> On Feb 14, 2016, at 12:51 PM, Lubomir I. Ivanov <[email protected]> wrote: > > On 14 February 2016 at 22:26, Dirk Hohndel <[email protected]> wrote: >> On Sun, Feb 14, 2016 at 10:46:15AM +0200, Willem Ferguson wrote: >>> I have many crashes. Happens before the dive list is displayed. Launches >>> successfully about 20% of the time. Apart from that, same symptoms observed >>> by others. >> >> Do you have another logcat with a crash or two in it? That's really more >> useful than just a report of "I have many crashes". >> >> I poked around in the function that was shown top of stack and nothing >> jumped out at me :-( >> >> What's odd is that it works one out of five times. I hate >> non-deterministic crashes... >> > > people seem to be using the website fabric.io and the Crashlytics SDK > for user crash reports on Android. > my understanding is that said SDK integrates code into one's > application (with advertised minimal overhead), so that once the Java > VM or NDK code crashes a report is sent to the fabric.io associated > account. > this means that even if the user has not reported the crash the > developers can examine it. > > someone posted a Qt tutorial: > https://confluence.polibrasnet.com.br:8443/display/PDC/Integrate+Qt+Android+Application+with+Crashlytics+Tools > > perhaps someone from the KDE / Qt-mobile community has experience with this.
I'd be rather worried that this will add third party tracking code into the binaries. The whole thing (if I read through the build process) seems quite hacky. So I'm not convinced that this is a good idea. That said, better analysis of crashes for the native code would be useful. Not sure. /D _______________________________________________ subsurface mailing list [email protected] http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
