Run the attached test and you'll see this output:

QDEBUG : tst_Untitled::testCase1() Key_Shift ("1000020", "") pressed with modifiersQFlags<Qt::KeyboardModifier>(ShiftModifier)
QDEBUG : tst_Untitled::testCase1() Key_Shift ("1000020", "") released with modifiersQFlags<Qt::KeyboardModifier>(ShiftModifier)

Then make the following changes

    QTest::qWait(1000);

//    QTest::keyPress(window, Qt::Key_Shift);
//    QTest::keyRelease(window, Qt::Key_Shift);

and click the shift key once before the test finishes, and you'll see this output:

QDEBUG : tst_Untitled::testCase1() Key_Shift ("1000020", "") pressed with modifiersQFlags<Qt::KeyboardModifier>(ShiftModifier)
QDEBUG : tst_Untitled::testCase1() Key_Shift ("1000020", "") released with modifiersQFlags<Qt::KeyboardModifier>(NoModifier)

This difference (along with the fact that shift event handling/behaviour doesn't seem to be documented) makes it impossible to test shift events in an application: if I handle modifiers one way, the test will fail, and if I do it the other way, the actual application won't work as expected.

The only documentation I can find is https://doc.qt.io/qt-5/qkeyevent.html#modifiers, and it says:

Returns the keyboard modifier flags that existed immediately after the event occurred.

So is Qt Test wrong?