General and Desktop

Oh yes, thank you. That was a total thinking failure. What I actually wanted to test was something like that: { Nix nix4(4, "Hello, I'm four"); *nix1 = nix4; } qWarning() << "nix1: " << nix1->display(4); And that works. *nix2 = *nix1; delete nix1 works too Thanks, solved!

@shavera said in Is there a canonical way to set up QApplication and Google Test together?: After many iterations, I've finally found what I think works well here: int main(int argc, char *argv[]) { QCoreApplication app{argc, argv}; QTimer::singleShot(0, [&]() { ::testing::InitGoogleTest(&argc, argv); auto testResult = RUN_ALL_TESTS(); app.exit(testResult); }); return app.exec(); } it works for me

@nyaruko said in Qt warning of type conversion already registered: I have a multi-thread Qt program... This is a common cause of issues of all flavours. It may be as simple as the same setup code for shared configuration, e.g. the metatype database, executing in multiple threads. @ocgltd said in Qt warning of type conversion already registered: Any way to track this down? You could provide a minimal program that invokes the message, some basic Qt version information, tool chain information, etc. Something that happens twice in ten years here and spans many Qt versions is not likely to have an immediate, "I've seen that before," response.

@Christian-Ehrlicher Thank you, I figured it out, using opt->rect's x and y attributes draws it in the correct place.

You are not linking against the library you are testing.

workaround: Add Qt::NoTitleBarBackgroundHint window flag

@SGaist I am on windows.

@Mark81 It is "difficult" because you do not have a proper foreign key relation, which is what QSqlRelationalTableModel requires. Having some other column (eventType) whose value dictates which of multiple tables (alarms, sessions, ...) is the foreign one in which to look up the key column's value is supported neither in SQL nor Qt's QSqlRelationalTableModel --- it is not allowed in the definition of a foreign key. Either think about redefining/rearchitecting your intended relationships which fits better with SQL or (in my opinion) give up trying to use QSqlRelationalTableModel. Instead read in the contents of all required/linked tables into their own normal QSqlTableModels, as well as the original "source" table, and implement your desired logic yourself in client code as required for your desired behaviour.

@nicker-player Set a stylesheet?!

I still don't see any relation to Qt here. We already told you more than once where you can get such random numbers from. Qt does not provide such things (not needed and to hard to implement it properly - there are enough good alternatives around)

@Pl45m4 What setup are you using? MacOS 15.1 Apple clang Qt 6.7 QApplication::font() returns QFont(.AppleSystemUIFont,13,-1,5,400,0,0,0,0,0,0,0,0,0,0,1).

@PPatel said in std::thread crash on ubuntu 20.04: statement std::thread t1(this). In addition to @SGaist suggestion, what is this here? Does it have anything to do with Qt? I am not a std::thread user but I must confess I do not understand which of its constructor overloads you are even invoking here (unless your this is an existing std::thread).

Bug posted as https://bugreports.qt.io/browse/QTBUG-133522

@Christian-Ehrlicher Thanks Christian!

@Christian-Ehrlicher I did. From what I understand it seems it run the code you highlighted before: [image: a224c421-8a50-4f97-acaf-76e810198b8d.png] but still no focus rect: [image: f3dd347b-9bfb-4294-ac34-9be208037db9.png] To further digging the issue I placed a QCheckBox on a form and I managed to make the focus rect appeared: [image: 889e0be2-00b8-4887-ad51-a9a69c53d13c.png] then I placed a breakpoint to inspect the properties of the state variable: QStyle::State_Enabled | QStyle::State_On | QStyle::State_HasFocus | QStyle::State_Active | QStyle::State_KeyboardFocusChange and as I've done before, I added them to my delegate: if (isChecked) opt.state.setFlag(QStyle::State_On); else opt.state.setFlag(QStyle::State_Off); if (_wdPress == i) opt.state.setFlag(QStyle::State_Sunken); opt.state.setFlag(QStyle::State_Enabled); opt.state.setFlag(QStyle::State_HasFocus); opt.state.setFlag(QStyle::State_Active); opt.state.setFlag(QStyle::State_KeyboardFocusChange); and now it works. But it works even removing the Enabled and Active state. And I clearly tried these flags before, as I wrote above. I have no time to reinstall Qt 6.8.1 to test again, but it seems something has changed in 6.8.2.

Hi and welcome to devnet, How did you install Qt ? Do you have cmake installed ?

@developer23 said in Include Qt6 libs in CMake project on macOS: In CMake I include Headers and nested folders and then compiler emit This sounds wrong. You should only tell CMake which Qt modules you need, like: find_package(Qt6 REQUIRED COMPONENTS Core) target_link_libraries(mytarget PRIVATE Qt6::Core) Also, there is no need to have module names in includes: #include <QChar>

@jsulm Yes, as mentioned in the link.

@JonB said in QSqlTableModel setFilter => Unable to execute statement: ? Escaping the column names and doing everything more thoughtfully would have prevented the crash... so "range" instead of range etc... I think not only any select statements were influenced but the whole table itself. That's why the DHCP service didn't start up anymore ;)