[Kde-bindings] SIGSEGV when running Qyoto apps?

Arno Rehn arno at arnorehn.de
Sun Feb 21 18:10:41 UTC 2010

On Sunday 21 February 2010 15:28:04 IBBoard wrote:
> I've been using Qyoto for a while now and am getting used to the
> signals/slots way of doing things. Generally everything is okay, and the
> link of MonoDevelop to Qt Creator is usable, but I do seem to be having
> some stability issues (which was one of my first concerns after comments
> from Erik/FireRabbit). Before I filed a bug I wanted to check whether
> there were any obvious causes I might be missing.
> Basically, after an indeterminate amount of time (sometimes in the
> middle of executing a menu action, sometimes while resizing panels,
> sometimes while just moving the mouse over the window) my app crashes
> and prints out:
> The application was terminated by a signal: SIGSEGV
> SIGSEGV is apparently a segfault from accessing incorrect memory areas.
> Given that my app is .Net and hence "managed" code then I wouldn't have
> thought it could cause it directly. I've tried a static reference to the
> main window in case that helped [1], with no luck. I've also tried using
> all the standard RPMs from openSUSE (mono-qt 4.3.1 with libqt4 4.5.3) as
> well as with newer versions of Qt (4.6.0).
> Are there any known problems that could cause this? Or any good ways to
> get some more information to determine if it is just versioning issues?
That seems like a problem in Qyoto, yes. First thing to try would be qyoto (or 
mono-qt as opensuse seems to call it) 4.4.0, because that's the current 
release. A proper backtrace would be nice, too (-debug packages should be 
available, I hope). If everything fails (or even before :)), create a small 
test application that exhibits this behaviour or point us to your code so we 
can test it ourselves (a small test app would be nicer, though ;)).

Quick guess: Mono's GC is again interfering with Qt's way of cleaning up 
objects. Maybe look if an QObject doesn't have a parent set and set one, if 
necessary. Dunno if that's the cause, though.

Arno Rehn
arno at arnorehn.de

More information about the Kde-bindings mailing list