ltoscano added inline comments. INLINE COMMENTS
> kossebau wrote in kaboutdata.h:314 > Why should it happen before? (sorry for being stubborn and curious here, I > want to make sure to understand all code, to then also write proper warnings > in the dox when needed, so people know they should not create it before). > > The thing I understood Albert to point to yesterday was KMainWindow > auto-filling the translator info in the application metadata (see > KMainWindowPrivate::init(...), > https://cgit.kde.org/kxmlgui.git/tree/src/kmainwindow.cpp#n239). But we also > found that this actually has nothing to do with the Q*App instance vs. > KAboutData instance (besides KMainWindow expecting the QApp instance to exist > and KApplicationData::setAppData() having been called). > The other mention was the TODO added by kf5/convert-kcmdlineargs.pl, > Sadly the commit which added that TODO does it only as side-effect and > without any reasoning: > https://cgit.kde.org/kde-dev-scripts.git/commit/kf5/convert-kcmdlineargs.pl?id=ac74bc07de810ef5a51a3030bfa96980f72ad8b1 > And from all code path I have seen there is nothing in the KAboutData > constructor which requires a QApp instance. And the related i18n calls also > are said nowhere said to require a QApp instance. > @dfaure The above being your commit, do you remember (he, only 2 years ago ;) > ) your motivation for that TODO? To put it straight: I don't know (and I can't investigate properly now). But if you create the KAboutData instance before the call to QApplication, part of the "About <foo>" dialog is untranslated. REPOSITORY R244 KCoreAddons REVISION DETAIL https://phabricator.kde.org/D5439 To: kossebau, #frameworks, stikonas, mpyne, aacid, ltoscano Cc: dfaure