Crash Mobile build 758 on Android 5.1.1

[Rick Walsh]

On 31 Jan 2016 20:54, "Willem Ferguson" <willemferguson at zoology.up.ac.za>
wrote:
>
> On 31/01/2016 10:49, Rick Walsh wrote:
>>
>>
>>
>> On 31 January 2016 at 19:37, Willem Ferguson <
willemferguson at zoology.up.ac.za> wrote:
>>>
>>> Rick,  unfortunately both of the two experimental builds crash on my
phone.
>>> Kind regards
>>> Willem
>>
>> That's not all bad - it should mean that the splash screen isn't the
problem.
>>
>> I've created and uploaded another, subsurface-mobile-e9e843d.apk
>>
>> This should be one commit after 675 (you last good one).  So if it
doesn't work e9e843d should be the offending commit, it it works, f466ee6
will be the problem.
>>
>> I've also created one for 1030cb2, which should be the same as 675, but
built on my machine.  Please test it if e9e843d fails.
>>
>> Rick
>>
>>
>>>
>>> On 31 Jan 2016 07:37, "Rick Walsh" <rickmwalsh at gmail.com> wrote:
>>>>
>>>>
>>>>
>>>> On 31 January 2016 at 15:23, Willem Ferguson <
willemferguson at zoology.up.ac.za> wrote:
>>>>>
>>>>> On 31/01/2016 06:04, Willem Ferguson wrote:
>>>>>>
>>>>>> These are the builds that I tried:
>>>>>>
>>>>>> 525 ok
>>>>>> 675 ok
>>>>>>
>>>>> 680 crashes
>>>>>
>>>>>
>>>> If my sums are correct, that tracks it to one of:
>>>> 15a0141 QML UI: Don't explicitly set width of GridLayout items
>>>> 88101e4 QML UI: don't shorten the buttons
>>>> 04e994b Subsurface-mobile on Android: change the splash screen
methodology
>>>> f466ee6 Don't blindly copy a pointer to the heap
>>>> e9e843d Close edit drawer when current dive changes
>>>>
>>>> I struggle to make sense of the logcat.  Does the splash screen
display for you?  If not, it might have been caused by 04e994b
'Subsurface-mobile on Android: change the splash screen methodology'
>>>>
>>>> I created packages for 04e994b and f466ee6, and uploaded them to the
same link.  Can you test them?  As I created them, they're signed with a
debug key, rather than Dirk's key, and built with qt5.5.1 (not 5.6 that
Dirk uses).
>>>>
>>>> Cheers,
>>>>
>>>> Rick
>>>>
> Rick, puzzling results. 675 definitely work (did that twice to eliminate
finger trouble). 1030cb2 does NOT work (also tested twice). The other two,
(e9e843 & f466ee6) consequently also do not work. The program starts up,
shows splash screen, goes through email/password protocol, spends quite a
bit of time loading the log from the cloud, and then crashes, consistent
with the suspicion that this deals with the gridlayout involved in
displaying the dive list. Do not understand why 675 works and 1030cb2 not.

That is interesting. Either there's something your phone doesn't like about
my builds (work fine on my phone), or I was wrong about 675 being commit
1030cb2. I assumed 758 was built from the curreny git master and I counted
back. But maybe 758 was a few commits earlier, in which case I think 675
would be a few commits before 1030cb2. Dirk can you confirm?

R
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.subsurface-divelog.org/pipermail/subsurface/attachments/20160131/b6e2f137/attachment-0001.html>