view: honor automatic placement when adjusting floating geometry
The view_adjust_floating_geometry function is called when un-maximizing
a window or changing the output layout to ensure that views are well
placed. Rather than always centering these views should they fall
offscren, use the automatic placement strategy if so configured.
Johan Malm [Thu, 28 Dec 2023 21:35:13 +0000 (21:35 +0000)]
ci: include scripts/find-banned.sh
...to search for functions which have been banned from our code base,
because they're too easy to misuse, and even if used correctly,
complicate audits, cause inconsistencies and/or make static analysis
harder.
Jared Baur [Fri, 15 Dec 2023 15:36:29 +0000 (07:36 -0800)]
Add `touchpad` device type
It is nice to have finer granularity for device types to allow for
configurations such as using `naturalScroll` on touchpads, but not on
regular pointer devices such as mice.
The AutoPlace action will apply placement_find_best() to an active view,
moving it to a position on its output that will minimize overlap with
other views.
Johan Malm [Thu, 28 Dec 2023 17:47:52 +0000 (17:47 +0000)]
keyboard: fix virtual keyboard bug
Do no process virtual keyboard keycodes (just the keysyms).
Reproduce bug by issuing `wlrctl keyboard type xyz` and observe only 'x'
when 'xyz' was expected.
The 'y' and 'z' were matched in match_keybinding() in the keycode section and
returned keybinds for `XF86_AudioLowerVolume` and `XF86_AudioRaiseVolume`
respectively.
Applications may respond to pending resize requests either by ignoring
them or substituting alternative sizes (for example, when mpv constrains
resizes to keep its aspect ratio fixed). In these cases, view->pending
will fall out of sync with the actual view geometry. This will cause
problems when subsequent operations (e.g., MoveToEdge) use the pending
geometry to decide where to place the window.
To fix this, reset view->pending to be equal view->current when either:
1. The requested size change has been commited, to the scene graph, and
no subsequent changes are pending; or
2. The requested size change has been ignored by the client.
Jens Peters [Fri, 22 Dec 2023 10:20:42 +0000 (11:20 +0100)]
xwayland: Prevents usage of invalid data pointer
There are situations where the data pointer of a parent is NULL.
Checking for NULL first fixes e.g. a segmentation fault in CLion
when opening the settings window from a popup menu.
See also https://github.com/labwc/labwc/issues/1351#issuecomment-1867475324
John Lindgren [Tue, 28 Nov 2023 05:38:18 +0000 (00:38 -0500)]
xwayland: fix issue with panel getting stuck offscreen
For unknown reasons, XWayland surfaces that are completely offscreen
seem not to generate commit events. In rare cases, this can prevent an
offscreen window from moving onscreen (since we wait for a commit event
that never occurs). As a workaround, move offscreen surfaces
immediately.
This fixes an issue that I can reproduce by having qmpanel displayed on
a (larger) external monitor, then undocking the laptop so that qmpanel
requests a simultaneous move+resize to the (smaller) laptop display.
John Lindgren [Sat, 16 Dec 2023 14:56:45 +0000 (09:56 -0500)]
theme: use non-hover button variants as fallback
Some themes don't have hover variants for button pixmaps.
It looks better visually to use the non-hover variants as fallbacks
rather than the built-in 6x6 pixmaps.
Johan Malm [Tue, 19 Dec 2023 17:45:43 +0000 (17:45 +0000)]
seat: do not update active_view on layer-shell keyboard focus
...because layer-shell clients are not views and we want to be able to
use foreign-toplevel protocol on the active view even if a client such as
a panel has taken keyboard focus.
fix: fall-through to default cursor in cursor_get_from_edge()
When assertions are disabled, providing an unexpected input to
cursor_get_from_edge will cause the non-void function to terminate
without a return value. This may be effectively unreachable in practice.
However, returning a default cursor as a fall-through case will both
silence a compiler warning and prevent catastrophy should the function
ever be called with a permitted value.
Johan Malm [Sat, 9 Dec 2023 19:37:35 +0000 (19:37 +0000)]
theme: use max_toggled_hover.xbm
...and treat max_hover_toggled.xbm as an alternative name supported for
compatibility reasons.
Use the following button filename schema: "BUTTON [TOGGLED] [STATE]"
with the words separted by underscore and with the following meaning:
- BUTTON can be one of 'max', 'iconify', 'close', 'menu'
- TOGGLED is either 'toggled' or nothing
- STATE is 'hover' or nothing.
This is consistent with the openbox.org wiki and it is believed that this
is how the vast majority of extant openbox themes out there are written.
But please be aware that it is actually different to vanilla Openbox which
uses: "BUTTON [STATE] [TOGGLED]" following a commit in 2014 which broke
themes and led to some distros patching Openbox:
https://github.com/danakj/openbox/commit/35e92e4c2a45b28d5c2c9b44b64aeb4222098c94
Arch Linux and Debian patch Openbox to keep the old syntax (the one that
this commit aligns us with).
https://gitlab.archlinux.org/archlinux/packaging/packages/openbox/-/blob/main/debian-887908.patch?ref_type=heads
Consolatis [Sat, 2 Dec 2023 14:05:26 +0000 (15:05 +0100)]
src/output.c: ensure we don't carry around a stale output pointer
Prevent carrying around the wlr_output->data pointer when
destroying an output. The set_gamma handler may be called
during the destruction of a wlr_output after our own
destroy handler has already been called. This patch resets
the wlr_output->data pointer in our destroy handler and
adds a check in the set_gamma handler to verify the output
is actually valid.
Jens Peters [Thu, 30 Nov 2023 22:25:13 +0000 (23:25 +0100)]
keyboard: allow back by arrow-up or arrow-left in cycle view OSD
All non-modifier keys cycle forward which makes sense for
e.g. tab but is not very intuitive for arrow-up or arrow-left.
Handle those keys separately to provide a feel of navigation
by arrow keys in the cycle view OSD.
John Lindgren [Wed, 15 Nov 2023 16:33:55 +0000 (11:33 -0500)]
keyboard: break up handle_compositor_keybindings()
This function has grown quite large over time. Breaking out various
smaller functions makes the logic easier to follow.
No functional change intended, but there is a minor logical change:
- Due to factoring out match_keybinding(), each keypress can only
match a single keybinding now. Previously, it was theoretically
possible for a single keypress to map to multiple keysyms which could
each match a different keybinding.
John Lindgren [Tue, 22 Nov 2022 10:11:50 +0000 (05:11 -0500)]
output,xwayland: Add support for _NET_WM_STRUT_PARTIAL
Account for space taken up by XWayland panels (as indicated by the
_NET_WM_STRUT_PARTIAL property) in the usable_area calculation.
This makes it possible to use labwc in a "transitional" setup, where it
replaces the X11 window manager and compositor, but most other parts of
a existing X11 desktop environment can still be used via XWayland.
(Some remaining drawbacks of such a setup would be the lack of desktop
icons, and native Wayland clients not showing up in X11-based taskbars.)
John Lindgren [Tue, 24 Oct 2023 17:49:15 +0000 (13:49 -0400)]
xwayland: set _NET_WORKAREA property based on usable area
XWayland clients use the _NET_WORKAREA root window property to determine
how much of the screen is not covered by panels/docks. The property is
used for example by Qt to determine areas of the screen that popup menus
should not overlap (see QScreen::availableVirtualGeometry).
Depends on wlroots MR:
https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/4406
v2: prevent calling wlr_xwayland_set_workareas() too early
v3: fix segfault at exit (server->xwayland == NULL)
We now delay requested resolution changes by the backend until
the next frame event which causes us to render the new content
on the already enlarged buffer. Before this change, an empty
(black) buffer would have been shown instead before the next
frame event caused a new render of the actual contents.
Keep commiting the new state and then scheduling a frame event
would not help as due to the commit call it would still show an
empty buffer in the meantime.
Just modifying wlr_output->pending wouldn't work either because
wlr_scene_output_commit() *completely* ignores it (and it will
be removed in future wlroots commits). For this reason we move
to wlr_scene_output_build_state() directly because it allows us
to supply the current wlr_output->pending state and thus apply
any resolution change in lockstep with new rendering. Result:
No more flickering in the wayland backend and resizing is again
smooth as butter.
This prevents constant flicker while resizing
when running nested via the wayland backend.
For the X11 backend (can be tested via `WLR_BACKENDS=x11 labwc`),
it is still rather janky but at least doesn't cause endless self-
resizing anymore.
projects / proto / labwc.git / log