Skip to content

Conversation

@Frontear
Copy link
Owner

@Frontear Frontear commented Nov 10, 2025

I'm going to try a slightly new way of performing staging cycles. Having them open from the very beginning and iterating on them. This will hopefully encourage me to add descriptive information for each commit immediately after it's been done, rather than waiting till the merge day.

I'm also aiming for a specific end-date so that I can perform merges in a scheduled manner. This particular date is a little longer than usual, but I've chosen it since Nix 25.11 will be finalized on that day.

Breaking Changes

New Features

  • feat: drop kernel tunings for power savings (3a0ce68)
  • feat: new my.defaults module for opt-in defaults (515f136)
  • feat: properly formatted README.md (7b6231c)
  • feat: add another PR patch to powertop (39fe9c5)
  • feat: introduce the typst extension for VSCode (a5ee795)
  • feat: bump install-nix-action for flake-check workflow (d9b0d0d)

Fixes

  • fix: add Nix 2.32 to nix-benchmark (217d46f)
  • fix: add comment to describe .gitignore entries (ed05328)
  • fix: sync DMS setting pre-start so that changes are visible (5c3ee02)
  • fix: move specialisations to their own file (8d052bc)
  • fix: enable thermald based on detected hardware, drop old enable values (4261b69)
  • fix: drop dead code from DMS configuration (f330839)
  • fix: remove duplicate thermald definition (03822e1)
  • fix: refactor chrony into module, cleanup hosts/laptop (2428337)
  • fix: remove the now deprecated persist-helper program (6e56c95)
  • fix: improve instructions in README.md (da4d91c)
  • fix: remove invalid mkfs.btrfs argument (f4708e9)
  • fix: make nixos-rollback work when booted in a specialisation (0513677)
  • fix: add OBS studio to the GNOME desktop (9bb37e9)
  • fix: repair an irreversible destruction of git history by a flake input (124b304)

@Frontear Frontear changed the title staging: 2025-11-10 -> ??? staging: 2025-11-10 -> 2025-11-30 Nov 10, 2025
This will only be changed for the templates so future projects can have them. Existing projects don't need to be migrated to this since it's not really that important of a change.
This was a rather big change, and was also rather spontaneous. Overall everything should be set up perfectly fine, but there are a couple things to note:

1. The initial settings from `default-settings.json` do not apply. The
   program needs to be restarted for them to work. I suspect this has
   something to do with DMS copying but not changing the settings. This
   will need to be fixed.
2. Night mode does not seem to toggle on if auto mode is enabled. This
   is peculiar but it's not super high priority so I don't really care
   too much.
3. DMS has it's own Polkit agent, I could drop the GNOME one for this.

Overall it looks nice, and it implements many of the things I didn't bother with. Once upon a time I would've wanted to write this stuff myself, but not these days. I'd rather just have something that works.
Before this fix, DMS was simply copying the default settings and then not parsing them. As such, the configuration of DMS was stuck as the default, and a restart of the service was necessary to see the changes. This would become a problem on next boot, since I use a tmpfs root, so the solution was to perform the copy ourselves rather than letting DMS do it.

On a separate note, I should investigate what would happen if the `settings.json` file was managed directly, rather than through this indirect approach. At the moment it doesn't really matter, since tmpfs root + destructive copy means that imperative changes won't stick around for very long, but it's worth investigating to see if I can avoid this indirection. Of course, the one thing that would suck about this would be losing the ability to make an imperative change, then checking what JSON key it corresponds to, and setting it in my Nix configuration. Maybe I should leave it as is, just for this fact alone.
I've always been so desperate for power savings on this laptop. I think it's just time to accept this laptop has bad battery, and I will have to live with keeping it connected via an AC charger 24/7.
I chose not to make it automatic, though I might change that decision down the line.
o7, you served me very well. Thank you :)
All of the instructions here are for my own use. I don't expect anyone else to blindly use them, and I certainly won't help them if they do.
This one prevents powertop from printing values for battery consumption if it fails to read that information from the kernel. Before this change, it was just printing random values, which is not very ideal.
See comments for more details regarding how this change works, and why it was needed.
I find it more useful here since I do most of my academic work here, and sometimes I need to submit a demonstration video.
At the recommendation of many users in the NixOS community, I have decided to try out Typst for writing academic reports. I actually recently finished writing a whole report with it, and I have to say I'm very pleased. It's a significant upgrade from LaTeX by being easy to write like Markdown, but has so many more features on top to make it as good, or maybe even better than other tools. The only problem I had was that I initially thought I could only use their website, but now that I've found a VSCode extension, I will be using that.
Just awful. It's honestly kind of annoyed me that the author did this. They've apparently just wiped the entire flake clean, and also wiped all of the git history pertaining to it. I cannot understand why they'd do that, but I have to say that's not only ridiculously irresponsible, but extremely annoying.

Well anyways, I'm fortunate enough to have had a copy of it in my nix store, so I've gone ahead and grabbed it from there and referenced it in-tree. It will be removed very shortly, but I needed to make this fix so that I could push the staging cycle through.
Thanks to a very dumb author for destroying git history, I had to fetch their flake source in-tree from my local store. Unfortunately, it seems that the version of Nix that my action currently uses has a regression that fails to accept relative paths.
@Frontear Frontear marked this pull request as ready for review December 1, 2025 02:42
@Frontear Frontear merged commit d9b0d0d into main Dec 1, 2025
1 check passed
@Frontear Frontear mentioned this pull request Dec 2, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Migrate from auto-cpufreq to TuneD or power-profiles-daemon

2 participants