1052272 - TEST-UNEXPECTED-FAIL | /builds/slave/test/gaia/apps/keyboard/test/marionette/uninstallation_test.js | Show Keyboard App after uninstallation "before each" hook

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Gaia::System::Input Mgmt, defect)

Description

[Rudy Lu [:rudyl] (inactive after 2015/7/3)]

example failure, 
https://tbpl.mozilla.org/php/getParsedLog.php?id=45723138&tree=Gaia-Try

--
2:08:18     INFO -    JavaScriptError: (17) TypeError: this.runningLayouts[layout.manifestURL] is undefined
22:08:18     INFO -    Remote Stack:
22:08:18     INFO -    @app://system.gaiamobile.org/js/input_frame_manager.js, line 57
22:08:18     INFO -        at Error.MarionetteError (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/error.js:67:13)
22:08:18     INFO -        at Object.Client._handleCallback (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:476:19)
22:08:18     INFO -        at /builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:510:21
22:08:18     INFO -        at TcpSync.send (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/drivers/tcp-sync.js:153:10)
22:08:18     INFO -        at Object.send (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:457:36)
22:08:18     INFO -        at Object.Client._sendCommand (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:503:19)
22:08:18     INFO -        at Object._executeScript (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:1468:19)
22:08:18     INFO -        at Object.executeAsyncScript (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:1249:19)
22:08:18     INFO -        at Object.Mgmt.getAll (/builds/slave/test/gaia/node_modules/marionette-apps/lib/mgmt.js:37:19)
22:08:18     INFO -        at list (/builds/slave/test/gaia/node_modules/marionette-apps/lib/list.js:14:20)
22:08:18     INFO -        at getApp (/builds/slave/test/gaia/node_modules/marionette-apps/lib/getapp.js:45:10)
22:08:18     INFO -        at switchToApp (/builds/slave/test/gaia/node_modules/marionette-apps/lib/switchtoapp.js:20:10)
22:08:18     INFO -        at Context.<anonymous> (/builds/slave/test/gaia/apps/keyboard/test/marionette/uninstallation_test.js:90:17)
22:08:18     INFO -        at callFn (/builds/slave/test/gaia/node_modules/mocha/lib/runnable.js:223:21)
22:08:18     INFO -        at Hook.Runnable.run (/builds/slave/test/gaia/node_modules/mocha/lib/runnable.js:216:7)
22:08:18     INFO -        at next (/builds/slave/test/gaia/node_modules/mocha/lib/runner.js:258:10)
22:08:18     INFO -        at /builds/slave/test/gaia/node_modules/mocha/lib/runner.js:270:7
22:08:18     INFO -        at done (/builds/slave/test/gaia/node_modules/mocha/lib/runnable.js:185:5)
22:08:18     INFO -        at /builds/slave/test/gaia/node_modules/mocha/lib/runnable.js:199:9
22:08:18     INFO -        at Object.executeHook (/builds/slave/test/gaia/node_modules/marionette-client/lib/marionette/client.js:369:18)
22:08:18     INFO -        at process._tickCallback (node.js:415:13)

Comment 1

[Rudy Lu [:rudyl] (inactive after 2015/7/3)]

Hi John,

This test was pretty stable before.
Do you think this intermittent failure has anything to do with the refactoring in Bug 1044744?

Thanks.

Comment 2

[John Lu [:mnjul][Please NI or I'll miss]]

I'll look into this. Keeping the NI flag to remind myself.

Comment 3

[John Lu [:mnjul][Please NI or I'll miss]]

It's pretty reproducible on my side. Taking this.

Comment 4

[John Lu [:mnjul][Please NI or I'll miss]]

Attachment: Proposed Patch (PR @ GitHub)

Cause: |KeyboardManager.removeKeyboard()| did not invalidate the information stored in |KeyboardManager.showingLayoutInfo|. Thus, when |KeyboardManager.resetShowingKeyboard| calls |InputFrameManager.resetFrame|, an invalid layout information may be passed into it, resulting in the javascript error.

Added a |resetShowingLayout| in removeKeybaord to mitigate this.

The non-invalidation had actually existed before bug 1044744. In previous versions, if you dump |frame.dataset.frameManifestURL| in |KeyboardManager.resetKeyboardFrame|, you can see that the operations are actually being acted on a defunct frame and those operations seem to only fail silently.

What's really strange is that this failing test seemed to always fail on my local machine, and, judging from the cause of the failure, should not be intermittent at all. Dunno why my original bug 1044744 patch at TBPL didn't go wrong.

Anyway, let's have the patch tested.

Comment 5

[John Lu [:mnjul][Please NI or I'll miss]]

Comment on attachment 8471417 [details] [review]
Proposed Patch (PR @ GitHub)

TBPL seems happy with this patch. Tim, could you review this? Thanks.

Comment 6

[John Lu [:mnjul][Please NI or I'll miss]]

Master: https://github.com/mozilla-b2g/gaia/commit/409833a50b449339599d4071a637ef8f099fe2ab