GitLab

Better metadata, better life

NO_CHANGELOG

CCBUG: 454756

Luca Carlon authored May 27, 2022 and Nate Graham committed May 27, 2022

See discussion in plasma-framework!500.

CCBUG: 411462
CCBUG: 332512

There's no currentSerial, this was presumably supposed to be lastInputSerial

- use QCOMPARE rather than == so that if this fails, we know
  how many activities there were rather than just "not two".

The test tries to start Weston and skips when it is not found

486d86bc broke this by verifying whether the executable is found

This caused the test to fail instead of skip when Weston is not installed

It doesn't really do anything useful, but is needed to build KIO

Otherwise consumers that expect it to be async break

BUG: 450342

See:
https://kde.org/info/security/advisory-20220131-1.txt
https://mail.kde.org/pipermail/kde-devel/2022-January/000943.html

Instead of /usr/include/KF5/kwindowsystem_version.h.

For more details see:
kservice!79

GIT_SILENT

Xlib functions which work with keyboard mappings have a (Display) internal
cache, which needs to be refreshed on mapping changes by calling
X(kb)RefreshKeyboardMapping on change notify events.

After Qt switched to XCB, that is no longer happening, and it's also not
possible to get to the XEvent needed for triggering the refresh. Switch
over to using the XCB equivalent and just load the symbols on every invocation.

BUG: 426684

There is already a dummy plugin that is loaded, if it is not possible to
locate a matching platform plugin. The only difference was, that the
Wayland platform plugin always returned true for compositingActive.
This is now done by the dummy plugin to not break anything.

This is more or less the same as desktopFileName, but made by GTK

GIT_SILENT

NO_CHANGELOG

Left-over from 328fe001

NO_CHANGELOG

Can't use 'TARGET Qt5::X11Extras' because it my be undefined at the time
this file is used, instead use Qt6::Gui which will be set by the find_dependency
call above.

NO_CHANGELOG

That method takes a char32_t, unlike its non-static namesake which only takes
char16_t and has an assert that checks the arg is <= '0xffff', and the unit
test was hitting that assert.

Thanks to Fabian Vogt for figuring it out.