[BUG] duplicate panels after edit with `nwg-panel-config`

[matthis-k]

Describe the bug more than one panel

To Reproduce Steps to reproduce the behavior:

1. open nwg-panel-config
2. Change a setting (eg remove modules on the right)
3. click Save & restart
4. now there should be 2 panels

Expected behavior The current one gets terminated and the new one starts

Screenshots

Desktop (please complete the following information):

• Linux distribution: Arch
• Sway version sway -v: 1.7
• Panel version nwg-panel -v: 0.6.1

Additional context Can provide if needed

[nwg-piotr]

Well, it's not that easy. It happens to me sometimes, but I haven't yet managed to make it happen on demand. The steps you described above don't reproduce the misbehavior on my machine. Further investigation needed.

[matthis-k]

I just also tested it on my desktop machine: also duplicates. Here the terminal output from nwg-panel-config that might be useful to you and a screenshot afterwards:

Running on sway
28 signal received
17 signal received
Restarting panels
Dependencies check: {'light': True, 'brightnessctl': True, 'pamixer': True, 'pactl': True, 'playerctl': True, 'netifaces': True, 'btmgmt': True, 'wlr-randr': False, 'upower': True, 'swaync': True}
Running instance killed, PID 2628553
Running on sway
Output: HDMI-A-1, position: bottom, layer: bottom, width: 1920, height: 30
Output: HDMI-A-1, position: top, layer: bottom, width: 1920, height: 30
Command 'upower -i $(upower -e | grep devices/battery) | grep --color=never -E 'state|to\ full|to\ empty|percentage'' returned non-zero exit status 1.
Output: DP-1, position: top, layer: bottom, width: 1920, height: 30
28 signal received
host.init(): running dasbus.EventLoop
Exception in thread Thread-14 (init):
Traceback (most recent call last):
  File "/usr/lib/python3.10/threading.py", line 1009, in _bootstrap_inner
    self.run()
StatusNotifierHostInterface -> watcher_available_handler
  File "/usr/lib/python3.10/threading.py", line 946, in run
    self._target(*self._args, **self._kwargs)
  File "/usr/lib/python3.10/site-packages/nwg_panel/modules/sni_system_tray/watcher.py", line 214, in init
    session_bus.register_service(WATCHER_SERVICE_NAME)
  File "/usr/lib/python3.10/site-packages/dasbus/connection.py", line 262, in register_service
    raise ConnectionError("Name request has failed: {}".format(result))
ConnectionError: Name request has failed: 2
No backlight controller was found, so we could not decide an automatic target. The current command will have no effect. Please use -L to find a target and then specify it with -s.

Command 'upower -i $(upower -e | grep devices/battery) | grep --color=never -E 'state|to\ full|to\ empty|percentage'' returned non-zero exit status 1.
No backlight controller was found, so we could not decide an automatic target. The current command will have no effect. Please use -L to find a target and then specify it with -s.

[matthis-k]

Note: I just noticed I also click apply, otherwise there is no duplication. Edit: Never mind, it seems to be rather random, just try it a few times. It also seems to work without an edit. As you can see here, the first time it just refreshes one of the bars, then it adds another one.

[nwg-piotr]

This is damned difficult to nail. I've just pressed "Save & restart" 50 times in a row, to no avail.

[matthis-k]

Maybe try using a different preset? I think the active one is preset 0 and i am starting preset 1.

[nwg-piotr]

I use preset-0 99% of time, and the bug shows up occasionally, say once 2 days. I think it must be located somewhere in main.py lines 290 - 299. I'll get back to the issue as soon as I finish what I'm working on now.

[JDPDO]

I've got the problem sometimes too, no matter what preset or custom config.

Maybe it helps to know, that killall nwg-panel only works if the described behaviour isn't present. If the panel gets duplicated only killall -S SIGKILL nwg-panel works. So maybe its something with the SIGTERM-handlers at line 76?

[nwg-piotr]

Good point. I'll take a closer look at it soon. Sorry for the delay. I've been struggling with several dozen of lines of code since Thursday, and got it to work last night at 4 am. Need to test them while I still understand how the hell it works. ;)

[nwg-piotr]

Hopefully it should be fixed now (on the timeout branch). I'll test till tomorrow and merge, if nothing unexpected happens.

[JDPDO]

Is'nt this bug fixed in merge f5622e1a35e63840fb1826a2ef35a8edb9dfcbc2?

[nwg-piotr]

That's what I meant.