https://bugs.kde.org/show_bug.cgi?id=524780
--- Comment #10 from [email protected] --- Thanks a lot for testing the patch, Ralf! The Qt5 result is exactly what I would expect: with Qt5, QMap::operator[] const returns a const copy, so the expensive detach never happens and the patch cannot make anything faster there. It's good to know that it doesn't make anything worse either. The problem only shows up in Qt6 builds. KMyMoney builds against Qt5 by default (BUILD_WITH_QT6 is OFF in cmake/modules/QtVersionOption.cmake), while the Flatpak manifest sets -DBUILD_WITH_QT6=ON. So all Flatpak builds (Flathub 5.2.2 and the nightly) are affected. With a build configured with -DBUILD_WITH_QT6=ON you should see the ~60 s for R000183 without the patch and ~16 s with it. Your standalone Qt6 comparison only covered QTextBrowser, so it could not show this part. Would you be willing to apply the patch, or include it in a merge request yourself? Reintroducing QWebEngine via Alkimia sounds like a good solution for the remaining QTextBrowser part. I'd be happy to test it. -- You are receiving this mail because: You are watching all bug changes.
