who don't have a Android accessibility enabled.

Comment 3

[Eitan Isaacson [:eeejay]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/106a74f83351

Comment 4

[David Bolter [:davidb] (NeedInfo me for attention)]

\o/

Comment 5

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/106a74f83351

Comment 6

[Matt Brubeck (:mbrubeck)]

It looks like this caused a regression in the Tp4 Mobile benchmark:

http://graphs.mozilla.org/graph.html#tests=[[84,63,20]]&sel=1336140433529.5078,1336318390388.1465&displayrange=7&datatype=running

Does AccessFu have any effect on the browser if it's not enabled?  (It looks like it shouldn't.)  Is there any reason it would be enabled on the Talos test slaves?

Comment 7

[Eitan Isaacson [:eeejay]]

I'm having a hard time understanding the test and the result. The only thing I could think of is the accessibility check on startup. that should probably be decoupled. I thought it was on a separate thread, but it is not. Could a regression in startup time affect this test?

http://mxr.mozilla.org/mozilla-central/source/mobile/android/base/GeckoAppShell.java#1848

If you think that might make sense, I could work on that tomorrow.

Comment 8

[Kartikaya Gupta (email:[e-mail])]

We should probably move both the Prompt:Show and Accessibility:IsEnabled chunks out into their own GeckoEventResponder implementations. Any string-related operations are expensive in Java, and doing two unnecessary string creations and compares on every message probably doesn't help us.

Comment 9

[Kartikaya Gupta (email:[e-mail])]

Filed bug 752905 for ^

Comment 10

[Eitan Isaacson [:eeejay]]

I think the patch in bug 752935 remedies the regression.