← Back to team overview

ubuntu-phone team mailing list archive

Re: Landing team 28.05.14

 

* Łukasz 'sil2100' Zemczak <lukasz.zemczak@xxxxxxxxxxxxx> wrote:
> Image #52 looked rather nice dogfooding wise (although some new 
> issues have been spotted) so we *might* promote #53

I wouldn't suggest promoting #53.  I ran into a nasty crash loop 
on it.

> * Links to auto-generated commit-logs of recent images:
> (Host might change in the nearest time)
> http://people.canonical.com/~lzemczak/landing-team/52.commitlog

Awesome.  It's a nice improvement over the logs we had before.  
:)

> * Known issues:
> 
> ** Pictures taken with flash are almost completely black (Bill)
> https://bugs.launchpad.net/camera-app/+bug/1322121
>  -> Bisecting and problem identification in progress - camera app might
> be rewritten soon

I tried this again, and it worked.  I think it involves a race 
condition that I didn't trigger on the first try.

However, after taking pictures and trying to open the gallery 
app, unity8 restarted.  And then in a later session (~30 minutes 
later?), I tried to run the camera app again to re-test, and got 
only a white screen (no crash, it just didn't start).  On a third 
try, it crashed unity8 on startup.

On my fourth try, the camera started successfully but then a 
picture with flash gave me a very very dark image (despite the 
post-pic preview looking bright and overexposed).

I've also noticed that, even when the flash works, the post-pic 
shot review has a completely different white balance and exposure 
than what the saved pic shows in the gallery.

After viewing the fourth result in gallery, I turned the screen 
off and then it turned itself back on to show me unity8 crashing 
again.  Then it did the same thing again a couple minutes later, 
and again a couple minutes after that.  I wasn't even using it at 
the time.

A quick look at "top" shows that qmlscene is pegging the CPU over 
100%, and so is maalit-server.  I tried killing those manually, 
and unity8 restarted again about 10 seconds later.  And, despite 
not having anything using up much CPU time, it then crashed again 
after a couple minutes.

Next up, I tried a reboot from the device's root shell.  This 
resulted in a black screen with the backlight on, not actually 
finishing a boot or responding to adb commands.

After about 10 minutes I gave up and did a hard reset, which got 
it back to a stable shell, at least while idle.


Otherwise, I noticed some other things in #53:

  - When receiving a call from someone in my contact list, the 
    person's name is displayed as "Unknown caller" if the screen 
    was off when the phone received the call.  If the screen was 
    on though, the correct name is displayed.  Not sure if this 
    is a new issue.

  - During a call, the loudspeaker mode significantly reduces the 
    volume of transmitted sound, making it harder for the person 
    on the other end to hear.  It actually transmits better using 
    non-loudspeaker mode, even from across the room.  (however, 
    without loudspeaker, playback for received audio is much 
    quieter so it's not practical to use the normal mode)  As far 
    as I can tell, this is because it uses a different hardware 
    mic for the two modes.

  - Media labels in the music and video scopes took a few seconds 
    to display, instead of the usual instant reaction.  However, 
    this might be related to the high CPU use I saw, since it was 
    snappy after a hard reset.

  - The "location detection" indicator stays checked now.  
    However, this doesn't actually make location services work, 
    and the GPS toggle still turns itself off.


-- Selene


Follow ups

References