Installation and Deployment

it is likely you are using wayland, not X11 Switch from Wayland to Xorg in Ubuntu At the login screen click on the cog icon beside the “Sign In” button. Select the option “Ubuntu on Xorg.” Enter your password and log in to your Ubuntu machine.

@Nikolay_spb said in qt.qpa.plugin: Could not load the Qt platform plugin "xcb" in: yes, i have desktop manager Would be nice to mention which one. Is X server running or do you use something else (like Wayland)?

@szerwi please uninstall using control panel

Ok thanks that worked.

@raven-worx thanks, I was able to compile! Out of curiosity, is there a script for Linux to export all the relevant environment variables like in Windows (the "Qt Console") ? I spent a couple of hours to figure out why it cannot find the tools (CMake and Ninja). Then I had to export their paths and finally it compiled.

@jsulm so far I tried to build a simple QtWidget "Hello world" application on my dev machine. Then I copied it to the target with this structure: $ tree . ├── test_qt6 │   └── bin │   ├── test_qt6 │   └── platforms │   └── libqeglfs.so ├── qt6_libs │ ├── libicudata.so │ ├── libicudata.so.56 │ ├──libicudata.so.56.1 │ ├──libicui18n.so │ ├──libicui18n.so.56 │ ├──libicui18n.so.56.1 │ ├── but running the application with: LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/user/qt6_libs /home/user/test_qt6/bin/./test_qt6 -platform eglfs it still searches for x11: ./test_qt6: error while loading shared libraries: libxkbcommon.so.0: cannot open shared object file: No such file or directory ps. sorry but the editor loses the indentation. The platform plugin file is actually inside the platforms directory UPDATE I tried to copy the missing library, now ldd finds all. But still does not run: EGLFS: Failed to open /dev/fb0 EGLFS: Can't continue without a display I'm working to check the permissions.... fixed. $ export QT_QPA_PLATFORM=eglfs $ export QT_QPA_EGLFS_FB=/dev/fb0 $ LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/user/qt6_libs ./test_qt6 -platform eglfs Could not initialize egl display Aborted (core dumped)

It's including your host headers instead of ones coming from the toolchain. Maybe try using absolute path in your sysroot flag: -sysroot /home/yourUserName/rpi/sysroot

What I just found out: If I start/restart nmake module-qtwebengine several times (after the last build failed as stated above) the process fails always at a different point (concernint the list of read files). Has anyone got an idea what this means? Thank you, R.

@Shaoning-Di said in Qt Cmake configuration has no path to a C++ compiler set: but it didn't work What exactly did not work?

Thank you, jsulm! Although I was able to start downloading, after a few seconds the connection was dropped and now it gives 503 Service Temporarily Unavailable error on this site :( Attila

@codeNinja95 Sorry but with QT installer framework I never found as to install vc runtime. But if you use the last installer of Postgresql it installs vc runtime.

There is a Qt plugin for VS. This can import Qt projects and handle most of the stuff automatically. I personally prefer to use qmake.exe -tp vc MyProject.pro to generate a VS project file. If you want to be portable (Linux/macOS) I would say it is the preferred way to always make changes to the .pro file (adding/deleting files) and regenerate the VS project from it. I believe (I am not a great fan of the Qt plugin for VS and don't use it) that the Qt plugin does all the stuff necessary for the project file. But, it does not have the Qt Designer built in. You need to use Qt's own tools to still edit UI-files. If ICC does not show up inside Qt Creator automatically, you can add it by hand and then add it to a kit. I am not sure how well that is supported. Is there any specific reason you want to use ICC? A while back (I believe it was ICC v14) I did some benchmarks and ICC did not optimize OO code. It was slightly faster with regular C code, but no meaningful optimizations for C++. Only this year did they start to use clang as backend and I would expect that this version of the current ICC could actually be better. If you don't make sure to use the clang-variant of ICC and don't have any other specific reasons to use it (performance at least wasn't a reason a while back), just use one of the default compilers on Windows.

Move python to a path without spaces

No, xcb will not work on windows. Tunneling X11 does not mean that the xcb stuff is done one your windows system but on your linux host.

Thanks for tip : Solved by : sudo apt install libopengl0 -y Can open now. I appreciate your help