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.
This commit adds missing error-handling to the creation of the tokener
instance. The stack depth parameter is used to initialize an array that
json-c prefaults ahead of time, causing INT_MAX to result in out of
memory errors.
Also drop the depth to 256 to prevent this OOM.
Though this fix is not very satisfactory -- json-c could be made to
not prefault -- it should do for now. At the very least, swaybar will
not crash.
Fixes#6126.
There's no inherent limit on the nesting Sway can generate, and the
default used by `json_tokener_new`, 32, can plausibly be hit during
regular usage.
Fixes#6115.
Only wl_pointer.motion was used to update pointer position, which would
cause issues if the pointer was not moved prior to wl_pointer.button.
This also fixes touch input through wl_pointer emulation, which fires
wl_pointer.button immediately after wl_pointer.enter.
Copied from a similar fix made to swaynag.
Closes: https://github.com/swaywm/sway/issues/6109
IconThemePath is not a standard property in XDG's StatusNotifierItem
specification, so missing this property should not be logged as an error.
This patch changes the log level to SWAY_DEBUG when swaybar queries the
value of IconThemePath so that swaybar won't log the returned message as
an error if IconThemePath does not exist.
Closes: https://github.com/swaywm/sway/issues/6092
Closes#5605
Text Subpixel antialiasing is :
- FreeType makes glyph bitmaps containing coverage percentage for each
subpixel, instead of pixel
- Then draw by performing the blend for each subpixel, instead of pixel (e.g.
dual-source blending in opengl)
And there's only one Alpha channel, so this extra coverage data can't
leave Cairo to reach the compositor through there.
Therefore, it can't work as intended if output text alpha != bar background
alpha. Disable it for those cases, enable it elsewhere
As for color emojis, they are RGBA bitmaps. If drawn with text alpha=1.0
and background alpha=1.0 (should be completely opaque bar), then with
'CAIRO_OPERATOR_SOURCE' then texels with alpha < 1.0 result in a blend with
whatever's behind the bar, instead of the bar background
When swaybar receives the following JSON body
[
{
"full_text": "foo",
"separator": false,
"separator_block_width": 0
},
{
"full_text": "bar"
}
]
it should not draw any separator or any space between the two blocks.
However, since swaybar calculates that separator_block_width
0 is too small to fit any configured separator, it will override the
separator_block_width with some non-zero value. This patch changes
that such that the necessary separator_block_width is only expanded if
the block has 'separator: true'.
This should be in line to what i3 does, as its documentation of the
i3bar protocol for separator states that "[...] if you disable the
separator line, there will still be a gap after the block, unless you
also use separator_block_width".
The following statusbar output is not considered by sway to be following
the swaybar-protocol:
{"version":1}[[{"full_text":"2.89","urgent":false}],
However this one is:
{"version":1}\n[[{"full_text":"2.89","urgent":false}],
Both outputs contain a header with the required values and an unfinished
array of objects with the required values, but the first one is showed
verbatim.
`determine_bar_visibility` stops and starts the status command process according to the bar’s visibility. If the bar was hidden during teardown, teardown would stall while waiting for the stopped status command process to exit.
This resumes a stopped status command during teardown and allows, for example, sway to reload or quit without leaving a swaybar instance behind each time.
Fixes#5536.
CONT before TERM as requested in review.
`$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.
For some reason my version of sway doesn't show workspace names:
$ swaymsg -t get_outputs
Output HDMI-A-1 '(null) (null) (null)' (inactive)
Output HDMI-A-2 '(null) (null) (null)' (inactive)
Which is weird, but it's no reason to crash swaybar. The field is
totally missing from the JSON, so it ends up doing strcmp(NULL, name)
which is undefined behavior.
See issue #5228. Currently, WL_OUTPUT_SUBPIXEL_NONE is ignored and
CAIRO_ANTIALIAS_SUBPIXEL is still set. This commit checks if subpixel is
set to none and if so, calls set_antialias with CAIRO_ANTIALIAS_GRAY.
This mirrors the functionality in Mako's
[PR261](https://github.com/emersion/mako/pull/261)
This removes any pending messages once the item is destroyed.
Furthermore, this installs SNI event calbacks asynchronously
in order to prevent sd-bus from bypassing pending messages.
24e8ba048a did not take scaling into account.
The hotspot size used pixel coordinates, the absolute coordinates were logical,
and the relative coordinates were completely wrong.
This commit makes all coordinates use logical values. If
`"float_event_coords":true` is sent in the handshake message, coordinates are
sent as floating-point values.
The "scale" field is an integer containing the scale value.
The function group_handler may get a nullptr as `new_group`. If that's
the case, return true, as if `new_group` was the empty string.
Also make the conversion to bool explicit when calling `strcmp`.
