Each tray client is directly reparented to the bar window. This saves us
the hassle of having to configure a basically useless tray window and
keeping its background pixmap in sync.
The only disadvantages are having to (un)map each client window
individually and calculating its position relative to the bar window
(which changes all the time) instead of relative to the tray window
(which only changes when clients are added/removed).
The client window has to be added to the save set after it has been
reparented. Otherwise if the reparenting fails weird stuff happens
(windows in the save set have to be child windows of windows created by
the current connection).
Closes: #1526Closes: #314
* debug log
* semi-working prototype. works on the left and the center but not on the right
* fixes formatting
* fixes tests
* - fixed tray_width_change signal
- implements suggestions
* - fixes error with tray positioning
* - tries to fix tests. Does not work
* - fixes tests
* - implemented suggestions
* reverted formatting in comake and doc
* - changed unique_ptr to const reference
* - fixed formatting errors in code
* - actually fixed formatting(ran clang-format)
* - implemented suggestions
* - Added CHANGELOG.md entry(not sure about wording)
* - removed bar_settings from tray_manager::setup
* - fixed issue from rebase
* - fixed issue with tests from rebase
* implemented suggestions
* Update CHANGELOG.md
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
Since polybar sets WM_NORMAL_HINTS, awesomeWM for some reason no longer
respect the position set by polybar before mapping.
reconfiguring the window position once again after mapping the window,
again positions it correctly.
The tray manager of polybar listens on multiple atoms for the background of the root window. On change of these atoms, it will redraw its window-background and message its tray-client to redraw also.
On fast changes of of background, this leads to immense messaging and eventual flickering of the systray.
This patch tries to soften the issue in a way, that tray-window and its client will only redraw, if the bar has transparency. If not, there should be no reason to redraw on wallpaper-change.
* Redraw on background change only if transparent
* Replace tab with spaces
When polybar is remapped (either through IPC or something like xdo
show), openbox positions it somewhere on the screen using the same
positioning algorithm as it would for regular windows.
If the position and size is set in the WM_SIZE_HINTS atom, openbox will
respect the position and size declared by the window.
Fixes#2021
* add units support (POINT, PIXEL, SPACE) for polybar
- add a size_with_unit struct
- add a geometry_format_values struct
- move dpi initialisation from renderer.cpp to bar.cpp
- add a string to size_with_unit converter
- add point support (with pt)
- add pixel support (with px)
* Fix unit test compilation
* clang-format
* Better names
The old names didn't really capture the purpose of the structs and
function.
space_type -> spacing_type
space_size -> spacing_val
size_type -> extent_type
geometry -> extent_val
geometry_format_values -> percentage_with_offset
* Remove parse_size_with_unit
No longer needed. The convert<spacing_val> function in config.cpp
already does all the work for us and always setting the type to pixel
was wrong.
In addition, line-size should not be of type spacing_val but extent_val.
* Cleanup
I tried to address most of my comments on the old PR
* Fix renderer width calculation
We can't just blindly add the x difference to the width because for
example the width should increase if x < width and the increase keeps
x < width.
Similarly, we can't just add the offset to the width.
* Rename geom_format_to_pixels to percentage_with_offset_to_pixel
* Cleanup
* Apply suggested changes from Patrick on GitHub
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
* Update src/components/bar.cpp
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
* Update src/components/config.cpp
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
* Update src/components/builder.cpp
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
* Update src/components/builder.cpp
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
* config: Use stod for parsing percentage
* Use stof instead of strtof
* units: Fix test edge cases
* Remove unnecessary clang-format toggle
* Use percentage_with_offset for margin-{top,bottom}
* Support negative extent values
* Rename unit to units and create a cpp file
* Move percentage_with_offset_to_pixel unit test to units
* Add unit tests for units_utils
* Clarify when and how negative spacing/extent is allowed
Negative spacing is never allowed and produces a config error.
Extents allow negative values in theory, but only a few use-cases accept
it.
Only the extent value used for the `%{O}` tag and the offset value in
percentage_with_offset can be negative. Everything else is capped below
at 0.
The final pixel value of percentage_with_offset also caps below at 0.
* Fix parsing errors not being caught in config
* Print a proper error message for uncaught exceptions
* Cleanup module::get_output
All changes preserve the existing semantics
* Stop using remove_trailing_space in module::get_output
Instead, we first check if the current tag is built, and only if it is,
the spacing is prepended.
* Remove unused imports
* Restore old behavior
If there are two tags and the second one isn't built (module::build
returns false), the space in between them is removed.
For example in the mpd module:
format-online = <toggle> <label-song> foo
If mpd is not running, the mpd module will return false when trying to
build the `<label-song>` tag. If we don't remove the space between
`<toggle>` and `<label-song>`, we end up with two spaces between
`<toggle>` and `foo`.
This change is to match the old behavior where at least one trailing
space character was removed from the builder.
* Add changelog entry
* Remove unused setting
* Use percentage with offset for tray-offset
Co-authored-by: Jérôme BOULMIER <jerome.boulmier@outlook.fr>
Co-authored-by: Joe Groocock <github@frebib.net>
This adds a new `tray-foreground` config option, and uses it instead of
the `tray-background` setting to build up the `_NET_SYSTEM_TRAY_COLORS`
property. I'm pretty sure that the previous behavior was a mistake (this
was introduced in 94298741b6 which is a
pretty large diff). For me and some other people, this results in a
black icon being drawn on top of a black background, which is pretty
useless! I would say that this diff fixes
https://github.com/polybar/polybar/issues/2235. Note: the old code dealt
with `unsigned int` and maxed values out at 0xff. The new code deals
with `uint16_t` and maxes values out at 0xffff. I haven't found the
relevant documentation to justify this change, but from manually
testing, I'm pretty confident this is the right change.
This all matches pretty closely with this code from i3:
[`i3bar/src/xcb.c::init_tray_colors`
code](43e805a00d/i3bar/src/xcb.c (L1490-L1522)),
which you can see also uses the bar's foreground color and maxes values
out at 0xffff, not 0xff.
If you merge this up, I think we should also update
https://github.com/polybar/polybar/wiki/Configuration#bar-settings to
mention the new settings.
Fixes#2235
* fix(monitor): do not include outputs when monitors are supported
Previously, when splitting an output into two monitors, `polybar -m`
would report both the splitted monitors and the output. This was not
caught by the the clone detection as the detection works by removing
monitors contained into another monitors (and monitors were excluded
from that logic) and we want the other way around: outputs covered by
monitors should be ignored.
Instead of trying to detect covered outputs, the solution is quite
simple: when monitors are supported, do not consider outputs, unless
we request all outputs. A monitor can be set primary (and RandR
reports primary outputs as primary monitors). The only information we
would miss from monitors are things like refresh rate and EDID. We
don't need that, so we are fine.
As monitors are only created for connected output (and they are in
this case "active") or disconnected output if they are mapped (and
they are in this case "inactive"), I am a bit unsure if we have
exactly the same behaviour as previously when `connected_only` is set
to `false`.
As some modules require an output, we keep the output in the
`monitor_t` structure and we ensure it is correctly set when using
monitors. A monitor can have 0 or several outputs. We only handle the
0 and 1 cases. When a monitor has more outputs, only the first one is
used. AFAIK, only the xbacklight module needs that and I think we are
fine waiting for a user needing this module and merging monitors.
The C++ binding fail to expose the `outputs()` method to iterate over
the outputs of a monitor. This seems to be a bug in XPP. The field is
correctly defined in the RandR XML file and it works with the Python
binding.
```xml
<struct name="MonitorInfo">
<field type="ATOM" name="name" />
<field type="BOOL" name="primary" />
<field type="BOOL" name="automatic" />
<field type="CARD16" name="nOutput" />
<field type="INT16" name="x" />
<field type="INT16" name="y" />
<field type="CARD16" name="width" /> <!-- pixels -->
<field type="CARD16" name="height" /> <!-- pixels -->
<field type="CARD32" name="width_in_millimeters" />
<field type="CARD32" name="height_in_millimeters" />
<list type="OUTPUT" name="outputs">
<fieldref>nOutput</fieldref>
</list>
</struct>
```
Falling back to C only to access the list of outputs is not enough
because the list is appended to the structure and not visible through
the public API. When copied, the structure loses the list of monitors.
Also, change the mention "XRandR monitor" to "no output" when there is
no output attached. People using monitors know what it means and it is
useful to catch a future regression where we don't have an output at
all (which would break the brightness plugin).
Fix#2481
* Update CHANGELOG.md
Co-authored-by: Patrick Ziegler <p.ziegler96@gmail.com>
XCB_BACK_PIXMAP_PARENT_RELATIVE requires that the client has the same
depth as the tray window.
There was an issue with dropbox having a depth of 32 and the tray window
having a depth of 24 that caused the configuration of the icon to fail.
It would then be displayed outside of the bar because the catch block
was not hit (different exception).
We now just don't configure XCB_CW_BACK_PIXMAP. This seems to work and
is also what stalonetray does.
This does not fix the issue with dropbox having an arbitrary background.
Fixes#1679Fixes#2430
* Add variant support to xkeyboard layout label
Solves #316
* Run style checks, I guess?
* Add comment
* Return nothing if there is no variant
* Update CHANGELOG
Before the module would just try to evenly distribute desktops
(workspaces) among the viewports.
But since `_NET_DESKTOP_VIEWPORT` actually maps desktops to viewports,
we can use that information to assign workspaces to the right viewport.
Fixes#1849Fixes#1764
A warning implies something went wrong and (possibly) the user should do
something about it. However, warnings are not always used this way.
For example:
* When a fallback value for a `${..}` reference is used, this shouldn't
produce a warning (or notice) since using fallbacks is not something
bad.
* pulse telling you that it uses the default sink because no sink was
specified also does not warrant a warning (even notice may be too
high).
* Whenever polybar shuts down it produces a "Termination signal
received..." warning. Since there isn't a more proper way to shut down
polybar, it should not produce a warning. Same argument for a
`screenchange-reload`
Before it would only reload if the size changed and even that was
reliable since the method relied on the order of the monitor list.
Now if the monitor list differs in any way (pos, dimension, primary,
output, name) a reload is issued
Because of how monitors are removed inside the loop and depending on the
monitor order a cloned monitor may be assigned a width of 0 but is never
actually removed resulting in polybar saying the bar is out of bounds
Fixes#1794
Premature optimization that tried to cache monitors but the cache did
not take into account the parameter values.
The call `get_monitors(..., ..., false, true);` would get all connected
and unconncected monitors a subsequent call
`get_monitors(..., ..., true, false);` would get back the same list of
monitors even though it requested only connected monitors.
Additionally `get_monitors` is never called periodically so the
optimization really didn't help much.
Fixes compilation under GCC 9
The default copy constructor implicit generation is deprecated by C++ standard.
The window& operator=(const xcb_window_t win); operator seems to be useless.
Fixes#1728
Ref jaagr/xpp#16
* `layout-icon-*` list that maps layouts to icons.
* `indicator-icon-*` list that maps indicators to off and on icons
* `label-indicator-off`
* `label-indicator-on` which replaces the now deprecated `label-indicator`
* `label-indicator-[on|off]-*` for each indicator. Overrides `label-indicator-on` and `label-indicator-off`
Fixes#1558Closes#1048
* add icon support for xkeyboard layouts
* removed unneeded #include
* add sperate %icon% token that can be used in <label-layout>
* removed unneeded #include
* added caps lock indicator (was mentioned in wiki, but not actually implememnted) and support for indicator icons
* a few more fixes to make sure existing user configs are not broken
* ready to go
* Added an option to replace xkb indicator names
* Added labels for each indicator state
* Removed print left on accident
* Fixed review comments
* Update src/modules/xkeyboard.cpp
Co-Authored-By: Gilnaa <gilad@naaman.io>
Polybar had issues when there is no background set or set by a tool like imagemagick which doesn't add the root pixmap to the root window properties.
There's not much we can do about it, but at least polybar doesn't crash anymore.
Fixes#1582Fixes#1585
* fix(tray_manager): only enable transparency if neccessary
Previously, we always enabled transparency
* fix(background_manager): avoid needless fetching
* fix(renderer): move logging message to correct place
* fix(background_manager): handle dummy pixmap (_XSETROOT_ID) right
* fix(background_manager): more initialization + don't free on error
Freeing on error is incorrect, since we could still be called again later in
which case we still need the resources.
* fix(background_manager): add more infos to trace logs
* fix(background): correct typo (XROOTMAP -> XROOTPMAP)
* fix(background_manager): do not report "no background" as error
* style(background_manager): use braces for if
Co-Authored-By: bennofs <benno.fuenfstueck@gmail.com>
* fix(background_manager): better error message for dummy pixmap
Co-Authored-By: bennofs <benno.fuenfstueck@gmail.com>
* style(background): some more style fixes
* fix(connection): initialize pixmap in all cases in root_pixmap()
* style(connection): improve readability using early return
This adds `monitor-exact = true` in the bar section
This also properly does best-match instead of first-match if multiple
matches exists. For example if there are two monitors HDMI2 and HDMI-2
and we try to match HDMI-2 with monitor-exact = false, until now HDMI2
would be matched. Now exact matches are always preferred.
Fixes#1532
This patch adds support for observing multiple slices of the desktop background.
This is used for the tray so that it doesn't have to rely on the bar's rect to
get the desktop background. In particular, it now handles the case where the
tray is not contained fully within the bar's outer rect (for example, when using tray-offset-{x,y})
Co-Authored-By: bennofs <benno.fuenfstueck@gmail.com>
We need to use positions relative to the position of the bar for indexing into
the background image slice, but the code used absolute ones.
This worked fine as long as absolute positions are the same as relative
positions (this is the case for a bar located at (0,0), so if bottom = false).
But for bottom bars (where the bar position is not (0,0)) this was wrong which
caused the tray background to be black (out of bounds for the background slice).
The systray only supports pseudo transparency (real transparency would require
much larger changes) so the real transparency should only be used for the bar itself.
This option is no longer necessary because the tray background color can now
simply be set to any (semi-)transparent color (just like the bar background).
We now take the bar position that the window manager gives us instead of trying
to calculate it ourselves. This is more correct when multiple bars are attached
to the same edge, as the window manager may move some of them in that
case (assuming override redirect is not enabled)
atoi, atof and so on have undefined behavior if anything goes wrong. We
now use strto*, but without error checking. In most places overflows and
the like *should* not happen. String to number conversions are only used
when reading data from other applications or from the config, if another
application gives unparsable strings or too large numbers, then most
likely there is something wrong with that application. If the error
comes from the user config, then the user has to live with values
provided by strto* on error (which are very reasonable)
Fixes#1201
