ubuntu-phone team mailing list archive

[Łukasz 'sil2100' Zemczak <lukasz.zemczak@xxxxxxxxxxxxx>]

Hi everyone,

On 14.03.2014 05:09, Leo Arias wrote:
> On Thu, Mar 13, 2014 at 7:59 PM, Ted Gould <ted@xxxxxxxxxx> wrote:
> 
>>  On Thu, 2014-03-13 at 16:16 -0600, Leo Arias wrote:
>>
>> On Thu, Mar 13, 2014 at 11:43 AM, Łukasz 'sil2100' Zemczak <
>> lukasz.zemczak@xxxxxxxxxxxxx> wrote:
>>
>> Hi,
>>
>>
>> On 12.03.2014 18:00, Didier Roche wrote:
>>
>>   We might need to have some view on what's seen on the screen during the
>> test-run in the lab... Right now it seems that url-dispatcher works
>> correctly, starts the test-app correctly but autopilot cannot detect it
>> as running. Debugging continues.
>>
>>
>>
>>  Is there an alternate way to get the name of the focused application in
>> unity8?
>>
>>
>> Yes, you can use the Window Stack DBus interface exported by Unity:
>>
>>
>> http://bazaar.launchpad.net/~indicator-applet-developers/hud/trunk.14.04/view/head:/data/com.canonical.Unity.WindowStack.xml
>>
>> You can call GetWindowStack and the first entry will be on top.
>>
> 
> 
> Thanks Ted!
> 
> I've written an alternate version that doesn't use autopilot to make the
> assertions.
> https://code.launchpad.net/~elopio/unity8/alternate_focused_window/+merge/210956
> 
> If this new test also fails, then the problem is not autopilot, it should
> be in unity or in url dispatcher (or on one of the helpers I used). If it
> passes, then the error should be either in autopilot or in the unity QML
> properties.
> 
> Both tests pass on my mako with the most recent devel-proposed image.
> 
> Łukasz, can you run my branch on the lab? And as you said, of course it
> would be really useful to get a video of what's happening on the screen
> during the test.
> 
> pura vida.
> 

Thanks guys for looking into that! Before running the branch you
mentioned, I asked Para to spin the unity8 test using Leo's earlier
refactoring branch - the one that is essentially fixing the FIXME in the
url-dispatcher test code.

https://code.launchpad.net/~elopio/unity8/fake_app_from_toolkit/+merge/208002

Not sure why, but it seems that using this branch the test no longer
fails. My understanding was that both the current approach and the UITK
do it in the same way. I wonder what could be the reason for that. When
looking at the UITK FakeApplication code I see some small differences,
mostly on how the desktop file is created - but all of those seem
cosmetic. And we even checked yesterday that both the .qml file and the
.desktop file created during the failing test have correct contents.

I'm puzzled - maybe I didn't look into it enough. But at least from the
testing on the smoketesting devices we know we can get rid of the
failure by getting that branch landed.

Best regards,

-- 
Łukasz 'sil2100' Zemczak
 lukasz.zemczak@xxxxxxxxxxxxx
 www.canonical.com