In service processes, `get_desktop_window` only creates a fake window
owned by the first calling thread, with only a handle and no client-side
object.
Before 9d6455f91cb7a7d166d5c4708d528b43a63eb4d7 `get_window_thread` was
returning an error (called from `send_message` > `process_message`) in
that case, as client-side object is missing.
After the commit, we only check handle existence and we now return
success and then try to send/process the desktop window message, waiting
for explorer startup which never happens.
Fixes: 9d6455f91cb7a7d166d5c4708d528b43a63eb4d7
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/8046
These patches fix painting in an application that draws outside of a CS_PARENTDC child
client area in its WM_PAINT handler.
Comments and suggestions are welcome.
--
v5: win32u: Don't clip update region to the window client rectangle.
win32u: GetUpdateRect() should clip update rectangle to the window client area.
win32u: GetUpdateRgn() should clip update region to the window client area.
win32u: Clip PAINTSTRUCT.rcPaint to the window client area.
server: Don't generate WM_PAINT messages for minimized windows.
server: For a CS_PARENTDC child use parent for visible region calculations.
server: If the being validated region covers whole window then validate everything.
https://gitlab.winehq.org/wine/wine/-/merge_requests/5665
Supersedes !7928 with some test tweaks:
* Test that shutdown waits for any queued handler,
* Use message loop only when necessary (DQTYPE_THREAD_CURRENT),
* Remove some non-deterministic refcount checks after shutdown, as dispatch thread might still hold some until it exits.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/8055
First push with null patch, to get current gitlab runner result in [#42287](https://gitlab.winehq.org/bernhardu/wine/-/pipelines/42287).
--
v2: timeout/tests: Ignore ctrl-c in the parent process.
https://gitlab.winehq.org/wine/wine/-/merge_requests/8051