ci: add treefmt as code formatting multiplexer, refactor CI to avoid duplication, reorg CI into DevOps workflow

[JakobLichterfeld]

As we do have a multitude of programming languages, running each linter separate is not efficient.

Treefmt solves this and comes with speed improvements as well. See the documentation for more information.

• treefmt.toml config
• lint whole project accordingly
  • exclude Grafana Dashboards
• CI
  • refactor CI to avoid duplication
  • reorg CI into DevOps workflow
  • allow ghcr_build parallel to elixir test
  • GitHub Action for treefmt or Pre-Commit Hook using treefmt-nix

closes #4218

[netlify]

✅ Deploy Preview for teslamate ready!

Name	Link
🔨 Latest commit	b48ac45
🔍 Latest deploy log	https://app.netlify.com/sites/teslamate/deploys/670b9662a55fee0008fafd51
😎 Deploy Preview	https://deploy-preview-4219--teslamate.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[swiffer]

👍, but would prefer formatting as created by grafana export for dashboards

[JakobLichterfeld]

👍, but would prefer formatting as created by grafana export for dashboards

Yes, I stumbled over that too. Will exclude the dashboard config from the prettier linter.

[JakobLichterfeld]

Will exclude the dashboard config from the prettier linter.

Done

[JakobLichterfeld]

CI ran successful on push but not on pull_request_target.

[JakobLichterfeld]

I will ensure pull_request_target only runs on real forks to avoid invalid cache name and unnecessary duplicate runs.

[brianmay]

I think the mix hash needs to be updated also:

error: hash mismatch in fixed-output derivation '/nix/store/gly3c0rmyx308dzyl4qssf0irh175wgh-teslamate-mix-deps-1.30.2-dev.drv':
         specified: sha256-ukE7PKrsRaw6TWDSqTJljsPJQedi39mZgGFrMINEJlw=
            got:    sha256-Y+CGgvnSCiiuyhtsQ+j0vayq1IHO5IEPVl+V/wwTd6w=
error: 1 dependencies of derivation '/nix/store/1ladsmz0i5x6zzx15bqqpjv2fywa746g-teslamate-1.30.2-dev.drv' failed to build

[brianmay]

Looks mostly good to me, just needs the two issues resolved.

[scottbot95]

So I did a little digging because it was bothering me that we have to copy the treefmt config in two places and I discovered a potential workaround. treefmt.settings is just used to generate treefmt.build.configFile and treefmt.build.configFile is what is actually used as the config file for the treefmt executable. So we could simply set treefmt.build.configFile ourselves with lib.mkForce and point it to the treefmt.toml in repo. That said, from their code it's unclear to me whether this is an intended extension point or merely an implementation detail so I'm not sure if we really want to rely on it. It's probably fine... What's the worst that could happen 😅

[JakobLichterfeld]

That said, from their code it's unclear to me whether this is an intended extension point or merely an implementation detail so I'm not sure if we really want to rely on it. It's probably fine... What's the worst that could happen 😅

Yeah, I found that hole in the fence too, but was too excited when I finally got it working to do another refactoring :-)

[brianmay]

This error was a bit unexpected:

$ nix flake check --override-input devenv-root "file+file://"<(printf %s "$PWD") .
warning: Git tree '/home/brian/tree/3rdparty/teslamate' is dirty
warning: not writing modified lock file of flake 'git+file:///home/brian/tree/3rdparty/teslamate':
• Updated input 'devenv-root':
    'file:///dev/null?narHash=sha256-d6xi4mKdjkX2JFicDIv5niSzpyI0m/Hnm8GGAIU04kY%3D'
  → 'file:///proc/self/fd/11?narHash=sha256-MR36Ywqs/zTxXcyqXW6MESFZ3X7OT91FF2wTpsZPoEg%3D'
error:
       … while checking flake output 'checks'

         at «none»:0: (source not available)

       … while checking the derivation 'checks.x86_64-linux.default'

         at «none»:0: (source not available)

       (stack trace truncated; use '--show-trace' to show the full trace)

       error: attribute 'nixosModules' missing

       at /nix/store/ngxcbaks38d1vpxflw88983n205bkfic-source/flake.nix:227:33:

          226|                     nixosModules.default = import ./module.nix { inherit self'; };
          227|                     imports = [ self'.nixosModules.default ];
             |                                 ^
          228|                     services.teslamate = {

[JakobLichterfeld]

Not sure yet what is happening with the CI. Elixir test fails even if all test cases pass. I assume this is due to the postgres17 update and was not found with CI before, as the build caching was used in the postgres17 pr.

Edit: Wow, GitHub Runners are really slow today. Will investigate further next week.

[brianmay]

Tests "succeeded" now, after I clicked retry.

Actually if you look at the logs, there was one scary looking error that isn't generating a test failure:

** (FunctionClauseError) no function clause matching in DateTime.to_naive/1
    (elixir 1.16.2) lib/calendar/datetime.ex:925: DateTime.to_naive({:ok, ~U[2003-07-01 00:00:00Z], 0})
    (teslamate 1.30.2-dev) lib/teslamate/log.ex:185: TeslaMate.Log.get_positions_without_elevation/2
    (teslamate 1.30.2-dev) lib/teslamate/terrain.ex:77: TeslaMate.Terrain.handle_event/4
    (stdlib 5.2.3.2) gen_statem.erl:1397: :gen_statem.loop_state_callback/11
    (stdlib 5.2.3.2) proc_lib.erl:241: :proc_lib.init_p_do_apply/3

relevant code is at lib/teslamate/log.ex:185 is:

    date_earliest =
      cond do
        min_id == 0 ->
          DateTime.add(DateTime.utc_now(), -10, :day)

        true ->
          DateTime.from_iso8601("2003-07-01T00:00:00Z")
      end

    naive_date_earliest = DateTime.to_naive(date_earliest)

Problem is DateTime.from_iso8601 returns a tuple, and could in theory fail. Sigh. Wish Elixir had strong type checking...

In actual practice, this probably function I think can't fail, the value is constant.

[brianmay]

I think the above is not a regression in this pull request, so opened a new issue. #4260. Probably better to merge it separately so it doesn't get lost.

Think it is safe to merge this PR now...

[brianmay]

Looks good to me now. Minor stuff can be cleaned up as we find it after this is merged.

[scottbot95]

Also looks good to me. I'll have a PR before to long to do some further clean-up on the nix code.

[JakobLichterfeld]

Tests "succeeded" now, after I clicked retry.

Yeah, GitHub Runners have been slow and I assume a timing error.

Actually if you look at the logs, there was one scary looking error that isn't generating a test failure:

Saw it and was confused as to why there was no failure.

Wish Elixir had strong type checking...

That explains it.

I think the above is not a regression in this pull request, so opened a new issue. #4260. Probably better to merge it separately so it doesn't get lost.

Ty!