On Monday 16 February 2015 13:22:30 Rex Dieter wrote: > Sune Vuorela wrote: > > On 2015-02-16, Rex Dieter <[email protected]> wrote: > >> * QT_BUILD_KEY handling > >> > >> Similar to webkit above, the configure bits about QT_BUILD_KEY only > >> handles up to gcc-4.x. > >> > >> For fedora 22, our gcc5 builds are done in a way that is (supposedly) ABI > >> compatible to gcc4, so I'm currently using: > >> http://pkgs.fedoraproject.org/cgit/qt.git/tree/qt-gcc5_compat_qt_build_ke > >> y.patch to ensure QT_BUILD_KEY matches (ie, it includes the same g++-4 > >> string as before). > >> > >> As far as upstream is concerned, I assume you'd prefer the default here > >> to be g++-5 ? > > > > The build key is used for ABI checks, so I think the 4 should still be > > used. e.g. you can build a qt plugin with g++-5 and it should still be > > loadable by the qt you buil with g++-4.x > > > > So I think your current patch is right. > > > >From my reading of > > http://developerblog.redhat.com/2015/02/10/gcc-5-in-fedora/ > > It may depend on the value of _GLIBCXX_USE_CXX11_ABI macro, but I admittedly > don't yet fully grasp the implications of it being set either way.
As far as I understand, _GLIBCXX_USE_CXX11_ABI only makes a difference in the ABI of the standard library. (libstdc++) Since the Qt ABI should not contains anything from the standard library, this macro should not have an impact on the binary compatibility with Qt. -- Olivier Woboq - Qt services and support - http://woboq.com - http://code.woboq.org _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
