Currently, swaybar does not gracefully die if it detects
that the dbus connection was lost. Although it's not recommended
to restart dbus without restarting the compositor, it can very
easily happen. In the case it does, compositor's tray should
not consume 100% cpu until it has to be force killed.
apply suggestions
just setting the bar to not running will call teardown and unref the
dbus.
Make the status command a process group leader and change the kill(2)
calls to target the new process group. Signals sent by swaybar will then
be received by both the status command and its children, if any. While
here, check the result of fork(2).
Without this, children spawned by the status command may not receive the
signals sent by swaybar. As a result, these children may be orphaned on
reload.
The issue could be shown by setting the bar to
bar {
status_command i3status | tee /tmp/i3status.out
}
which would leave orphaned processes for each reload of sway
$ ps o pid,ppid,cmd | grep i3status | grep -v grep
43633 43624 sh -c i3status | tee /tmp/i3status.out
43634 43633 i3status
43635 43633 tee /tmp/i3status.out
$ swaymsg reload
$ ps o pid,ppid,cmd | grep i3status | grep -v grep
43634 1 i3status
43635 1 tee /tmp/i3status.out
43801 43788 sh -c i3status | tee /tmp/i3status.out
43802 43801 i3status
43803 43801 tee /tmp/i3status.out
This fixes#5584.
`$WAYLAND_SOCKET` is unset by `wl_display_connect` after it has
successfully connected to the wayland socket.
However, subprocesses spawned by swaybar (status-command) don't have
access to waybar's fds as $WAYLAND_SOCKET is O_CLOEXEC. This means any
status command which itself tries to connect to wayland will fail if
this environment variable is set.
Reorder display and status-command initialization so that this variable
is not set and add an assert so we can enforce this invariant in future.
This adds complete support for the barconfig_update ipc event. This also
changes the bar command and subcommand handlers to correctly emit the
event. This makes it so all bar subcommands other than id and
swaybar_command are dynamically changeable at runtime. sway-bar.5 has
been updated accordingly
This allows swaybar to become visible when the mode changes (to any
mode other than the default). swaybar will be hidden again when the
modifier is pressed and released or when switching back to the default
mode.
This also applies the same logic to visible by urgency to hide swaybar
when the modifier is pressed and released.
These changes are to match i3's behavior.
This commit mostly duplicates the wlr_log functions, although
with a sway_* prefix. (This is very similar to PR #2009.)
However, the logging function no longer needs to be replaceable,
so sway_log_init's second argument is used to set the exit
callback for sway_abort.
wlr_log_init is still invoked in sway/main.c
This commit makes it easier to remove the wlroots dependency for
the helper programs swaymsg, swaybg, swaybar, and swaynag.
This allows `bar output` and `bar tray_output` to specify an output
identifier. Output names should still work as well.
This parses the output identifier from the xdg_output description,
which wlroots currently sets to `make model serial (name)`. Since this
could change in the future, all identifier comparisons are guarded by
NULL-checks in case the description cannot be parsed to an identifier.
This fixes a few issues with swaybar's cursor scaling:
1. The cursor scale is now changed when the output scale changes
2. The cursor scale is no longer bound by the max output scale when
swaybar is launched
3. Related to the previous item, the cursor is no longer tiny on low
scale outputs after the max output scale has changed
This also bumps up `wl_compositor` to version 4 to allow usage of
`wl_surface_damage_buffer`.
* The loop functions are now prefixed with `loop_`.
* It is now easy to add timers to the loop.
* Timers are implemented using pollfd and timerfd, rather than manually
checking them when any other event happens to arrive.
Previously, when the bar was hidden, the height would be set to 0.
This meant that if the bar was empty upon reshow, it would not render
since the height was still 0, which made it seem there was a problem.
Now, the height is not reset, but the width is, to indicate upon reshow
that the layer surface needed reconfiguring.
As well as adding the hidden_state property to the bar config struct,
this commit handles barconfig_update events when the mode or
hidden_state changes, and uses a new function determine_bar_visibility
to hide or show the bar as required, using, respectively,
destroy_layer_surface, which is also newly added, and add_layer_surface,
which has been changed to allow dynamically adding the surface.
This distinguishes the binding mode from the distinct config mode, as
well as removing mode_pango_markup from the config struct where it
should not be present.
Sway sets a default status_command which runs date every second. This
patch removes this behaviour so the user can have a NULL status bar if
desired.
I had to swap swaybar's event_loop_poll and wl_display_flush so that it
would map the initial surface.
