General and Desktop

@SGaist Thank you for replying! I found the problem. I had a problem importing VTK-related dependency libraries. After I re-imported the related libraries, the function returned to normal.✌️

@Thaxila said in QtWebEngineWidgets not showing checkboxes properly: I am now trying compiling the source code with Mingw64 on Windows to see if that will get rid of the bugs we see on Windows. Since QtWebEngine (actually the underlying chromium engine) only builds under MSVC, not MinGW, under Windows this is a wasted effort....

@Mitchal-Dichter said in Is there a simple way to make a triangle in Qt3D?: working directly with Direct3D, Metal, and Vulkan and skipping Qt Even though this is a Qt Forum here and Qt is in fact a great framework, sometimes you better look for something else. In your case, if you plan to render everything directly, you could also pick an immediate GUI, like ImGUI together with a Vulkan (on mac through MoltenVK), D3D or OpenGL backend. Fee free to post some updates on this or showcase your approach later :)

Thank you I'll go the inotify way. I just need to check that "commits" to the settings file are atomic so I won't be reading garbage (I read somewhere that QSettings handles well that).

I have read somewhere that Qt installed through pip is independent from the Qt installed through your system packages. Duplicate libs+plugins+styles etc ... so your program built against the pip installed Qt won't have access to the Qt installed by your distro vendor.

i got it event->modifiers() & Qt::ShiftModifier

@Jammin44fm said in Problem running application on Linux. Ubuntu 22.04 LTS + Wayland platform.: I've decided to go with just the wayland-egl platform, as that is what is suggested by the Qt docs as being the ideal target platform. Where did you read this? In case it was in this page, do note that it is about making a Wayland compositor with Qt - not a client application (and talks about the EGL hardware integration plugin, which is only tangentially related to the Wayland-EGL platform plugin). Applications that just run on a Linux desktop against a generic compositor (i.e KDE/Sway/GNOME/etc.) and don't have a particular reason to care about which protocol exactly is used to manage the shared graphics buffers - should use the generic Wayland platform plugin. EGL used to have some support issues around certain compositors and particularly nVidia drivers, so it makes sense for it to be something you explicitly opt into if you insist on using it. The fact this works when i specify the platform to use, indicates that most things are in place, I'm just stumped as to why it wont use the wayland-egl by default!! That's not how it works with platform plugins. Qt won't just load the first one that happens to be available/working - many of the plugins shipped with Qt are for specialized purposes; it would quite confusing indeed if an application would just proceed to load the minimal plugin by default if the i.e. the xcb plugin is missing or broken! Essentially, if no platform plugins are explicitly requested via command line or environment variable, there is a compiled in default that says what to use. On generic Linux this is xcb , and if Wayland is detected to be running then also wayland (which is provided by libqwayland-generic.so). As to why wayland-egl isn't tried by default as well, see the above.

@SGaist Ya I've done plenty of digging over there. Maybe it helped because I finally did manage to make some progress (kind of). The ninja apidoc command now fails due to a number of exceptions like the one below. I couldn't come across any discussion of paths being misrepresented like this so I'm guessing I made a careless mistake somewhere - just not sure how to go about determining where exactly without a never ending trial/error. qt.shiboken.doc: (doc) While handling <snippet> in PySide6.QtXml.QDomNode, line 4: Failed to open file "\Qt\6.9.1\Src:C:\Users\Nicholas\Packages\BuildQt\pyside-setup\build\.venv\build\pyside6\doc\base\codesnippets:\Qt\6.9.1\Src\qtbase\src\xml\doc\snippets\code\src_xml_dom_qdom.py" for reading: The filename, directory name, or volume label syntax is incorrect.)

@pehg hi, While taking a quick look at the header file mentioned in the article I stumbled upon this issue on the llvm project that might explain the situation. I think you don't have __cpp_lib_three_way_comparison defined with clang 18.

You may wish to download the source code for DeepSkyStacker 5.1.0 which was first released step in the journey from MFC to Qt (5.1.0 is a hybrid). You can DL the source code from here: https://github.com/deepskystacker/DSS/archive/refs/tags/5.1.0.zip. The file of interest is DeepSkyStacker\DeepSkyStacker.cpp HtH David

@The-Force said in QT 5.15 Windows Crash on deallocation of QModelIndexList: __acrt_first_block == header ...usually points at out-of-bounds access. if (!list.empty() || ui->classList->model()->rowCount() == 0) ui->shipCombo->setCurrentIndex(list.first().row()); If list is empty and the model's row count is zero, list.first() causes a read out of bounds and is probably the reason for the crash. Something like if (!list.empty() && ui->classList->model()->rowCount() != 0) ui->shipCombo->setCurrentIndex(list.first().row()); ...would make more sense to me.

@SGaist 6.9.1 + vs2022

@Navadvipa-Chandra-das said in Not compilated Qt from sources: failed because the system could not detect libzstd.dll ... Although the instructions for installing Qt do not mention the zlib library as required for installation! That's strange; I never needed libzstd or zlib to build Qt on Windows. Zstandard and zlib are optional components. If you don't have them, your config.summary file should contain something like this: Support enabled for: Using pkg-config ....................... no Using vcpkg ............................ no udev ................................... no OpenSSL ................................ yes Qt directly linked to OpenSSL ........ no OpenSSL 1.1 ............................ no OpenSSL 3.0 ............................ yes Using system zlib ...................... no Zstandard support ...................... no Thread support ......................... yes

@schrute What I found to be effective, is to leave it as a bordered window, and just intercept the WM_NCCALCSIZE message, so that it does nothing. This basically gives you a bordered window, where you stop windows from drawing the borders. This way, when you implement your own title bar, aero snap will still work.

I will not look at qml but maybe it's a good starting point for others who want to debug this.

I found these Natvis visualizers to be working fine with Visual Studio Code, if that helps you: https://github.com/narnaud/natvis4qt/tree/main/natvis QJson value visualizers are only included in the qt6.natvis file.

I fixed this issue. It was a Qt 6.9.0 bug, but Qt 6.9.1 has the fix

@jawadulhassan Hi Thank you for the hint, the setModal(true) and setWindowModality(Qt::ApplicationModal) has no effect. the activateWindow() where do you mean I should locate it? If I see it right, the opening of the editor is done in the QAbstractItemView. Do you have a sample which fixes this issue?

The solution is qt6-help-devel