Am Freitag, 16. Oktober 2015, 21:54:50 schrieb René J.V. Bertin: > On Friday October 16 2015 01:17:50 Friedrich W. H. Kossebau wrote: > > On my system with gcc version 5.1.1 it seems that does not result in a > > Not, I've never seen this kind of error with gcc. Nor very often with clang, > fortunately... > > A workaround/fix might be to make KoSharedLoadingData a normally exported > > class with implementation of constructor/destructor inside libflake? > > It would seem so. I've had to do a similar fix with KDE PIM's ktimetracker; > there, the issue was not a mysterious error message in the logs, but > downright failure of a dynamic_cast.
By that error I assume it fails to do the dynamic_cast in Calligra as well, due to missing typeinfo? Just that the unsucessful cast might be error- handled, so does not result in a crash. Your patch not tested, but looks fine to me, especially this brings KoSharedLoadingData now on par with KoSharedSavingData :) Please commit to calligra/2.9 and master. Can cherry-pick to master myself, if you do not have that branch present. Cheers Friedrich _______________________________________________ calligra-devel mailing list calligra-devel@kde.org https://mail.kde.org/mailman/listinfo/calligra-devel
