881273 - Marionette tests that use wifi are failing after flashing 2.0

Status: RESOLVED FIXED

(Remote Protocol :: Marionette, defect)

Description

[Florin Strugariu [:Bebe]]

STR:
1. Flash Unagi with latest FF OS
2. Run https://github.com/mozilla/gaia-ui-tests/blob/master/gaiatest/tests/unit/test_connect_to_local_area_network.py

Actual:
The test if failing with:
https://gist.github.com/bebef1987/5749168#file-trace

you can find the logcat at https://gist.github.com/bebef1987/5749168#file-logcat

the build id is from https://pvtbuilds.mozilla.org/pub/mozilla.org/b2g/nightly/mozilla-central-unagi-eng/latest/unagi.zip 10-06-2012

if we run the test with a breakpoint at https://github.com/mozilla/gaia-ui-tests/blob/master/gaiatest/gaia_test.py#L486 the test starts passing  sporadically

Comment 1

[Florin Strugariu [:Bebe]]

This might not be a isolated issue with the wifi.

I think (not sure) after running some more tests that self.marionette.execute_async_script might have some issues and because of that all the cool JS that we run with it might fail.

What's interesting is that we have the same stack trace in the rest of the tests

JavascriptException: TypeError: backgroundColor is undefined
stacktrace:
@app://system.gaiamobile.org/js/window_manager.js, line 469

Comment 2

[Jonathan Griffin (:jgriffin)]

This either means some of the gaia-ui-tests code is using gaia's window manager incorrectly for 1.2, or gaia's window manager is itself working incorrectly, and we just happen to be catching that error in our window.onerror handler.

Comment 3

[Zac C (:zac)]

Dave researched this a bit earlier and thinks it is an actual gaia regression and in github we could see a fix committed already.

We're going to try again with a fresh releng build tomorrow and see how it goes.

Comment 4

[Jonathan Griffin (:jgriffin)]

I think this has been resolved as a gaia issue.