]> git.mdlowis.com Git - proto/labwc.git/log
proto/labwc.git
5 weeks agodocs: tab to space: 4, trailing newline: -1, fit some lines to 80-cols master
Tomi Ollila [Tue, 25 Nov 2025 21:03:46 +0000 (23:03 +0200)]
docs: tab to space: 4, trailing newline: -1, fit some lines to 80-cols

5 weeks agorcxml: minor updates for deprecation message
tokyo4j [Thu, 11 Dec 2025 18:58:51 +0000 (03:58 +0900)]
rcxml: minor updates for deprecation message

6 weeks agoNEWS.md: interim update
Johan Malm [Sat, 6 Dec 2025 14:24:39 +0000 (14:24 +0000)]
NEWS.md: interim update

...at start of cool-down period for 0.9.3

6 weeks agodesktop-entry: better handle desktop files with dots in their name
Consolatis [Tue, 9 Dec 2025 08:11:25 +0000 (09:11 +0100)]
desktop-entry: better handle desktop files with dots in their name

This fixes - among others - cases like `R.E.P.O..desktop`.

Due to the duplicated dot at the end we were matching against
a 0 sized string which was always true and thus would always
match the desktop file, regardless of what the app-id was.

A related issue is that an app-id of `osomething` would match
a desktop file name called `R.E.P.O.desktop` due to the string
being size of 1.

Fix this by requiring the partial desktop filename string to
be at least 3 characters wide.

Fixes #3257

6 weeks agowindow-switcher: add `order` parameter to allow stable window list ordering
Maik Broemme [Mon, 8 Dec 2025 17:54:23 +0000 (18:54 +0100)]
window-switcher: add `order` parameter to allow stable window list ordering

Add a new configuration option to control the window switcher traversal order.

`order="focus"` cycling is convenient for quick toggling, but some users - me as well -
prefer a stable taskbar-like order which can now be achieved with `order="age"`.

6 weeks agolabwc-config(5): clarify touchpad and touchscreen sections
Johan Malm [Sat, 6 Dec 2025 20:05:36 +0000 (20:05 +0000)]
labwc-config(5): clarify touchpad and touchscreen sections

6 weeks agocursor: don't synthesize relative motion events from absolute events
John Lindgren [Sat, 6 Dec 2025 17:49:09 +0000 (12:49 -0500)]
cursor: don't synthesize relative motion events from absolute events

It seems to have been inherited behavior from tinywl, but it's not clear
what purpose it serves, and it causes a couple of issues:

- A new absolute position that's discontinuous with the previous cursor
  position can produce unexpectedly large relative motion deltas. This
  can occur for example when multiple input devices are active, or in
  nested/VM scenarios when the pointer leaves the windowed output and
  re-enters at a different point.

- When the cursor position is locked via constraint, the computed deltas
  continue to get larger as the absolute event position diverges further
  from the locked position. This led to the mouse pointer going crazy in
  applications that use the relative events, such as games under Wine/
  Wayland.

6 weeks agoosd-classic: initialize struct fields in order
John Lindgren [Sun, 7 Dec 2025 05:12:27 +0000 (00:12 -0500)]
osd-classic: initialize struct fields in order

6 weeks agoosd-classic: remove unused variable (buf)
John Lindgren [Sun, 7 Dec 2025 05:10:57 +0000 (00:10 -0500)]
osd-classic: remove unused variable (buf)

6 weeks agorcxml: declare variable at initialization
John Lindgren [Sun, 7 Dec 2025 05:09:45 +0000 (00:09 -0500)]
rcxml: declare variable at initialization

6 weeks agonode: update node_descriptor_create() comment
John Lindgren [Sun, 7 Dec 2025 05:08:07 +0000 (00:08 -0500)]
node: update node_descriptor_create() comment

6 weeks agodocs/rc.xml: populate <theme><name> to avoid empty string
Johan Malm [Sat, 6 Dec 2025 14:28:04 +0000 (14:28 +0000)]
docs/rc.xml: populate <theme><name> to avoid empty string

6 weeks agocursor: allow movement until entering the constraint surface
John Lindgren [Thu, 4 Dec 2025 08:02:03 +0000 (03:02 -0500)]
cursor: allow movement until entering the constraint surface

Fixes an issue where the cursor would get stuck (immovable) outside the
window of a Wine/Wayland game, if it was already outside when the game
started (common with a 4:3 game on a 16:9 screen). Now one can manually
move the cursor into the game window, at which point it becomes locked.

This is a minimal/interim fix. Ideally we should warp the cursor into
the constraint area automatically, but that would be a bit more work.

The change to apply_constraint() just turns an assert-failure into a
safe no-op return, since the function is now entered for "locked" as
well as "confined" constraint types.

6 weeks agosnapping: replace <snapping><range> with <snapping><range inner="" outer=""> (#3241)
elviosak [Sat, 6 Dec 2025 07:09:28 +0000 (04:09 -0300)]
snapping: replace <snapping><range> with <snapping><range inner="" outer=""> (#3241)

<inner>/<outer> configure the size of snapping area on output edges with/without adjacent outputs.

6 weeks agodocs/rc.xml.all: add scrollMethod option
Standreas [Fri, 5 Dec 2025 18:46:18 +0000 (19:46 +0100)]
docs/rc.xml.all: add scrollMethod option

6 weeks agorcxml: discourage empty strings in rc.xml configuration
tokyo4j [Wed, 3 Dec 2025 09:26:44 +0000 (18:26 +0900)]
rcxml: discourage empty strings in rc.xml configuration

Background:
I rewrote the config parser in 9462457..2f414a4, but it broke certain
configurations by changing how empty strings are handled: they were mostly
just ignored before my parser rewrite, but after that, they are
interpreted as just empty strings (output="" is considered as 'output named ""').

Though that was unintentional, I believe ignoring empty strings was not a
good idea in the first place, as we already allow empty strings for
certain configurations (e.g. `<desktop prefix="">`), which makes the
parser's behavior inconsistent.

Change:
So let's clarify that we intend to read empty strings as empty strings.
As a preparation, this commit adds warnings for empty strings we are
currently ignoring, so that users can be informed that we intend to just
read empty strings (e.g. `<theme name="">`) as empty strings in the future.
I removed existing empty strings in `rc.xml.all` to avoid warnings when
reading it.

7 weeks agoAdd directional options to `Resize` action:
Maik Broemme [Fri, 28 Nov 2025 19:51:00 +0000 (20:51 +0100)]
Add directional options to `Resize` action:

This introduces an optional "direction" argument to the Resize action,
mirroring Fluxbox's StartResizing [corner] behavior.

Supported values (case-insensitive) are:
up-left, up, up-right, left, right, down-left, down, down-right.

If no direction is specified, the existing behavior is preserved and the
resize edges are inferred from the current pointer position. The action
documentation has been updated to describe the new argument.

7 weeks agoxwayland: flush XCB connection to mitigate race between Raise and input
John Lindgren [Tue, 2 Dec 2025 18:33:07 +0000 (13:33 -0500)]
xwayland: flush XCB connection to mitigate race between Raise and input

7 weeks agocycle: update config to <osd output="all|cursor|focused">
tokyo4j [Tue, 2 Dec 2025 15:41:04 +0000 (00:41 +0900)]
cycle: update config to <osd output="all|cursor|focused">

I think `<windowSwitcher><osd output="keyboard">` is a bit unclear and
hard to interpret as "show OSD in the output with keyboard focus". Also,
we use "cursor" instead of "pointer" in other configurations like
`<placement policy="cursor">` and `<action name="ShowMenu" atCursor="">`.

So let's replace `output="all|pointer|keyboard"` with
`output="all|cursor|focused"`. In documentation, I reordered them to
`output="all|focused|cursor"` as "focused" feels like a bit more
sophisticated and general policy.

7 weeks agocycle: some minor updates to documentation and deprecation warnings
tokyo4j [Tue, 2 Dec 2025 15:36:30 +0000 (00:36 +0900)]
cycle: some minor updates to documentation and deprecation warnings

7 weeks agosrc/interactive.c: fix typo
tokyo4j [Tue, 2 Dec 2025 11:26:50 +0000 (20:26 +0900)]
src/interactive.c: fix typo

7 weeks agoxdg: don't send xdg_wm_base::ping on creation of new surface
John Lindgren [Tue, 2 Dec 2025 02:53:00 +0000 (21:53 -0500)]
xdg: don't send xdg_wm_base::ping on creation of new surface

Although this wasn't causing any known issues, here too we weren't
handling the ping_timeout event, so it seems useless.

7 weeks agoxwayland: don't send _NET_WM_PING to new windows
John Lindgren [Tue, 2 Dec 2025 02:49:31 +0000 (21:49 -0500)]
xwayland: don't send _NET_WM_PING to new windows

No one remembers why we were doing this, and we do not handle the
ping_timeout event, so it seems pointless. Also it appears to confuse
some older X clients.

7 weeks agoxdg: center small fullscreen views and add black background fill
John Lindgren [Mon, 1 Dec 2025 20:13:07 +0000 (15:13 -0500)]
xdg: center small fullscreen views and add black background fill

7 weeks agoview: use wlr_output_layout_get_box()
John Lindgren [Sun, 30 Nov 2025 23:34:25 +0000 (18:34 -0500)]
view: use wlr_output_layout_get_box()

7 weeks agoview: move view_moved() out of view_impl_apply_geometry()
John Lindgren [Mon, 1 Dec 2025 18:58:18 +0000 (13:58 -0500)]
view: move view_moved() out of view_impl_apply_geometry()

7 weeks agobox: factor out box_center()
John Lindgren [Wed, 26 Nov 2025 04:25:47 +0000 (23:25 -0500)]
box: factor out box_center()

7 weeks agocycle: remember cycled window list in server->cycle.views
tokyo4j [Sun, 30 Nov 2025 08:47:40 +0000 (17:47 +0900)]
cycle: remember cycled window list in server->cycle.views

This allows changing the cycled order in the future, e.g. focused order vs
created order.

Functionally, this commit also changes the initially selected window;
before this commit, the previous/next of the topmost window was always
selected, but now the previous/next of the active window is selected first
if it is in the cycled list. This won't change behaviors for most users,
but this ensures that the user can go back to the focused window with
Alt-Tab + Alt-Shift-Tab even when it is not the topmost window.

This commit fixes the TODO in the previous commit by trying to preserve
the selected view when a view is destroyed during window cycling.

7 weeks agocycle: clarify the lifecycle of window switcher
tokyo4j [Sun, 30 Nov 2025 13:10:58 +0000 (22:10 +0900)]
cycle: clarify the lifecycle of window switcher

This commit clarifies the lifecycle of the window switcher (cycle) by:
- init_cycle(): initializes the window switcher (e.g. OSD).
- update_cycle(): updates the window switcher states including OSD,
  preview and outlines.
- destroy_cycle(): clears all the window switcher states.

This commit temporarily regresses by not trying to preserve the selected
view when a view is destroyed. This will be addressed in the next commit.

7 weeks agocycle: factor out get_osd_impl()
tokyo4j [Sun, 30 Nov 2025 13:07:18 +0000 (22:07 +0900)]
cycle: factor out get_osd_impl()

7 weeks agocycle: move & rename destroy_osd_scenes()
tokyo4j [Sun, 30 Nov 2025 13:03:56 +0000 (22:03 +0900)]
cycle: move & rename destroy_osd_scenes()

7 weeks agocycle: use dummy node to remember the scene position of previewed window
tokyo4j [Sun, 30 Nov 2025 11:05:42 +0000 (20:05 +0900)]
cycle: use dummy node to remember the scene position of previewed window

With this approach, we only need to store cycle->dummy_node instead of
cycle->preview_anchor and cycle->preview_parent and we don't need to care
about annoying cases like when the previewed window has no siblings or
when a window tracked by cycle->preview_anchor is destroyed.

7 weeks agosrc/cycle/cycle.c: remove node->enabled check before _set_enabled()
tokyo4j [Sun, 30 Nov 2025 11:04:25 +0000 (20:04 +0900)]
src/cycle/cycle.c: remove node->enabled check before _set_enabled()

wlr_scene_node_set_enabled() just returns early when it doesn't change
enabled state.

7 weeks agocursor: update focus on entering SSD if followMouse=yes
tokyo4j [Sun, 30 Nov 2025 11:45:30 +0000 (20:45 +0900)]
cursor: update focus on entering SSD if followMouse=yes

This fixes a known regression in 885919fc that cursor entering the
titlebar (and other SSD parts) doesn't update the keyboard focus even when
followMouse=yes.

7 weeks agocursor: clarify the semantics of update_pressed_surface()
tokyo4j [Sat, 15 Nov 2025 18:39:49 +0000 (03:39 +0900)]
cursor: clarify the semantics of update_pressed_surface()

This should not change any behaviors. This is mainly just a preparation
for the next commit.

7 weeks agocursor: generalize seat->pressed
tokyo4j [Thu, 20 Nov 2025 16:25:19 +0000 (01:25 +0900)]
cursor: generalize seat->pressed

This commit moduralize seat_{set,reset}_pressed() into
cursor_context_save() so that we can also have seat->hovered later.

7 weeks agoUse "cycle" instead of "osd" across the codebase
tokyo4j [Fri, 28 Nov 2025 17:41:54 +0000 (02:41 +0900)]
Use "cycle" instead of "osd" across the codebase

We were using the word "osd" to describe the window switcher, but it can
be used with on-screen display (OSD) disabled by
`<windowSwitcher><osd show="false">`. Let's use "cycle" instead to avoid
confusion.

7 weeks agoRename osd.{h,c} to cycle.{h,c}
tokyo4j [Fri, 28 Nov 2025 17:09:24 +0000 (02:09 +0900)]
Rename osd.{h,c} to cycle.{h,c}

7 weeks agoxwayland: hide content tree at re-map if shaded
John Lindgren [Mon, 24 Nov 2025 17:00:34 +0000 (12:00 -0500)]
xwayland: hide content tree at re-map if shaded

7 weeks agoview: add defensive checks for null content_tree
John Lindgren [Mon, 24 Nov 2025 15:35:21 +0000 (10:35 -0500)]
view: add defensive checks for null content_tree

7 weeks agoxwayland-unmanaged: remove obsolete workaround
John Lindgren [Tue, 18 Nov 2025 07:37:25 +0000 (02:37 -0500)]
xwayland-unmanaged: remove obsolete workaround

7 weeks agoxwayland: associate/dissociate/map/unmap cleanups
John Lindgren [Mon, 24 Nov 2025 16:27:19 +0000 (11:27 -0500)]
xwayland: associate/dissociate/map/unmap cleanups

- connect/disconnect map handlers in set_surface()
- call set_surface() at time of associate/dissociate

This separates the concepts of "associate" and "map" more clearly.

It's no longer necessary to listen for wlr_surface "destroy" event,
because dissociate is always received first.

Also, view->content_tree is now destroyed and set to NULL at unmap.
Previously, we relied on wlr_scene to destroy it automatically when
the surface was destroyed, but kept a potentially dangling pointer in
view->content_tree until next map. Similar change for unmanaged.

v2: comment updates

7 weeks agoview: deduplicate foreign-toplevel creation code
tokyo4j [Sun, 23 Nov 2025 05:37:12 +0000 (14:37 +0900)]
view: deduplicate foreign-toplevel creation code

7 weeks agoRemove view_impl->map and view_impl->unmap
tokyo4j [Sun, 23 Nov 2025 05:24:42 +0000 (14:24 +0900)]
Remove view_impl->map and view_impl->unmap

jlindgren: data parameter to handle_unmap() is NULL

8 weeks agooutput: avoid use of wlr_scene_output.WLR_PRIVATE.index
John Lindgren [Tue, 25 Nov 2025 22:36:02 +0000 (17:36 -0500)]
output: avoid use of wlr_scene_output.WLR_PRIVATE.index

We were only using it to allow quick bitset comparisons of sets of
outputs (such as view->outputs). We can maintain our own bit IDs for
this purpose and avoid using the private wlroots field.

Note: from my reading of wlr_scene_output_create(), it appears to
always take the lowest unused index, resulting in aggressive re-use of
index values when outputs are disconnected and reconnected. I've tried
to make re-use as infrequent as possible. This could theoretically
reduce the chance of a mix-up in view_update_outputs(), although I'm
not aware of any practical scenario where it matters.

v2: prevent adding more than 64 outputs

8 weeks agoCI: add pcre2 dep for FreeBSD to work around git error
Consolatis [Sun, 23 Nov 2025 02:23:59 +0000 (03:23 +0100)]
CI: add pcre2 dep for FreeBSD to work around git error

2 months agoview: separate (un)minimize and (un)map logic
John Lindgren [Fri, 21 Nov 2025 00:41:00 +0000 (19:41 -0500)]
view: separate (un)minimize and (un)map logic

Map/unmap logic is currently re-used for minimize/unminimize, but lots
of it doesn't actually apply in that case. This is both confusing and
creates some extra complexity, such as:

 - extra "client_request" parameter to unmap(), in which case it has to
   still do some cleanup even if view->mapped is already false

 - various "view->mapped || view->minimized" checks when we really just
   mean "is the view mapped"

To clean this all up, let's put the logic that really is common into
a new view_update_visiblity() function, and stop using map/unmap for
minimize/unminimize.

Note that this changes the meaning of "view->mapped", which used to
mean "mapped and not minimized" but now really just means "mapped".
I left some "view->mapped" conditions as-is (rather than changing to
"view->mapped && !view->minimized") where it seemed to make sense.

v2: add view_update_visibility() as suggested by tokyo4j

2 months agowlr-foreign: set all initial states correctly
John Lindgren [Tue, 18 Nov 2025 03:53:53 +0000 (22:53 -0500)]
wlr-foreign: set all initial states correctly

Two were missing: "minimized" and "activated".

At least "minimized" can be set before map, so the initial state needs
to be read and set on the foreign-toplevel.

I think in the current code, the foreign-toplevel is always created
before a view is activated, but for future-proofing, let's not rely
on this, and set "activated" as well. There's no harm since wlroots
optimizes away any redundant state changes.

2 months agoNEWS.md: interim update
Johan Malm [Sat, 15 Nov 2025 14:51:20 +0000 (14:51 +0000)]
NEWS.md: interim update

2 months agoxwayland: remove obsolete (since wlroots 0.17) NULL check
John Lindgren [Tue, 11 Nov 2025 21:45:28 +0000 (16:45 -0500)]
xwayland: remove obsolete (since wlroots 0.17) NULL check

Since d7dc6e01b446 ("Chase wlroots: Unified mapping"), the map event
is emitted by the wlr_surface itself, thus it cannot be NULL.

Change the "if" to an "assert" accordingly.

2 months agocursor: remove unneeded process_cursor_motion_out_of_surface()
Hiroaki Yamamoto [Sat, 15 Nov 2025 11:14:26 +0000 (20:14 +0900)]
cursor: remove unneeded process_cursor_motion_out_of_surface()

2 months agoxdg: make use of xdg_toplevel_view_get_parent()
John Lindgren [Wed, 12 Nov 2025 15:44:04 +0000 (10:44 -0500)]
xdg: make use of xdg_toplevel_view_get_parent()

2 months agoview: factor out view_impl_init_foreign_toplevel()
John Lindgren [Wed, 12 Nov 2025 15:43:22 +0000 (10:43 -0500)]
view: factor out view_impl_init_foreign_toplevel()

2 months agoxdg: remove unnecessary cast
John Lindgren [Wed, 12 Nov 2025 15:20:03 +0000 (10:20 -0500)]
xdg: remove unnecessary cast

2 months agoseat: Call configure_libinput() for keyboards
David Turner [Wed, 12 Nov 2025 16:48:48 +0000 (16:48 +0000)]
seat: Call configure_libinput() for keyboards

Call configure_libinput() for keyboards in new_keyboard() and
seat_reconfigure().

This allows the use of the sendEventsMode configuration option on
keyboards in order to disable keyboard input, for example:

    <libinput>
        <device category="  RPI Wired Keyboard 1">
            <sendEventsMode>no</sendEventsMode>
        </device>
    </libinput>

2 months agosrc/osd/osd.c: fix indentations for switch-case
tokyo4j [Tue, 11 Nov 2025 10:59:30 +0000 (19:59 +0900)]
src/osd/osd.c: fix indentations for switch-case

2 months agoosd: add multi-monitor support
Samet Aylak [Mon, 10 Nov 2025 04:16:26 +0000 (23:16 -0500)]
osd: add multi-monitor support

Adds `output` attribute to control which monitor(s) display the window
switcher OSD. Supports three modes:
- "all": display on all monitors (default)
- "pointer": display on monitor with mouse cursor
- "keyboard": display on monitor with keyboard focus

The configuration structure is also refactored to nest OSD-specific
settings (show, style, output, thumbnailLabelFormat) under an <osd>
element within <windowSwitcher>, improving logical organization.

2 months agoview: less hacky support for minimize-before-map
John Lindgren [Sat, 8 Nov 2025 03:38:32 +0000 (22:38 -0500)]
view: less hacky support for minimize-before-map

The previous "minimal fix" (5148c2aa3140) worked but was a bit of a
hack, as it basically un-minimized and then immediately minimized the
view again at map. It's not actually too difficult to make the map
handlers aware of minimized views, eliminating the need for the hack.

Note: this depends on the previous commit ("xwayland: connect commit
and surface_destroy handlers together") otherwise the xwayland map
handler registers the commit handler twice, leading to a crash.

2 months agoxwayland: connect commit and surface_destroy handlers together
John Lindgren [Sat, 8 Nov 2025 03:20:18 +0000 (22:20 -0500)]
xwayland: connect commit and surface_destroy handlers together

Factor out set_surface() which consolidates connecting/disconnecting
the wlr_surface event listeners in one place.

In theory, this means we can receive commit events for minimized views.
However, with a test app that resizes itself, I didn't see any change,
i.e. the commits still don't come through until un-minimize. It's
possible they are being filtered at wlroots or protocol level.

Also remove an old, semi-related TODO from view.c.

2 months agoosd: make empty string "" in thumbnailLabelFormat to not show any text (#3196)
elviosak [Sun, 9 Nov 2025 10:20:08 +0000 (07:20 -0300)]
osd: make empty string "" in thumbnailLabelFormat to not show any text (#3196)

2 months agooutput: factor out output_set_has_fullscreen_view()
John Lindgren [Sat, 8 Nov 2025 14:40:26 +0000 (09:40 -0500)]
output: factor out output_set_has_fullscreen_view()

2 months agoview: update top layer visiblity at unmap instead of destroy
John Lindgren [Fri, 7 Nov 2025 19:32:30 +0000 (14:32 -0500)]
view: update top layer visiblity at unmap instead of destroy

It's possible for a fullscreen xwayland view to be unmapped without
being destroyed. In this case, we need to update top layer visibility,
otherwise panels and the like will remain hidden.

Since unmap is always called before destroy, it's sufficient to do the
update only in view_impl_unmap() and not in view_destroy().

Adaptive sync logic needs work still, but I tried to minimize changes
to it since I don't have hardware to test it.

2 months agoosd: add thumbnailLabelFormat to windowSwitcher (#3187)
elviosak [Wed, 5 Nov 2025 09:23:15 +0000 (06:23 -0300)]
osd: add thumbnailLabelFormat to windowSwitcher (#3187)

This commit adds `<windowSwitcher thumbnailLabelFormat="%T">` to configure the label text in each item in the thumbnail-style window switcher. Its format follows `<fields><field content="custom" format="">`.

2 months agodocs: remove repeated words
elviosak [Tue, 4 Nov 2025 21:51:58 +0000 (18:51 -0300)]
docs: remove repeated words

2 months agoosd: allow clicking an osd item to focus the window
tokyo4j [Wed, 29 Oct 2025 18:22:45 +0000 (03:22 +0900)]
osd: allow clicking an osd item to focus the window

2 months agoosd: generalize osd_{classic,thumbnail}_item
tokyo4j [Wed, 29 Oct 2025 17:51:05 +0000 (02:51 +0900)]
osd: generalize osd_{classic,thumbnail}_item

This allows us to share common codes for dealing with osd items.
For example:
- Get the clicked osd item to focus its associated window
- Scroll the items when their total height is taller than output height

2 months agoosd: rename *_scene_item to *_item
tokyo4j [Wed, 29 Oct 2025 18:32:55 +0000 (03:32 +0900)]
osd: rename *_scene_item to *_item

2 months agoosd: move end_cycling() into osd_finish()
tokyo4j [Wed, 29 Oct 2025 18:06:15 +0000 (03:06 +0900)]
osd: move end_cycling() into osd_finish()

...so that we can use `osd_finish()` to support clicking an osd item to
focus its associated window.

2 months agoosd-classic: substitute `theme->osd_border_width+switcher_theme->padding`
tokyo4j [Sun, 2 Nov 2025 03:18:11 +0000 (12:18 +0900)]
osd-classic: substitute `theme->osd_border_width+switcher_theme->padding`

2 months agoworkspaces: place osd at the center of output rather than usable area
tokyo4j [Sun, 2 Nov 2025 02:37:57 +0000 (11:37 +0900)]
workspaces: place osd at the center of output rather than usable area

2 months agoosd: place osd at the center of output rather than usable area
tokyo4j [Sun, 2 Nov 2025 02:29:50 +0000 (11:29 +0900)]
osd: place osd at the center of output rather than usable area

2 months agoosd: use wlr_output_effective_resolution() to get output geometry
tokyo4j [Sat, 1 Nov 2025 16:12:42 +0000 (01:12 +0900)]
osd: use wlr_output_effective_resolution() to get output geometry

Before this commit, output transformations were not taken into account for
parcent-based osd widths like:

osd.window-switcher.style-classic.width: 80%
osd.window-switcher.style-thumbnail.width.max: 80%

2 months agoTranslation updates from weblate
Weblate [Thu, 30 Oct 2025 06:38:35 +0000 (07:38 +0100)]
Translation updates from weblate

Co-authored-by: Weblate <noreply@weblate.org>
Co-authored-by: Yaron Shahrabani <sh.yaron@gmail.com>
Co-authored-by: sudoasd <asdwasd@meiliboxi.fi>
Co-authored-by: zenfas <zenfas66@gmail.com>
Translate-URL: https://translate.lxqt-project.org/projects/labwc/labwc/fi/
Translate-URL: https://translate.lxqt-project.org/projects/labwc/labwc/he/
Translate-URL: https://translate.lxqt-project.org/projects/labwc/labwc/vi/
Translation: Labwc/labwc

2 months agoosd-classic: fix miscalculation for osd width in percentage
tokyo4j [Sat, 1 Nov 2025 13:43:45 +0000 (22:43 +0900)]
osd-classic: fix miscalculation for osd width in percentage

Fixes a regression in 75eb370.

2 months agoosd-thumbnail: fix miscalculation for item geometries
tokyo4j [Sat, 1 Nov 2025 13:49:57 +0000 (22:49 +0900)]
osd-thumbnail: fix miscalculation for item geometries

2 months agoosd-thumbnail: s/thumb_theme/switcher_them/
tokyo4j [Sat, 1 Nov 2025 13:56:45 +0000 (22:56 +0900)]
osd-thumbnail: s/thumb_theme/switcher_them/

Just to align with other parts of the codebase.

2 months agoseat: use CONNECT_SIGNAL() for new_virtual_pointer event
tokyo4j [Sat, 25 Oct 2025 05:15:37 +0000 (14:15 +0900)]
seat: use CONNECT_SIGNAL() for new_virtual_pointer event

2 months agoinput/cursor.c: avoid consuming scroll on client mousebind
elviosak [Fri, 24 Oct 2025 20:26:18 +0000 (17:26 -0300)]
input/cursor.c: avoid consuming scroll on client mousebind

3 months agodocs/: remove XML declarations in .xml files
Johan Malm [Sun, 19 Oct 2025 18:30:38 +0000 (19:30 +0100)]
docs/: remove XML declarations in .xml files

...because we don't like them anymore and libxml2 does not consider them
mandatory.

3 months agodesktop: work around client-side rounding issues at right/bottom pixel
John Lindgren [Mon, 20 Oct 2025 02:15:37 +0000 (22:15 -0400)]
desktop: work around client-side rounding issues at right/bottom pixel

This also avoids a similar server-side rounding issue with some
combinations of wlroots and libwayland versions. See:

- https://gitlab.freedesktop.org/wayland/wayland/-/issues/555
- https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/5159

Tested with qmpanel (similar to lxqt-panel) at 1x and 2x output scale.
Does not help if scaling is done client-side, e.g. QT_SCALE_FACTOR=2.

Fixes: #2379
Fixes: #3099
3 months agolabnag: use overlay layer by default
Johan Malm [Sun, 19 Oct 2025 18:41:24 +0000 (19:41 +0100)]
labnag: use overlay layer by default

...so that the dialog is still visible when some client is using
fullscreen mode.

3 months agoNEWS.md: interim update
Johan Malm [Sun, 19 Oct 2025 17:48:38 +0000 (18:48 +0100)]
NEWS.md: interim update

3 months agoscripts/check: run style check for test files
tokyo4j [Sun, 19 Oct 2025 19:02:34 +0000 (04:02 +0900)]
scripts/check: run style check for test files

Co-authored-by: @johanmalm
3 months agoCI: also run on t/ changes
tokyo4j [Thu, 16 Oct 2025 07:43:59 +0000 (16:43 +0900)]
CI: also run on t/ changes

3 months agotests: remove placeholders for arguments
tokyo4j [Wed, 15 Oct 2025 21:27:20 +0000 (06:27 +0900)]
tests: remove placeholders for arguments

3 months agoDon't remove newlines when parsing config, menu and XBM
tokyo4j [Wed, 15 Oct 2025 07:36:01 +0000 (16:36 +0900)]
Don't remove newlines when parsing config, menu and XBM

Removing newlines in rc.xml and menu.xml caused parser error with
following content:

<!--
 -
 - Some comments
 -
-->

...though it is a valid XML.

Let's not do that. I moved `grab_file()` to `buf.c` and renamed it to
`buf_from_file()`, because it now directly touches `struct buf` and
I don't like having a source file only for one function.

3 months agoTranslation updates from weblate
Weblate [Sat, 11 Oct 2025 20:01:24 +0000 (22:01 +0200)]
Translation updates from weblate

Co-authored-by: alvaroelpob <alvaropobladoresteban9@gmail.com>
Co-authored-by: p-bo <pavel.borecki@gmail.com>
Translate-URL: https://translate.lxqt-project.org/projects/labwc/labwc/ca/
Translate-URL: https://translate.lxqt-project.org/projects/labwc/labwc/cs/
Translation: Labwc/labwc

3 months agotheme: replace THEME_(IN)ACTIVE with SSD_(IN)ACTIVE
tokyo4j [Tue, 14 Oct 2025 04:44:44 +0000 (13:44 +0900)]
theme: replace THEME_(IN)ACTIVE with SSD_(IN)ACTIVE

3 months agowindowswitcher: show 's' as "state" for shaded views
Tomi Ollila [Sun, 6 Jul 2025 21:53:36 +0000 (00:53 +0300)]
windowswitcher: show 's' as "state" for shaded views

While at it sorted the code to show 'm' before 's' and 's' before 'M'
- from the least visible to the most visible state.

3 months agoimplement is_modal_dialog() for xdg shell
Tobias Bengfort [Thu, 9 Oct 2025 06:59:40 +0000 (08:59 +0200)]
implement is_modal_dialog() for xdg shell

3 months agorcxml: call labnag with --keyboard-focus on-demand by default
tokyo4j [Mon, 13 Oct 2025 09:58:23 +0000 (18:58 +0900)]
rcxml: call labnag with --keyboard-focus on-demand by default

3 months agolabnag: add --keyboard-focus option
tokyo4j [Wed, 1 Oct 2025 05:49:23 +0000 (14:49 +0900)]
labnag: add --keyboard-focus option

The new `--keyboard-focus [none|on-demand|exclusive]` option (default:
`none`) allows to some keyboard controls in labnag:

  Right-arrow or Tab: move the button selection to the right
  Left-arrow or Shift-Tab: move the button selection to the left
  Enter: press the selected button
  Escape: close labnag

The selected button is highlighted with the inner 1px border. Maybe we can
instead use different colors for the selected button, but I prefer the
inner border for now because it doesn't require us to add new color
options or make them inherit labwc's theme.

3 months agolabnag: remove redundant lines in conf_init()
tokyo4j [Thu, 2 Oct 2025 16:13:39 +0000 (01:13 +0900)]
labnag: remove redundant lines in conf_init()

3 months agolabnag: s/LAB_EXIT_TIMEOUT/LAB_EXIT_CANCELLED/
tokyo4j [Thu, 2 Oct 2025 05:43:25 +0000 (14:43 +0900)]
labnag: s/LAB_EXIT_TIMEOUT/LAB_EXIT_CANCELLED/

3 months agoosd: allow window switcher to temporary unshade windows
Consolatis [Fri, 3 Oct 2025 00:10:00 +0000 (02:10 +0200)]
osd: allow window switcher to temporary unshade windows

This can be configured with a new unshade="yes|no"
argument for windowSwitcher in rc.xml

Fixes: #3111
3 months agoview: store title/app_id in view
tokyo4j [Wed, 13 Aug 2025 08:40:25 +0000 (17:40 +0900)]
view: store title/app_id in view

This simplifies our codes and eliminates duplicated
`view.events.new_{title,app_id}` events. This should not change any
behaviors.

3 months agoosd-thumbnail: make sure item->{normal,active}_title are non-null
tokyo4j [Sun, 12 Oct 2025 14:59:04 +0000 (23:59 +0900)]
osd-thumbnail: make sure item->{normal,active}_title are non-null

The if-statement doesn't make sense, because `view_get_string_prop()`
never returns NULL. And if it did, it would cause segfault in
`osd_thumbnail_update()`.

3 months agomenu: remove redundant cast
John Lindgren [Sun, 12 Oct 2025 01:47:40 +0000 (21:47 -0400)]
menu: remove redundant cast

3 months agoosd,ssd: don't cast away const
John Lindgren [Wed, 1 Oct 2025 16:41:00 +0000 (12:41 -0400)]
osd,ssd: don't cast away const

3 months agomenu: use xmlFree() for return value of xmlGetProp()
John Lindgren [Mon, 29 Sep 2025 18:51:17 +0000 (14:51 -0400)]
menu: use xmlFree() for return value of xmlGetProp()