Investigate effort/pros/cons of discontinuing UWP support in Qt 6 [Spike]

Note: This report is merely for gaining an estimate of the required work and potential benefits/drawbacks of removing UWP support. It does not imply any decision.

Risks:

• Reduce list of supported platforms
  • Windows Phone: dead
  • UWP on desktop Windows: How many users?
  • Xbox One: any users?
  • HoloLens: Hard to say
  • Windows IoT devices: What are MS's plans going forward with their IoT offering? Do they want to concentrate on Azure or do they plan devices?

Benefits:

• Reduced maintenance:
  • qmake/build system adaptions (manifests, packaging)
  • Maintenance of ANGLE's UWP functionality
  • Native winrt network backend (sockets, hostinfo, ssl) can be removed
  • Native winrt font database implementation can be removed
  • winrt qpa plugin can be removed
  • native uiautomation implementation (accessibility) can be removed
  • IFDEFery in qtbase in various places
    • core application
    • file system/path handling
    • threading
    • event/timer handling
    • shared memory
    • various locations in widgets
  • Rather big outstanding issue: main thread != UI thread. Kind of worked around, but not fixed
  • winrtrunner no longer needed
• Auto tests:
  • Significantly reduced CI integration times
    • qtbase takes around 1 h without winrt, winrt takes around 2 hours for a full test run
  • IFDEFery in auto tests, checking for winrt
    • Some of these should check, if tests are running in a sandboxed environment instead

Informational:

• Backends created for UWP can be used on desktop Windows (>= 10). These backends will not be dropped. Examples:
  • Sensors
  • Location
  • Bluetooth (LE)

Estimated workload: 5d