Don't immediately ASSERT when the hyprland socket operations fail.
Instead, retry them a few times and then return gracefully instead of
crashing.
https://github.com/scorpion-26/gBar/issues/32
Some apps have a tendency to deadlock after registration (notably
KeePassXC). Now we flush before querying the item properties, so any
pending answer should have left the bus, hopefully avoiding the deadlock
ItemPropertyChanged is called multiple times per actual icon change.
senderName is the same, objectPath is the same, but our registered name
for the callback differs.
We can't use the provided name for item searching, since that *can*
differ from the one we registered with (Looking at you, Discord. Discord
registers with org.kde.StatusNotifierItem-*, but calls back with :*.*)
Since then the object path will sometimes be wrong for the name, we just
use the one from the icon we originally registered with. This causes
multiple icons to reload though
Potential fix for https://github.com/scorpion-26/gBar/issues/26
Giving each main box element (left, center, right) caused the window to
resize and clip out of the monitor when
e.g. the right part was larger than the 1/3 of the window.
Now we have a size for the center widget and try to manually center a box with
that size by setting the left widget's size accordingly. This allows the
right widget to have more room, so it can go right up to the center widget without issues.
For it to actually do something, we need to click twice within 2s. The
clicked button also changes color for visual indication on what was
pressed
See https://github.com/scorpion-26/gBar/issues/17
Implements a rough outline of the SNI (StatusNotifierItem) d-bus
protocol for tray icons.
Note: This is currently *very* WIP
Full implementation will close https://github.com/scorpion-26/gBar/issues/5
The workspace of a second monitor would not be activated, so
lastActiveWorkspace of that would be nullptr. This would also happen for
the first monitor (First and second swapped)
after Hyprland boot.
Now the WS of the first monitor is fully activated by default and every
other monitor has at least lastActiveWorkspace set (Assumes, that WS
with MonID + 1 is on that monitor).
Instead of always polling audio volume, gBar subscribes to events,
causing the update to only happen on demand.
Plus we can block poll updates when there is audio change through gBar,
which results in a much smoother slider.
This change reduces (at least on my end) CPU times from 4-6% per core
to 2-4% per core.
Due to the fact, that we needed to manually propagate hover events from
the slider, two hover events are sent when coming from the side of the
slider. This we need to await BOTH close events until we can call the
leave function.
This commit adds a new widget for the bar: A Microphone widget. It is
disabled by default and can can
be enabled with AudioInput: true.
Implements a flyin as well.
https://github.com/scorpion-26/gBar/issues/5 for the ping
When PA_CONTEXT_READY is called multiple times, PulseAudio::FlusLoop() would get stuck to to integer underflow.
Now there are multiple things preventing that.
Possible fix for https://github.com/scorpion-26/gBar/issues/7
There seems to be a floating point error somewhere though, since
sometimes it goes up by AudioScrollSpeed - 1 instead of
AudioScrollSpeed...
For the ping https://github.com/scorpion-26/gBar/issues/5
Scrolling the workspace widgets now scroll through the workspaces. The
exact behaviour is governed by
'WorkspaceScrollOnMonitor'(either 'm[+/-]1' or 'e[+/-]1'). The direction
can be inverted with 'WorkspaceScrollInvert'.
For the ping: https://github.com/scorpion-26/gBar/issues/5
On my system, the bluetooth widget wouldn't execute the first time the
button is clicked. This probably due to the fact I forgot a NULL at the
end of the execl call. Now I just call system and "call" it a day
This lifts the requirement of needing to disable the dependencies by
hand if you don't have them installed.
It also enables us to have a single package, that works for all.
gLib will not immediately run the callback after the timeout has been
added.
This caused everything to be a few ms late.
Now the timeout will trigger early by default.
Also added an example line to the config at data/config.
Config syntax is: "WorkspaceSymbol-[workspace_number]: [symbol]".
If no default is provided it will default to the circle with dot.
