https://bugs.kde.org/show_bug.cgi?id=525665
--- Comment #13 from Branislav Klocok <[email protected]> --- I can confirm this with the minimal example from comment 0 and the unmodified 0.7.0.1 package, without rebuilding anything: * QML_DISK_CACHE=aot-bytecode: all handles work, no loop warnings * QML_DISK_CACHE=aot: broken as before, 57 loop warnings from the same three qrc: locations * QML_DISK_CACHE=aot-native: all handles work, no warnings The only difference between the first two runs is whether the native code generated ahead of time is used, so the breakage comes from that native code and not from the bytecode or the QML source. The aot-native result looks contradictory, but my guess is that without bytecode the compiled units from the library are not used at all and everything is compiled at runtime. I have not verified that. Together with the rebuilt 0.6.2.1 breaking against Qt 6.11.2, this points at what qmlcachegen generates for these bindings rather than at a change in KQuickImageEditor. I have not tried building with --only-bytecode, so I cannot say yet whether that works as a workaround on the library side. Tested with kquickimageeditor6-imports 0.7.0.1 and Qt 6.11.2 on openSUSE Tumbleweed. -- You are receiving this mail because: You are watching all bug changes.
