plasma5: 5.22.0 -> 5.23.1

[CertainLach]

Motivation for this change

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandbox = true set in nix.conf? (See Nix manual)
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• 21.11 Release Notes (or backporting 21.05 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.

[InternetUnexplorer]

FYI: there's a patch to fix the "Other" option in sddm. It will be included in 5.23.1, but many distros are backporting it:

https://invent.kde.org/plasma/plasma-workspace/-/commit/d5a3e749a30613294f41386180aaf31dfb1a9bee

Not sure if it's worth backporting it here though since reviewing Plasma updates usually takes some time and so 5.23.1 might be out by the time this is ready to merge…

[CertainLach]

@InternetUnexplorer included patch

[andrevmatos]

This seems to have some intersection with #137669 / #141128 from staging, maybe basing it on top of that would be useful to ensure the kde-frameworks PRs are used as well?

[andrevmatos]

Nitpick: Wouldn't better to simply add a chmod +x on preBuild or something like that, instead of a full patch? Would be easier to overwrite/inspect and less prone to getting out of date.

[andrevmatos]

This version with kf5.87 has the same bug I saw in #141670: clock is not shown, and clicking on its space shows calendar with some mock values and non-interpolated format strings:

[mohe2015]

This version with kf5.87 has the same bug I saw in #141670: clock is not shown, and clicking on its space shows calendar with some mock values and non-interpolated format strings:

For some unknown reason I just got this with nixos-unstable. Maybe it doesn't have a causality with this update? Or did something update in nixos-unstable. I don't have the time to further investigate right now.

good: e4ef597

bad: 9aeeb75

[mohe2015]

@andrevmatos Could you please test with reverting 763a001 and 4832352
That fixed it for me on unstable where I also had that bug. Didn't test with this PR though.

[mohe2015]

#100057 (comment) maybe also try deleting the cache first (I didn't have time yet to do that)

[SuperSandro2000]

@andrevmatos Could you please test with reverting 763a001 and 4832352 That fixed it for me on unstable where I also had that bug. Didn't test with this PR though.

Didn't change anything for me.

[mohe2015]

@andrevmatos Possibly you had the same problem that I had. The referenced commits reset my default to wayland. Check in your settings/on the login screen if that's the case. Otherwise this may be something different.

[oxalica]

5.23.1 is already out.

Could you also include the patch for https://bugs.kde.org/show_bug.cgi?id=438839 ? Since we have plasma wayland now, this crash issue is quite easy to run into if you have only one monitor.

[mohe2015]

@andrevmatos Possibly you had the same problem that I had. The referenced commits reset my default to wayland. Check in your settings/on the login screen if that's the case. Otherwise this may be something different.

For wayland #100057 (comment) worked for me.

[nrdxp]

Given that Plasma is an important part of the NixOS ISO, and the release feature freeze is looming, would you be willing to rebase on the latest Plasma as @oxalica has suggested? We really should have the latest version at release time if at all feasible.

[oxalica]

I tried this PR and it works almost fine on X11 and no critical issues on wayland. Here are some issues I met,

1. Both analog and digital clock widgets are broken under wayland, as mentioned in plasma5: 5.22.0 -> 5.23.1 #139657 (comment). But on X11, they are fine.
2. Cannot use systemd boot anymore by setting systemdBoot=true or the option runUsingSystemd implemented in nixos/plasma5: make it run using systemd and wayland #117102 . It work before this PR for both X11 and wayland, but after this PR, neither of them works. You can check that plasma systemd user units are not even started and plasmashell is running out of control of systemd.
3. The wayland crash issue I mentioned is still not resolved after the patch applied. This is an upstream bug and I already reported it in that bug tracking thread.

Note that (2) can be the cause of (1) since @mohe2015 in #100057 (comment) reported that without systemdBoot, clock widget is also broken under wayland on current plasma 5.22.0

[iam-cult]

The Wayland clock widget issue is resolved by not declaring any timezone in the system configuration, but it then defaults to UTC.

[FedX-sudo]

The date bug has been ongoing, and is present in 5.22.5. It brakes timedatectl (it returns UTC only) as well, but date works fine.

[ttuegel]

Why is this patch no longer needed?

[andrevmatos]

mobile-broadband path is now fetched from CMake, instead of being hardcoded; this makes it work ootb on nixos

[ttuegel]

SDDM defaulted to the Plasma Wayland session for me, which gives a bad experience because Wayland is so broken. Is that an intentional effect of this pull request, or is it just some bad state left over state from my testing something else?

[InternetUnexplorer]

SDDM defaulted to the Plasma Wayland session for me, which gives a bad experience because Wayland is so broken. Is that an intentional effect of this pull request, or is it just some bad state left over state from my testing something else?

This happened to me too. I think it's because the sessions are sorted alphabetically and it chooses the first one.

There seems to be sddm/sddm#326.

[ttuegel]

SDDM defaulted to the Plasma Wayland session for me, which gives a bad experience because Wayland is so broken. Is that an intentional effect of this pull request, or is it just some bad state left over state from my testing something else?

There is also discussion of this on #100057, so at least we know it's unrelated to this pull request.

[oxalica]

🤔 Seems 5.23.2 is just released one day ago. https://invent.kde.org/plasma/plasma-workspace/-/tree/v5.23.2