kbuildsycoca5 crashing

[Jeremy Whiting]

I only built kf5/kde applications. Though I've no idea what compiler
was used to build the Qt packages. probably g++ 4.9 from arch stable I
bet. /me removes testing, downgrades the dozen packages from it and
will rebuild. Thanks for the tips both of you. Hopefully I can finish
the 15.04.1 release tuesday from within a plasma 5 session instead of
this fallback gnome session :) (and have konversation running again
also...)

thanks,
Jeremy

On Sun, May 10, 2015 at 9:46 AM, Albert Astals Cid <aacid at kde.org> wrote:
> El Diumenge, 10 de maig de 2015, a les 08:57:28, Jeremy Whiting va escriure:
>> Also, none of the recent patches on
>> https://projects.archlinux.org/svntogit/packages.git/log/trunk?h=packages/qt
>> 5 make me nervous, all seem to be pretty straightforward. Could it be lower
>> level as in my glibc is too new, or the compiler itself (GCC 5.1.0)
>
> I hear nothing else than scary stories about people mixing stuff compiled with
> the new gcc and older versions, there's supposedly a switch that makes the abi
> be compatible, but the switch seems to be failing at properly make things
> compatible.
>
> I'd suggest to either recompile everything or to downgrade to old gcc.
>
> Cheers,
>   Albert
>
>>
>> On Sun, May 10, 2015 at 8:54 AM, Jeremy Whiting <jpwhiting at kde.org> wrote:
>> > I see. Some kde applications segfault (klettres, okular) other's run
>> > ok (dolphin, khangman) I haven't tried others. As for qt binaries I
>> > checked the binaries that come with the arch package, and assistant,
>> > designer, qdbusviewer all work fine. Is there one in particular I
>> > should check?
>> >
>> > On Sun, May 10, 2015 at 8:45 AM, David Faure <faure at kde.org> wrote:
>> >> On Sunday 10 May 2015 08:24:48 Jeremy Whiting wrote:
>> >>> Rebuilt kservice (and nuked my build folder and /usr/local just to be
>> >>> doubly sure.) kbuildsycoca5 still crashes with this same backtrace, no
>> >>> idea why. Could it be a bug in QtCore? Arch testing packages of Qt?
>> >>
>> >> I doubt it. More likely a compiler bug or a BIC or a miscompile.
>> >>
>> >> Try running a few Qt examples from your Qt build, to at least make sure
>> >> that works.
>> >>
>> >>> 0xffffffff is -1, no? did it get decreased one too many times somehow?
>> >>
>> >> You don't understand. This isn't the value of the refcount, it's the
>> >> address in memory of the variable holding the refcount.
>> >>
>> >> Which means a serious mixup somewhere lowlevel (e.g. compiler).
>> >>
>> >> --
>> >> David Faure, faure at kde.org, http://www.davidfaure.fr
>> >> Working on KDE Frameworks 5
>>
>> _______________________________________________
>> Kde-frameworks-devel mailing list
>> Kde-frameworks-devel at kde.org
>> https://mail.kde.org/mailman/listinfo/kde-frameworks-devel
>