Installation and Deployment

Also happens in 6.6.3

// Run windeployqt // --> <PropertyGroup> <Cmd><![CDATA["$(QtToolsPath)\windeployqt.exe"]]></Cmd> <Cmd>$(Cmd) --list target</Cmd> <Cmd>$(Cmd) @(Args->'%(Identity)',' ')</Cmd> <Cmd>$(Cmd) @(PosArgs->'%(Identity)',' ')</Cmd> </PropertyGroup> <Message Importance="high" Text="Qt Deploy: $(Cmd)" /> <Exec Command="$(Cmd) > &quot;$(QtDeployLog)&quot;" EnvironmentVariables="$(QtDeployEnv)" /> <!-- /////////////////////////////////////////////////////////////////////////////////////////////// // Read deploy log into QtDeployed list // --> <ReadLinesFromFile File="$(QtDeployLog)"> <Output TaskParameter="Lines" ItemName="QtDeployed" /> </ReadLinesFromFile> <ItemGroup> <QtDeployed Remove="@(QtDeployed)" Condition="!Exists('%(Fullpath)')"/> </ItemGroup>

This part in the qtdeploy.targets is the cause.
Apparently the first few outputs of windeployqt.exe are lines like this:

Adding in plugin type generic for module: Qt6Gui Adding in plugin type iconengines for module: Qt6Gui ...

before outputting one .dll/.qm file per line.

<ItemGroup> <QtDeployed Remove="@(QtDeployed)" Condition="!Exists('%(Fullpath)')"/> </ItemGroup>

Since this part cant resolve these lines to paths, it throws the error.
That's the cause, but I'm also not good enough at Visual Studio/Qt magic to work around this.
Is there a way to inject a step into this targets file to filter the first few lines? Or just skip that step?

Edit: turns out the "--list target" parameter for windeployqt is specifically there to stop those
Edit: https://bugreports.qt.io/browse/QTBUG-122257 has been reported and fixed as of 6.8 apparently, so I guess the only fix is going to a newer Qt Version.

Okay, I have now built a minimal QtWebengineCore and QtCore app. This also does not run, and I receive no feedback when I call it from the console. However, the environment is suitable for a large QtGui app.

I have copied the sysroots directory into my system where Qt compilation environment is set up. Below is the only file available in the sysroots directory.

tisdk/build/arago-tmp-default-glibc/sysroots/am335x-evm/imgdata/tisdk-default-image.env

Note: I am using a am335x-evm custom board.

Please suggest how to include the sysroots directory into the compilation process through command line arguments. Also, let me know what the contents of the sysroots directory should look like.

This solves any display related issues from within docker containers
docker run -it --name $CONTAINER_NAME --network host -e DISPLAY -e TERM -e QT_X11_NO_MITSHM=1 $IMAGE_NAME bash

I am seeing similar if not the same issuing using Qt 6.7.3, the missing files are:

libunwind.dll libc++.dll

Copying those files from C:\Qt\6.7.3\llvm-mingw_64\bin fixes it.

@Christian-Ehrlicher: With Ubuntu's cross compiler "arm-linux-gnueabi-..." a build was possible, not yet tested on target.
Thanks!

Came across the same issue today with Qt 6.7.2 on macOS Ventura when building my app with GCC 14.2.0 (in C++23 mode). That is, because Qt was built with Apple Clang (14.0.0) which does not set _GLIBCXX_USE_CXX11_ABI=1.

Inline methods (like QString::toStdString, QVector::toStdVector, etc...) are fine because they are compiled with the same C++ ABI as used in the application.

That does no longer seems to be the case, because QByteArray::toStdString() is not defined in its header.

Just for reference, I get the following linker errors:

Undefined symbols for architecture x86_64: "QByteArray::toStdString[abi:cxx11]() const", referenced from: QString::toStdString[abi:cxx11]() const in main.cpp.o ld: symbol(s) not found for architecture x86_64

Hi,

In addition to what @Axel-Spoerl wrote: are you using the Windows Subsystem for Linux ?
Since you mention KStars, they have a Windows build so why not use that one directly ?

@mdebellis said in Finding a Kit for Basic Tutorial:

I want to use QT with Python.

In that case you should use PyQt (https://wiki.python.org/moin/PyQt) or PySide (https://wiki.qt.io/Qt_for_Python).
The online installer you used is for C++ development.

@davidwedel
So it looks like maybe someone was playing with Qt and switched default debugging off and that's how it got released with this file in it, even if no Qt! :) Naughty Fedora people!

What did you set to change the output directory of your libs?
I'm having the same issue, can't change the output directory of the qt_generate_deploy_app_script. When I tried by setting CMAKE_INSTALL_LIBDIR, I get RPATH errors and can't make the package successfully.

is it there any incompatibility between Jetson Nano Hardware/Sofware and qtWebView? (aarch64+Ubuntu 18.04)

Thanks for your answers.

@Scarab said in Static Qt Error: Qt version is not properly installed, please run make install:

"pass the relocatable flag to configure". I don't know how to do that either, there is no configure field in the Qt versions tab

"configure" is the command you execute before you build Qt. You need to pass relocatable flag to configure before you actually build Qt.

I think you just have to keep trying different mirrors. Are you claiming every one of them returns Forbidden? That would be different.

Hi and welcome to devnet,

You should take a look at the content of the tests run by the configure script to see what exactly failed with regard to EGLFS.

The NVIDIA forum might also contain some hints.

Then explicitly call the version of Python you want to use when executing your script.

As for conda VS venv, it highly depends on your goal.
You might also want to consider using something like uv that will help you manage your dependencies and environnement.

Since you are starting fresh, I would recommend testing uv.

Hi,

What does ~file qmake6 return ?

dnf -y update this morning.

Skipped:
qt5-qtbase-5.15.15-1.fc40.x86_64 qt5-qtwebengine-freeworld-5.15.17-1.fc40.x86_64

When configure Qt, did you add the following arguments?
It seems that the package won't be built unless you explicitly use them.
-eglfs -qpa wayland-egl -qpa xcb