761067 - Key events is being mistakenly sent twice with nested <iframe mozbrowser>

Status: RESOLVED INVALID

(Firefox OS Graveyard :: General, defect)

Description

[Tim Guan-tin Chien [:timdream] (please needinfo)]

With bug 757486 landed and bug 755648 pending, we now rely on <iframe mozbrowser> to forward keys to the parent frame.

However, keys from nested <iframe mozbrowser> is being forward to the System app more than once. For the current Gaia, this includes, at least,

A) Browser App browser content

STR:

1) In Gaia, go to browser app and navigator to any web address
2) Interact with the web content by panning the view, have the view iframe gaining focus
3) Try to adjust the volume

Expected:

* Volume indication should appear and go up/down one level

Actual:

* It goes up/down by two level because the key events is being sent twice.

Comment 1

[Mounir Lamouri (:mounir)]

Can you write a smaller testcase? At least help to identify the origin of the bug? On top of my head, I don't see what could produce that except if, for some reasons, an event was sent to the system app by a specific path and another was sent trough the mozbrowser hierarchy...

Comment 2

[Justin Lebar (not reading bugmail)]

It's not totally clear from comment 0, but but it sounds like this happens specifically when you have 

  system app
    <iframe mozbrowser>
      <iframe mozbrowser>

Comment 3

[Mounir Lamouri (:mounir)]

Attachment: Hypotethic test case

With the attached test case, everything is working fine. Please, check locally.

Comment 4

[Tim Guan-tin Chien [:timdream] (please needinfo)]

https://gist.github.com/2817145

I can reproduce the behavior with your test case, and mine.

I've specifically stopped shell.js from generating key events so the events must have come from mozbrowser

Comment 5

[Mounir Lamouri (:mounir)]

I don't understand. You are calling addEventListener twice, one for each phase so it is expected that you get the event being handled twice, one for each phase. Or do you mean you get two handling for each phases? I can't reproduce that locally.

Comment 6

[Tim Guan-tin Chien [:timdream] (please needinfo)]

(In reply to Mounir Lamouri (:volkmar) (:mounir) from comment #5)
> I don't understand. You are calling addEventListener twice, one for each
> phase so it is expected that you get the event being handled twice, one for
> each phase. Or do you mean you get two handling for each phases? I can't
> reproduce that locally.

Yes, I am getting two handling for each phases, unfortunately. :(

I am building SGS2 out of mozilla-b2g/b2g repo and Gecko from mozilla-central BTW.

Comment 7

[David Flanagan [:djf]]

In my work on https://bugzilla.mozilla.org/show_bug.cgi?id=762362 I'm seeing something similar.

In b2g/chrome/content/shell.js, I'm getting two keydown keypress keydown keypress keyup keyup for the sleep, volume up and volume down hardware buttons.

For the soft Home button, I only get keydown keypress keyup.

I see this on a GB otoro with whatever is in today's in mozilla-b2g gecko repo.

Comment 8

[David Flanagan [:djf]]

Adding dhylands, who thinks there might be a low-level hardware button debouncing issue.

Comment 9

[Tim Guan-tin Chien [:timdream] (please needinfo)]

Doesn't happen anymore as we have changed the way hardware keys are handled.

Comment 10

[Gabriele Svelto [:gsvelto]]

Just noticed that we still have a comment regarding this bug in the code:

http://hg.mozilla.org/mozilla-central/file/87730d939be7/b2g/chrome/content/shell.js#l476

Comment 9 seems to suggest that we don't need the debouncing keys anymore, shall I file a follow-up to remove that code?