ubuntu-phone team mailing list archive

[Ricardo Salveti de Araujo <ricardo.salveti@xxxxxxxxxxxxx>]

On Fri, Feb 28, 2014 at 3:57 AM, Ricardo Salveti de Araujo
<ricardo.salveti@xxxxxxxxxxxxx> wrote:
> On Thu, Feb 27, 2014 at 2:48 PM, Didier Roche <didrocks@xxxxxxxxxx> wrote:
>> Hey,
>>
>> We got more infos since yesterday and have an action plan. No image promoted
>> (and not for the foreseeable future), but we'll enable resume partial
>> landings.
>>
>> I/ First, the image #210 was produced, containing:
>> * dialer-app with the autopilot test fix from Chris (and the dashboard
>> confirms the flakyness being gone)
>> * the pulseaudio fix is in, and we see that pulseaudio is really idling now
>> and not taking 2% of one core. Nice work!
>> * bluetooth-touch to get bluetooth activation on flo
>> * various indicator-datetime crash fixes for the dekstop
>>
>> We still got some desktop-only landings.
>>
>> II/ Remaining issues:
>>
>> So, most of the flakyness and crashes are identified as being the same
>> source than the qmlscene crash we discussed some weeks ago. It's just more
>> randomly happening now. This crash doesn't seem to be easily fixable and is
>> in the Qt 5.0 V8 stack. Qt 5.2 (with the use of their new V4 QML specific
>> javascript engine) won't have that issue.
>>
>> The other listed issues are:
>>
>> 1. Voice call not working properly after the first call (Ricardo/David H.)
>> David was pinged on that one and looking with Ricardo is now focusing on
>> that one.
>
> Fix in proposed, needs a new image once in release (android and pulseaudio).
>
>> 2. Camera freezes after taking photo on Nexus 4 (which is making the camera
>> AP tests failing) (Ricardo/Sergio)
>> Sergio is working on that one. Ricardo will join force on that one.

Still in progress, but at I already understood what is happening. The
freeze is basically a mutex lock when there's a buffer to be dequeued,
but the app/aal layer ignores it when taking pictures (it works only
when the buffer is dequeued right before you hit the button to take a
picture).

I still need to investigate a bit more to see what can be changed in
the app/aal layer to make sure everything is properly dequeued before
moving forward, but unfortunately I'm not that familiar with the app
still.

Hope to have more progress during the day.

On a different topic, our team also landed a new ofono fixing quite a
few bugs and also removing the python 2 dependency, but it seems we're
getting quite a few crashes at the dashboard. When tested locally
(manually and with autopilot), everything worked nicely, so I wonder
if this could be a side effect of the phonesim setup steps that were
changed over the last few days.

Tony, can you give a hand to see if this can be fixed today still?
Otherwise it might just be easier to revert the package instead.

Cheers,
-- 
Ricardo Salveti de Araujo