example +kitten panel throws X opcode error

[dpnova]

dpn at werk in ~
○ kitty +kitten panel sh -c 'printf "\n\n\nHello, world."; sleep 5s'
X Error of failed request:  BadAtom (invalid Atom parameter)
  Major opcode of failed request:  18 (X_ChangeProperty)
  Atom id in failed request:  0x0
  Serial number of failed request:  296
  Current serial number in output stream:  299

My googling normally associates this error with fairly old remote desktop issues.

Running xmonad via gdm on vanilla ubuntu 24.04:

[dpnova]

So, it doesn't throw the same error when I boot into gnome shell on X. I don't see the panel window, but focus does shift off the terminal while it's running.

I'm not sure what the differences between xmonad on X and gnome on X would be....

[kovidgoyal]

Presumably your window manager doesnt support the properties the kitten needs to make panels. Report it to the WM developers. As for GNOME be sure you are running it in X11 mode not wayland mode, GNOME is generally hostile to other programs drawing desktop elements, so try with something else. All that said, if you really want to get the most out of panels switch to wayland, they are far more powerful/flexible there (with the exception of GNOME).

[dpnova]

Thanks for the reply - totally expect that this isn't an issue with kitty itself but thought you may have had some idea about what set of conditions would lead to this error. No stress if there isn't anything jumping out at you.

As much as I try to get rid of xmonad, I can't seem to find a replacement I'm happy with yet... so X continues to be part of my life 😬

I appreciate the reply, thanks!

edit: yep, gnome was in X mode, i saw the note in the docs about it.

[kovidgoyal]

It's pretty trivial code, it just sets the NET_WM_WINDOW_TYPE_DOCK and NET_WM_STRUT_PARTIAL on the window, presumably xmonad doesnt support one of these.

[dpnova]

Ah, with that info I can go digging - I'll have a play with seeing if there's any event changes required in x11_window.c

Thanks!

[dpnova]

checked in with the xmonad irc

[11:59:43 AM] xmonad does not have any panels built in, you're expected to run an external panel
[11:59:54 AM] so we don't create any of the panel atoms

I gotta get back to work, but when I get a moment I'll see if I can PR up something to create the atoms.

In my googling around I spotted code like this which seems to create the atoms up front

[kovidgoyal]

atoms are created at startup in detectEWMH()

[dpnova]

I pulled main to have a dig into the atoms and built locally:

± kitty/launcher/kitty +kitten panel sh -c 'printf "\n\n\nHello, world."; sleep 5s'
Cannot create panels as the window manager/compositor does not support the neccessary protocols

not sure if related but also noticed this (kitty runs fine despite the error messages) running a pretty minimal xmonad setup, so i'm not that surprised:

dpn at werk in ~/w/kitty on master
± kitty/launcher/kitty                                                             
[0.213] [glfw error 65544]: process_desktop_settings: failed with error: [org.freedesktop.DBus.Error.UnknownMethod] No such interface “org.freedesktop.portal.Settings” on object at path /org/freedesktop/portal/desktop
[0.213] [glfw error 65544]: Notify: Failed to get server capabilities error: [org.freedesktop.DBus.Error.ServiceUnknown] The name org.freedesktop.Notifications was not provided by any .service files

will log what is going on with the atoms to see if something is up there...

[dpnova]

have put a bit of logging around the atom handling and I have everything required there, but the newer error message got me looking deeper and upon checking the source - this is potentially about not having layer shell support...

Have noted your work here too f04fc25

[dpnova]

dropping back to 8e66127 puts me back where I was with my version installed via nightly.

Obviously happy to help test any X11 related stuff you may be doing :)

[kovidgoyal]

On Sat, May 03, 2025 at 06:15:56PM -0700, David Novakovic wrote: I pulled main to have a dig into the atoms and built locally: ``` ± kitty/launcher/kitty +kitten panel sh -c 'printf "\n\n\nHello, world."; sleep 5s' Cannot create panels as the window manager/compositor does not support the neccessary protocols ```

Yes, now kitty checks for support of the needed atoms and bails out early if the window manager doesn't support them instead of generating the badatom errors.

not sure if related but also noticed this (kitty runs fine despite the error messages) running a pretty minimal xmonad setup, so i'm not that surprised:

Completely unrelated, you need to have the settings and notify desktop portals installed to avoid those error messages. They have nothing to do with panels.

[kovidgoyal]

Oh and panels use layer shell, the changes are that now X11 support has been added for desktop background and quick access layer shell windows, in addition to dock panels. See #8592

[dpnova]

that's awesome.

When i dug into the atoms they were all being created ok, but the check on the layer shell support failed.

I'm not familiar with wayland at all - but I'll dig into the source a bit more and see if I can find something more specific about my xorg+xmonad setup that is failing and try rectify.

I appreciate your help and information with this

[kovidgoyal]

On Sun, May 04, 2025 at 12:31:00AM -0700, David Novakovic wrote: that's awesome. When i dug into the atoms they were all being created ok, but the check on the layer shell support failed. I'm not familiar with wayland at all - but I'll dig into the source a bit more and see if I can find something more specific about my setup that is failing and try rectify.

This doesnt have anything to do with wayland. Its called layer shell because it was originally implemented on wayland using the wayland layer shell protocol, but the x11 implementation is completely independent and works by setting various NET_WM properties on the window. The check fails when the window manager does not report the required atoms in the NET_SUPPORTED property of the root X11 window. See the function detectEWMH() in x11_init.c

[dpnova]

This doesnt have anything to do with wayland.

Yep, sorry I assumed I'd need to understand the wayland impl somewhat to grok how the x11 impl was done.

I added a bunch of debug logging around detectEWMH - but in my testing it seemed to be creating the atoms ok. I've since binned the code, I'll reproduce though if that's where you think something is going awry.

Thanks again