v7.9 crashes on startup

[GoogleCodeExporter]

Good morning,

up to now I was (! am ...) using v7.7 which is working well apart from the 
known bugs.

What steps will reproduce the problem?
1. Download the new apk from Google drive and install it (both versions of v7.9 
I have downloaded from Google drive so far have the same problem)
2. Start it
3. During a longer pause the message "Radiobeacon ist angehalten" appears 
before the main screen of Radiobeacon is shown.

I tried it with the old databases\radiobeacon and databases\radiobeacon-journal 
 of v7.7 and I tried it with reinstalling radiobeacon.apk v7.9 after I deleted 
those two files. It doesn't make a difference.

On what operating system?
Samsung Galaxy S2 (GT9100) with rooted stock kernel and exposed framework 
installed.

Michael

Original issue reported on code.google.com by ms2618...@gmail.com on 18 Dec 2013 at 6:34

[GoogleCodeExporter]

Hey Michael,

thanks for reporting.

I discovered some ANR after starting a new scan, but that was due to some 
problem in the wifi catalog database. In that case an wifi catalog update 
(Settings --> Download wifi catalog) solves the problem.

But from your description I understand that you get an ANR on the start screen 
(the sessions list)?

Original comment by wish7code on 18 Dec 2013 at 10:41

• Changed state: Accepted

[GoogleCodeExporter]

Hi,

the ANR happens before the start screen or any other message of the radiobeacon 
app  appears.
Updating the WIFI catalog didn't solve the problem.

Michael

Original comment by ms2618...@gmail.com on 19 Dec 2013 at 5:52

[GoogleCodeExporter]

Thanks for clarification. Could you provide a logcat dump (e.g. using 
https://play.google.com/store/apps/details?id=com.nolanlawson.logcat&hl=de)?

I tried to reproduce the isse, but haven't found anything suspious so far.
I installed 0.7.9 on my old LG GT540 phone to rule out memory issues (Android 
2.3, 256 MB RAM), but seems to be fine. I also tried to install 0.7.7 from 
F-Droid and the migrate to the 0.7.9 binaries to check if database migration 
might fail. But again, this is fine...

Really strange..

Original comment by wish7code on 19 Dec 2013 at 9:02

[GoogleCodeExporter]

Hi,

with some delay -but here it is.

M

Original comment by ms2618...@gmail.com on 21 Dec 2013 at 10:06

Attachments:

• 2013-12-20-06-39-52.txt

[GoogleCodeExporter]

The issue has been identified (a missing column in database migrations in fact) 
and will be fixed with next release soon after Christmas.. 

I'll keep you updated..

Original comment by wish7code on 23 Dec 2013 at 12:26

• Changed state: Started

[GoogleCodeExporter]

Hey Michael,

[https://googledrive.com/host/0B97gHr4MqjHpM2h0QVR5SWJOcGs binaries] have been 
updated.

If you prefer to build your own binaries, check out r758a2667cfda

Cheers
Toby

Original comment by wish7code on 26 Dec 2013 at 7:04

[GoogleCodeExporter]

Good morning Toby,

the build is better but not perfect ... now the ANR happens later:

I installed the update over my old installation (v7.7).
The start screen of Radiobeacon appears as it should.
But then if I restart my old session (see logcat attached, it happens at 
21:26:42) or try to start a new session the application crashes. 

Cheers 

Michael

Original comment by ms2618...@gmail.com on 27 Dec 2013 at 6:40

Attachments:

• 2013-12-26-21-26-26.txt

[GoogleCodeExporter]

Already learnt a lot about database migrations pitfalls:-)

Here we go with some new 
[https://googledrive.com/host/0B97gHr4MqjHpM2h0QVR5SWJOcGs/ binaries] or 
re569140e44af respectively..

Basically some database migrations were ignored, suggested solution solved the 
problem: 
http://stackoverflow.com/questions/14419358/confusion-how-does-sqliteopenhelper-
onupgrade-behave-and-together-with-impor

Original comment by wish7code on 27 Dec 2013 at 11:54

[GoogleCodeExporter]

Hi Toby

New version - new issues. As I am not sure whether they are due to the database 
migration I post it here instead of opening a new issue.

The new version starts fine (see screenshot). When I restart the session the 
number of wifis and cells ios not shown until after a gps fix has happened (see 
screenshot).

After the fix the operator name of the cellular is shown as "DEFACE" instead as 
"E-Plus" as it was in earlier versions. New cells are added to the database 
with operator "DEFACE" instead of "E-plus".

When I switch between OVERVIEW, WIFIS, CELLS and MAP it takes quite a time and 
after switching several times Radiobeacon crashes (see attached catlogs).

Cheers Michael

Original comment by ms2618...@gmail.com on 29 Dec 2013 at 7:37

Attachments:

• Screenshot_2013-12-28-20-39-15.png
• Screenshot_2013-12-28-20-39-23.png
• 2013-12-28-08-37-04.txt
• 2013-12-28-08-33-44.txt

[GoogleCodeExporter]

Hey Michael,

> When I restart the session the number of wifis and cells ios not shown until 
after a gps fix has happened (see screenshot).

Fixed with ra40cdfeda135 

> After the fix the operator name of the cellular is shown as "DEFACE" instead 
as "E-Plus" as it was in earlier versions. 

That's surprising.. the cell logger itself didn't change and I didn't find much 
while googling. Nevertheless when searching the string 'DEFACE', I stumbled 
upon an interesting discussion thread: 
http://www.heise.de/security/news/foren/S-Re-Wer-Root-hat-benutzt-Xposed-XPrivac
y/forum-270936/msg-24483539/read/ Are you using Xposed and Xprivacy?

I'm not familiar with Xposed: do you have the option to switch it off 
temporarily, to check whether this is somehow related?

> When I switch between OVERVIEW, WIFIS, CELLS and MAP it takes quite a time 
and after switching several times Radiobeacon crashes (see attached catlogs).

I'll have to investigate this in more detail. It looks like a problem with the 
UI lifecycle, but I wasn't yet able to reproduce the crash.

As always thx for reporting.. 
Cheers
Toby

Original comment by wish7code on 30 Dec 2013 at 4:17

[GoogleCodeExporter]

Hi Toby,

"That's surprising.. the cell logger itself didn't change and I didn't find 
much while googling. Nevertheless when searching the string 'DEFACE', I 
stumbled upon an interesting discussion thread: 
http://www.heise.de/security/news/foren/S-Re-Wer-Root-hat-benutzt-Xposed-XPrivac
y/forum-270936/msg-24483539/read/ Are you using Xposed and Xprivacy?"

Sorry, my fault. Yes, I am using X-Privacy and up to a few days ago the 
radiobeacon client had no restrictions. But with all the uüdates of X-Privacy 
and Radiobeacon I must have made a checkmark on the wrong place ... After I 
unticked the box it's working again.
Will the wrong entries be overwritten automatically when there are new 
measurements?

I wish you all the best for 2014.

Cheers 

Michael

Original comment by ms2618...@gmail.com on 31 Dec 2013 at 6:49

[GoogleCodeExporter]

All the best for you too!!

> Will the wrong entries be overwritten automatically when there are new 
measurements?

Unfortunately in case X-Privacy is working properly it probably hasn't masked 
the cell provider only (DEFACE instead of E-Plus) but also the complete rest of 
other data, namely the cell id. So these measurements are probably lost.

Neverthless maybe we can still rescue some parts of your tracks. You could copy 
the database to your desktop pc and send it to me for further examination..

Open a command line prompt and enter
adb shell
sudo
chmod 777 /data/data/org.openbmap/databases/radiobeacon
exit
adb pull /data/data/org.openbmap/databases/radiobeacon

and you'll get the database copied to your desktop pc..

PS: After some tests I can also confirm the performance issues you mentioned. 
Although the Nexus 4 doesn't crash, the performance becomes very bad on large 
tracks (in my case 7900 wifis + 800 cells). Will be fixed ASAP..

Cheers 
Toby 

Original comment by wish7code on 2 Jan 2014 at 7:39

[GoogleCodeExporter]

Hi Toby

as I didn't hear from you after I had sent you the database file I had a
closer look at it by myself. I think I can manage the DEFACE problem as
XPRIVACY didn't do a perfect job. The cid has been read out in most cases
despite of XPRIVACY being active ... In most cases I have (or will have in
the near future) correct entries of the particular cells without the "help"
of XPRIVACY. Therefore I think I can change the data manually if the
recorded positions are ok (I suppose this doesn't interfere with the way
the number of recorded cells is counted by radiobeacon).

But the other questions are still open:

a) How  can I import the cell backup files of opensignalmap sessions to the
local radiobeacon database? (I sent you a small example file as private
mail)
b) How can I upload the cell and wifi backup files of opensignalmap
sessions to openbmap? (I am not very interested in the wifi data, so I
don't want to keep them on the phone to avoid more performance issues)?

c u

Michael

Original comment by ms2618...@gmail.com on 13 Jan 2014 at 5:59

[GoogleCodeExporter]

Sorry Michael, I have been very busy lately and didn't reply:-(

> I suppose this doesn't interfere with the way the number of recorded cells is 
counted by radiobeacon

No, the number of cells is there for information purposes only, so no problem..

> How  can I import the cell backup files of opensignalmap sessions to the 
local radiobeacon database? (I sent you a small example file as private mail)

The file itself looks good: it has all the relevant cell info, namely cell type 
(e.g. UMTS), cell id, lac, mcc + mnc and psc and of course gps. The difficulty 
comes from radiobeacon's relational database structure, i.e there are separate 
tables for scans, gps positions and cells. In other words, bringing the 
opensignalmap data to the LOCAL database is rather complicated, as you would 
need to add a entry in each single table.

Nevertheless I could write a small converter to include it in the SERVER 
database. It should then be just a couple of python lines only, as the server 
database structure is more easy to handle.

I would like to start that, when I finished work on the server side.
Keep the files for the moment, we'll find a solution :-)

Cheers 
Toby

Original comment by wish7code on 13 Jan 2014 at 8:38