This commit is a follow up to 7bf1521363,
ensuring that if a user has changed global padding options, that they
will be preserved from the initialized window manager state and applied
on top of the dumped state which is being restored.
This commit introduces an if let binding to only process windows which
still exist when attempting to enforce workspace rules.
Previously, calls to functions such as Window::exe might have returned
an error if a window which had been destroyed but not yet removed from
the state was examined by the enforce_workspace_rules fn. Now, such
windows will fail the if let binding and be skipped entirely, eventually
being removed by the core event processing loop.
This commit adds an early exit from the border manager's event
processing loop whenever a window which still exists in the state but
has been destroyed is encountered. Instead of returning an error, the
'containers loop will now skip ahead to the next iteration.
This commit also makes an adjustment to the frequency with which the
reaper sends border manager notifications - a single notification is now
sent at the end of each iteration if necessary, rather than one
notification per workspace.
If a user triggers the workspace reconciliator by clicking on an app in
the start bar or via alt-tab, a notification should be sent to
subscribers such as komorebi-bar so that the focused workspace can be
updated.
The various komorebi reconciliators and manager modules don't emit
events to subscribers themselves (yet?), so for now we can pass on the
uncloak event.
Maybe we can look into expanding the Notification enum in the future.
fix#1211
This commit adds mutex lock scoping in
WindowManager::enforce_workspace_rule to avoid a deadlock when
should_update_focused_workspace evaluates to true.
fix#1212
This commit adds a new komorebic command "eager-focus", which takes a
full case-sensitive exe identifier as an argument. When komorebi
receives this message, it will look through each monitor and workspace
for the first matching managed window and then focus it.
This allows users who have well defined workspaces and rules to bind
semantic hotkeys to commands like "komorebic eager-focus Discord.exe" to
immediately jump to applications instead of mentally looking up their
assigned workspaces or positions within container stacks.
This commit adds a new field to the static config file, "remove_titlebar_applications", which allows
users to now use the full range of matching strategies to identify applications for which titlebars
should be removed. This is heavily discouraged for a number of reasons, and is unlikely to work with
a wide range of applications which now draw their own titlebar regions. The previous advice to use
in-application configuration settings to hide title bars if they exist is still valid.
resolve#805
This commit adds the following new socket messages and commands:
- `EnforceWorkspaceRules`: resets the `already_moved_window_handles` and
calls `enforce_workspace_rules` so that all workspace rules, including
initial workspace rules are applied again
- `enforce-workspace-rules`: cli command which sends the
EnforceWorkspaceRules socket message
This commit creates a new `SocketMessage` called `StopIgnoreRestore`
which makes komorebi stop without calling `window.restore()` on all
windows. This way every maximized window will stay maximized once you
start komorebi again and it is able to use the previous `State`.
If it fails to restore the previous state you might have to call
`komorebic restore-windows` in case you had hidden windows, for example
when when using the `window_hiding_behaviour` as `Hide`, or you can
simply unminimize them if you were using `Cloak` or `Minimize`.
This commit makes sure we focus the previously focused workspace on all
monitors, load it and update it and in the end focus the actual focused
monitor and workspace pair calling `update_focused_workspace` to make
sure it updates the workspace and gives focus to the focused window.
This commit adds changes to the main wm process to dump a state file to
temp_dir() when the process is exited either via komorebic stop or
ctrl-c, and to automatically try to reload that dumped state file if it
exists on the next run.
A new flag "--clean-state" has been added to both komorebi.exe and the
komorebic start command to override this behaviour.
The dumped state file can only be applied if the number of connected
monitors matches the number of monitors recorded in the state, and if
every HWND listed in the state file still exists.
This is validated by calling Window.exe(), which under the hood checks
for the continued existence of the process associated with the HWND.
Only the "workspace" subsection of the state for each matching
connecting monitor will be applied.
This commit adds various transparency related global configuration
values to GlobalState, which is can be queried via the komorebic
global-state command.
resolve#1182
This commit makes it so a floating window only has the floating border
when it is focused, if not it has the `Unfocused` border. It also makes
the 'focused_container' have the `Unfocused` border when it is not the
foreground window, for example when we have a floating window focused
instead.
This commit also changes the border's `window_kind` so that the stored
borders actually have that value so we can check it later (This value
wasn't being updated).
This commit also makes it so we properly invalidate the borders in the
situations discussed above (for example when changing focus to/from a
floating window we need the floating window border to update its ZOrder
as well as the previously focused window).
Lastly this commit, changes the `WM_PAINT` code part of the border so
that it now sets the position of border so that the border's ZOrder
updates to it's tracking window ZOrder.
This commit introduces a number of changes to the border manager module
to enable borders to track the movements of windows as they are being
animated.
As part of these changes, the code paths for borders to track user
movement of windows have also been overhauled.
The biggest conceptual change introduced here is borrowed from
@lukeyou05's work on tacky-borders, where the primary event listener of
the komorebi process now forwards EVENT_OBJECT_LOCATIONCHANGE and
EVENT_OBJECT_DESTROY messages from application windows directly on to
their borders.
These events are handled directly in the border window callbacks,
outside of the main border manager module event processing loop.
In order to handle these events more performantly in the border window
callbacks, a number of state trackers have been added to the Border
struct.
When handling EVENT_OBJECT_NAMECHANGE, these values are read directly
from the struct, whereas when handling WM_PAINT, which is sent by the
system whenever we invalidate a border window, we update the state
values on the Border structs from the various atomic configuration
variables in the mod.rs file.
Another trick I borrowed from tacky-borders is to store a pointer to the
Border object alongside a border window whenever it is created with
CreateWindowExW, which can be accessed within the callback as
GWLP_USERDATA.
There is some unfortunate introduction of unsafe code to make this
happen, but the callback uses null checks to exit the callback early to
ensure (to the best of my ability) that there are no pointer
dereferencing issues once we start making border changes in the context
of the callback.
There are a few other Direct2D related optimizations throughout this
commit, mainly avoiding the recreation of objects like brush properties
and brushes.
Finally, the border_z_order option is now deprecated as the border
window is now tracking the z-ordering of the application window it is
associated with by default - this should resolve a whole host of subtle
border z-ordering issues, especially when dragging windows around using
the mouse.
This work would not have been possible without the guidance of
@lukeyou05, so if you like this feature, please make sure you thank him
too!
This commit fixes an issue where when trying to move floating windows or
windows on a floating workspace across boundaries to another monitor
using the `move_container_in_direction` it wouldn't move the floating
windows physically, although it moved them internally on komorebi,
resulting in weird and wrong behavior.
This commit creates a new method on `Monitor` to
`add_container_with_direction` which takes a move direction and then
uses the same logic that was previously on the
`move_container_in_direction` function.
It changes the `move_container_to_monitor` function to take an optional
move direction which if it is some will have this function call the new
method `add_container_with_direction` instead of just `add_container`.
Lastly the `move_container_in_direction` function now when it realizes
the move will be across monitors simply calls the
`move_container_to_monitor` with the direction that was initially given
to it.
These changes require that all callers of `move_container_to_monitor`
add an direction option, instead of passing `None` on all of them, a new
helper function was created, named `direction_from_monitor_idx` which
calculates the direction a move will have from the currently focused
monitor and the target monitor return `None` if they are the same or
returning `Some(direction)` if not. This way now all commands that call
a move across monitor will use the logic to check from the direction if
it should add the container on front or end.
With these changes now all the code related to moving a window across
monitors using a command should be on one place only making sure that in
the future any change required only needs to be done on one place,
instead of having to do it on `move_container_to_monitor` and
`move_container_in_direction` as before!
This commit is an interactive squashed rebase of the following commits
from PR #1154:
8f4bc101bc
fix(wm): move floats in direction across monitors
This commit fixes an issue where when trying to move floating windows or
windows on a floating workspace across boundaries to another monitor
using the `move_container_in_direction` it wouldn't move the floating
windows physically, although it moved them internally on komorebi,
resulting in weird and wrong behavior.
This commit creates a new method on `Monitor` to
`add_container_with_direction` which takes a move direction and then
uses the same logic that was previously on the
`move_container_in_direction` function.
It changes the `move_container_to_monitor` function to take an optional
move direction which if it is some will have this function call the new
method `add_container_with_direction` instead of just `add_container`.
Lastly the `move_container_in_direction` function now when it realizes
the move will be across monitors simply calls the
`move_container_to_monitor` with the direction that was initially given
to it.
These changes require that all callers of `move_container_to_monitor`
add an direction option, instead of passing `None` on all of them, a new
helper function was created, named `direction_from_monitor_idx` which
calculates the direction a move will have from the currently focused
monitor and the target monitor return `None` if they are the same or
returning `Some(direction)` if not. This way now all commands that call
a move across monitor will use the logic to check from the direction if
it should add the container on front or end.
3b20e4b2fe
refactor(wm): use helper function on move to workspace
Use the same `add_container_with_direction` function on
`move_container_to_workspace` as it is being used on
`move_container_to_monitor` or `move_container_in_direction`.
This way we bring parity between all methods and make it easier to
change the way a container is added on a monitor workspace when taking
the move direction into consideration.
83f222fe84
* fix(wm): correctly define moves across monitors
Moves within the same workspace were being considered as moves across
monitors when the workspace was floating (not tiled).
This commit fixes this by changing the way we first define if a move was
across monitor or not.
We now search for the moved window on all workspaces and check if its
monitor index is different from the target monitor index (the monitor
where the move ended).
a02694348e
* fix(wm): ignore moves/resizes on floating workspaces
This commit makes sure that moves or resizes within a floating workspace
(i.e. not tiled) will be ignored, unless the move is across monitors.
We don't care about the positions or sizes of windows within a floating
workspace!
4bf24f81e0
* fix(wm): avoid workspace load on cross monitor moves
This commit replaces the `window_manager.focus_workspace` call with a
`monitor.focus_workspace` which doesn't load the workspace. There is no
need to load the workspace when moving windows across monitors since
those workspaces will already be loaded, we simply need to update them.
Loading the workspace would cause some issues as well, like when moving
a window to a floating workspace which already contained a window that
matched some `floating_windows` rules was always putting the
"floating_window" on top of the window we just moved with a bunch of
focus flickering. This is fixed with this commit.
cb53f463ae
* fix(wm): avoid workspace load on command move across monitor
If the move happens between the already focused workspaces of two
monitors we shouldn't load the workspace, since it is already loaded and
it will cause changes on focused windows, which might result on the
window we just moved not being focused.
When moving maximized floating windows across monitors they were
magically disappearing!
The window would be on the correct place, with the correct coordinates
and size, its styles wouldn't change it would still have the `VISIBLE`
style, however the window was invisible.
If we used the system move to try to move it sometimes we would be able
to see a bar on the top of the monitor and if we moved the window with
the keyboard on the direction of another monitor then the window would
start showing up on that monitor... So it was visible on that monitor
but not on the one we just moved it into.
After some investigation I decided to atribute that behavior to magic,
since I couldn't find any other plausible explanation, if someone knows
about this please tell me, I too would like to learn the ways of this
dark mysteries from the deep of the Windows OS.
On a serious note, this commit creates a workaround for this by simply
unmaximazing the window first (it's not restore, it doesn't change the
size) then it moves the window (if animations are enabled it proceeds to
wait for the animation to finish...), then it maximizes the window
again.
Previously when moving floating windows across monitors we would keep
the size of the window as it was. For most cases this would be ok.
However for users with monitors with completely different sizes this
could result on a window that would fill across monitors when moving
from the bigger monitor to the smaller monitor.
This commit, attempts to resize the windows proportionally to the
monitors' sizes.
There is currently a slight issue with some apps (so far I've only
noticed it on 'Wezterm'...) where if the DPIs across monitors are
different they don't seem to fully get the OS DPI change completely, but
it seems that setting the `Wezterm` compatibility high DPI scaling
override to "System" on the app's executable properties, fixes the
issue.
Since this is only 1 app (so far...) and only when the scales between
monitors are different I decided to commit this anyway.
This will do more good than harm, since in the cases it was misbehaving
with 'Wezterm' the result would be a wrongly resized window that is
still completely visible on the target monitor anyway and the override
fix seems to be good so far.
This commit is comprised of the following interactively rebased commits
from PR #1002 by @thearturca.
1a184a4442
refactor(animation): move animations to its own mod
First step for more rusty version animations. The goal is to make
animations more generic so its easier to add new animations to komorebi!
d3ac6b72c2
refactor(animation): reduce mutex calls on `ANIMATION_STYLE`
8a42b738fe
refactor(animation): introduce `Lerp` trait
e449861c10
refactor(animation): generalized ANIMATION_MANAGER
Instead of a isize key for the ANIMATION_MANAGER HashMap, now we use a
String key. For window move animation, the key would be
`window_move:{hwnd}`.
This allows us to use single manager for more types of animations.
67b2a7a284
feat(animation): introduce `AnimationPrefix` enum
8290f143a6
feat(animation): introduce `RenderDispatcher` trait
2400d757fe
feat(animation): implement window transparency animation
This commit also fixes graceful shutdown of animations by disabling them
before exit and wait for all remaining animations for 20 seconds.
44189d8382
refactor(animation): move generation of `animation key` to `RenderDispatcher`
e502cb3ffb
refactor(animation): rename `animation` mod to `engine`
Linter was upset about this:
> error: module has the same name as its containing module
369107f5e0
feat(config): adds per animation configuration options
Originally static config only allowed global config for animations.
Since this refactor introduces the abilty to add more type of
animations, this change allows us to configure `enabled`, `duration` and
`style` state per animation type.
Now each of them take either the raw value or a JSON object where keys
are the animation types and values are desired config value. Also adds
support for per animation configuration for komorebic commands.
There was an issue where if you changed focus across monitors and the
target monitor had a maximized window it would focus one of the
containers beneath it instead. And if there were no containers it
wouldn't focus anything, but instead keep focus on the previous monitor.
This commit fixes that issue.
This commit ensures that when switching to a workspace which contains a
floating window or a maximized window, either the maximized window or
the first floating window index will be focused.
This is to prevent windows which have been hidden on the previous
workspace from retaining keyboard focus.
fix#1130
This commit adds various widget grouping and transparency options to
komorebi-bar, and is comprised of the individual commits listed below,
worked on in PR #1108, squashed into one.
e8f5952abb
* adding RenderConfig, and some test frames on widgets
0a5e0a4c0a
* no clone
a5a7d6906c
* comment
6a91dd46cd
* ignore unused
80f0214e47
* Group enum, Copy RenderConfig
fbe5e2c1f7
* Group -> Grouping
ce49b433f9
* GroupingConfig
f446a6a45f
* "fmt --check" fix (thanks VS)
d188222be7
* added widget grouping and group module
1008ec2031
* rounding from settings, and apply_on_side
7fff6d29a9
* dereferencing
655e8ce4c1
* AlphaColour, transparency, bar background, more grouping config options
cba0fcd882
* added RoundingConfig
ec5f7dc82d
* handling grouping edge case for komorebi focus window
12117b832b
* changed default values
645c46beb8
* background color using theme color, AlphaColour.to_color32_or, updating json format for Grouping and RoundingConfig
10d2ab21c7
* hot-reload on grouping
d88774328a
* grouping correction on init
2cd237fd0d
* added shadow to grouping, optional width on grouping stroke
4f4b617f26
* grouping on bar, converting AlphaColour from_rgba_unmultiplied, simplified grouping
3808fcec8f
* widget rounding based on grouping, atomic background color, simplified config, style on grouping
be45d14f6d
* renamed Side to Alignment, group spacing
be45d14f6d
* proper widget spacing based on alignment
b43a5bda69
* added widget_spacing to config
c18e5f4dbe
* test commit
cba2b2f7ac
* refactoring of render and grouping, widget spacing WIP
9311cb00ec
* simplify no_spacing
36c267246b
* correct spacing on komorebi and network widgets (WIP)
85a41bf5b2
* correct widget spacing on all widgets
50b49ccf69
* refactoring widget spacing
9ec67ad988
* account for ui item_spacing when setting the widget_spacing
e88a2fd9c0
* format
This commit overhauls the "Komorebi" borders implementation to use
Direct2D, which enables anti-aliasing for rounded borders.
A lot of the heavy lifting was done by @lukeyou05 in the tacky-borders
project, which this commit largely adapts to komorebi. @lukeyou05
provided an incredible amount of guidance and feedback on the
implementation of this feature on the komorebi Discord.
This commit is a squashed interactive rebase of the following commits:
238271a71e
feat(borders): initial impl of direct2d border drawing
5525a382b9
feat(borders): avoid multiple render target creation calls
431970d7b6
feat(borders): reduce redraws to improve perf
47cb19e54a
feat(borders): remove black pixels around direct2d corners
3857d1a46c
feat(borders): clean up render targets on destroy
This commit adds a new komorebi command "cycle-stack-index" which allows
the user to manipulate the index position of the focused window in the
focused stack by swapping it with either the previous or the next window
until the desired index position has been found.
After investigating further the issue where commands would randomly stop
working, we've noticed that the issue seems to be that somehow the
listening thread gets stuck reading the unix socket, as in it
continuously tries to read a socket on a connection that is not sending
anything anymore. The result would be that komorebi would no longer be
able to receive commands until it was restarted.
This fix adds a read timeout of 1s and it spawns a new thread to handle
the stream reading and process of cmds. So in case this happens again,
that specific processing thread will only be stuck for 1s but the rest
of komorebi will never get stuck and should keep working normally.
This change prevents the move animation from playing again for each
window in the stack. Tested with all hiding behaviors. Looks good so
far.
resolve#1029
This commit ensures that when attempting to reload the static
configuration file after the user has imperatively created new
workspaces, the configuration reload logic will not attempt to load a
workspace config for those imperatively created workspaces.
This commit introduces a new komorebic command, close-workspace. This
command will remove the focused workspace from the window manager state
if the following conditions are met:
1. The number of workspaces on the focused monitor are >1
2. The workspace is empty
3. The workspace is unnamed
The third condition is to ensure that we are not removing workspaces
which have been declared in the static configuration file.
This commit ensures that if an IdWithIdentifer without an explicitly set
matching strategy makes it through to should_act_individual, it will be
treated the same as MatchingStrategy::Legacy instead of causing a
runtime panic.
This commit switches all relevant commands to treat the v2
applications.json asc format as the default format in all commands.
The v1 applications.yaml file will still be processed correctly if
passed.
This commit fixes the issue related to moving windows to/from a floating
workspace to a tiled workspace.
Previously the start of the move would be ignored however when moving
back from a tiled workspace since it didn't know about the existance of
that window it would also "move" that workspace focused tiled window
without physically moving it, leaving it in a weird state that seemed
like it was unmanaged.
This commit changes the way this mouse moves are handled and now also
handles moving `floating_windows` and even monocle or maximized windows.
This commit allows calls to Border::destroy to fail when called in the
context of border_manager::destroy_all_borders. This is important in the
context of the retile command, which calls this function, to not leave
the retile in an inconsistent state.
LGUG2Z