Do not let sendMotionEventsToFocused() steal pointer focus from
non-window surfaces such as waybar when follow_mouse is disabled.
This keeps keyboard focus changes from workspace switches from moving
pointer focus off the bar under the cursor, which otherwise makes
workspace scrolling on waybar behave one event behind.
* fix scrolling view move when a grouped window is closed - prevent redundant space assignment for a grouped window that's about to be unmapped, add a `FOCUS_REASON` for unmapping grouped windows
* fix focus_fallback after unmapping a window floating moving view
* fix view change when changing workspaces (special and normal).
add an optional `RECALCULATE_REASON` parameter to `recalculate()` calls of all algorithms.
add an optional `RECALCULATE_MONITOR_REASON` parameter to `recalculateMonitor()` calls.
add 2 new `FOCUS_REASON`s, for workspace change and special workspace toggle
* Fix hyprctl keyword causing scrolling view move.
rename INPUT_MODE_KB to INPUT_MODE_HARD since it seems to have little to do with keybinds
* fix fullscreen/maximise moving the scrolling view
* fix opening of layer 3 programs moving scrolling view
* refactor, clean up and clang-format
* fix bad decisions
* fix includes, clang-format.
* remove TODO
* `focuswindow` no longer moves scrolling view when `follow_focus = 0`
* fix: changes in a group's current window no longer casues scrolling view shift when `follow_focus = 0`
* fix: `moveintogroup`, `movewindoworgroup` no longer causes scrolling view change if `follow_focus = 0`
* clang-format
* add `RECALCULATE_REASON_UNKNOWN`, `RECALCULATE_MONITOR_REASON_UNKNOWN`, `REMOVE_FROM_GROUP_REASON_UNKNOWN`
remove optional wrapper from their enums, refactor code to work with the change
* dispatches that call `switchtoWindow()` in CKeybindManager no longer cause scrolling view to move if `follow_focus = 0`.
They still cause the move if `follow_focus = 1`
Dispatches effected by this: `movefocus`, `cyclenext`, `focuscurrentorlast`, `focuscurrentorurgent`
* Fix for Lua: switchToWondow
* Fix for Lua: focus({window})
* Fix for Lua: Move Window Into Group
* clang-format
* Drop `hyprctl keyword` change
* test: `focus({window})` with `follow_focus = false`
* test: `focus({window})` with `follow_focus = true`
* test: test Focus Fallback
* test: Focus Fallback with groups
* test: workspace change
* test: special workspace toggle
* test: workspace change - special to special
also minor refactoring
* test: close window in a group
* test: move window into group `follow_focus = false`
minor fix in test: expect -> assert
* test: move window into group `follow_focus = true`
* test: new layer
* test: maximise
fix in test: expect to assert.
* test: fullscreen
* test: `dispatch window.focus({ direction })` when `follow_focus = false`
* test: `dispatch window.focus({ direction })` when `follow_focus = true`
test-fix: expect -> assert
* test: move focus within group when `follow_focus = false`
* test: move focus within group when `follow_focus = true`
* test-fix: kill all layers before killing all windows in layers test. prevents internal post test cleanup function failing to kill all layers
test-fix: maximized and fullscreen swapped between their tests
test-fix: testScrollingViewBehaviourMoveFocusInGroupFollowFocusTrue missing layout initialization
test-fix: testScrollingViewBehaviourWorkspaceChange, testScrollingViewBehaviourCloseWindowInGroup have stale window class in a comment
clang-format
* catch up with main: move to new test function
* clean-up the clean-up part of the tests: redundant
* clang-format
* pain
* suffering: fix rebase fail
* test: manually kill all layers since the cleanup function fails
* remove test: fullscreen and maximise. scrolling has its own behaviour now
* fix fp16 blur with invalidation
* restore tonemap code
* fix fade outs
* add sdr compat for unaware surfaces
* use a better work space, not linear, only for sdr
* fix hl tf -> wp tf
* cache workbuffer descriptions
* linear not normalised WIP
* clear offload fb
* CM math
* fix includes
* use precalculated cm for shadows
* inner glow cm
* quad cm
* fix unmodified copy
* codegen
* cache color CM
* mark blur dirty when CM/DS changes
* avoid icc workbuffer
* fp16 blur finish hack
* fix output hdr luminances
* use srgb for mirroring/sharing with fp16
* defaults people
---------
Co-authored-by: Vaxry <vaxry@vaxry.net>
* render: scale background to monitor resolution
Currently the background is not scaled, e.g. 8k size image always uses
~128Mb of VRAM.
We could reduce the consumption for low-end devices and monitors with
lower resolution:
- ~8Mb for 1080p
- ~16Mb for 2k
- ~32Mb for 4k
* render: use gpu to scale bgtex and fix missing reset on size change
* tests: skip pointer tests when pointer input is non-functional
* tests: skip pointer tests in CI due to unreliable cursor behavior
* tests: skip pointer tests when no reliable input environment is available
* tests: skip pointer tests when pointer behavior is unreliable
* tests: temporarily disable pointer tests due to unstable CI environment
* tests: enforce deterministic pointer behavior (flat accel + fixed sensitivity)
* tests: temporarily disable pointer tests due to unstable CI environment
The protocol seems to allow this, because it explicitly mentions either
session lock surface or compositor blanking the output in this sentence:
> The locked event must not be sent until a new "locked" frame
(either from a session lock surface or the compositor blanking the output)
has been presented on all outputs and no security sensitive normal/unlocked
content is possibly visible.
Since `locked_screen_delay` is capped to 5 seconds,
after the 5 seconds this timer is registered for, we assume that
all outputs are covered either by lockedead or by an actual lock frame.
Thus sending locked is ok.
Virt
Vaxry
littleblack111
UjinT34
Matteo Golinelli
Maxim Andreev
Vaxry
erstarr
phanium
Mihai Fufezan
Nakaeri
ImperishableSecret
ArchSavandMio Argillander
fazzi
davc0n
Visal Vijay
Lichie
Maximilian Seidler
Kyler Clay