nixos/sway: workaround idle inhibit not working in Firefox

[teutat3s]

Things done

Related to #57602 (comment).

Source: https://www.reddit.com/r/swaywm/comments/1dqud2a/how_to_get_firefox_to_inhibit_idle_when_watching/

Firefox supports the Wayland inhibit protocol, but it attempts to use the DBus interfaces first. However, the gtk portal has an issue where it returns success even though the wlr portal/sway doesn't have an implementation for the inhibit method, see flatpak/xdg-desktop-portal-gtk#465.

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[teutat3s]

@ofborg test sway

[JohnRTitor]

Thanks for the review request, I do not use Sway as such I can not test it. This should get some attention from sway maintainers. @Synthetica9 @RicArch97

[RicArch97]

So is this like a "hack" to make Firefox always inhibit idle when playing audio/video, regardless of the application window state?

Something like for_window [app_id="Firefox"] inhibit_idle fullscreen already works for me to inhibit idle when watching fullscreen video, without this fix.

I think we have to be careful, to no go beyond user expectations regarding idle behavior. If Firefox is supposed to always inhibit idle while playing video's (or even just audio) regardless of application windows state, and this PR fixes that, then i think it's an OK change. Otherwise, we might have to create an extra option to toggle this new behavior.

[teutat3s]

This PR makes idle inhibit work in Firefox like expected, e.g. the screen will not lock when playing a video.
I don't consider it a hack like other workarounds that exist (no lock when audio is playing).

After applying this PR, this can be tested by starting a video in a tab, switching to the terminal to enter the following command, then focus back to Firefox:

echo "mouse over a Firefox window!"; sleep 2; swaymsg -t get_tree | jq '.. | select(.type?) | select(.focused)'

[nixos-discourse]

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/no-sleep-while-firefox-is-running/54755/5

[fpletz]

Tested that this works as intended.

[github-actions]

Successfully created backport PR for release-24.05:

• [Backport release-24.05] nixos/sway: workaround idle inhibit not working in Firefox #350955

[midirhee12]

@teutat3s This is incorrect. xdg.portal.config.sway.default should still be lib.mkDefault [ "wlr" "gtk" ]. You cannot only have gtk; you also need wlr portal. Also, you should have added the mkDefault. Now users of the module cannot modify the portals when necessary.

[JohnRTitor]

Yes, only apply it to xdg.portal.config.sway.default would have been appropiate, my mistake for not pointing that out!

[SuperSandro2000]

Now users of the module cannot modify the portals when necessary.

What is the usecase to not have gtk and wlr without using mkForce?