postgres: fix double cfg on pg_stat_statements install

[ahuston-0]

Things done

• Made sure, no settings are changed by default
• Tested changes on a real world deployment

Note: I'm not running with cfg.enableAllPreloadedLibraries or cfg.configurePgStatStatementsenabled, so testing this mostly just included checking the settings existed and that my dotfiles built.

Addresses the following build failure:

… while evaluating the option `system.build.toplevel':
       … while evaluating definitions from `/nix/store/6win34qzpvg1vahlc4iq0nhy1hbh90jg-source/nixos/modules/system/activation/top-level.nix':
       … while evaluating the option `warnings':
       … while evaluating definitions from `/nix/store/6win34qzpvg1vahlc4iq0nhy1hbh90jg-source/nixos/modules/system/boot/systemd.nix':
       … while evaluating the option `systemd.services.postgresql.serviceConfig':
       … while evaluating definitions from `/nix/store/6win34qzpvg1vahlc4iq0nhy1hbh90jg-source/nixos/modules/system/boot/systemd.nix':
       … while evaluating the option `systemd.services.postgresql.postStart':
       … while evaluating definitions from `/nix/store/cx38kwf2qq0ybr0pkinhij8z6x997phv-source/modules/postgres.nix':
       (stack trace truncated; use '--show-trace' to show the full, detailed trace)
       error: attribute 'cfg' missing
       at /nix/store/cx38kwf2qq0ybr0pkinhij8z6x997phv-source/modules/postgres.nix:263:59:
          262|             # based on https://git.catgirl.cloud/999eagle/dotfiles-nix/-/blob/main/modules/system/server/postgres/default.nix#L294-302
          263|             (lib.mkIf (cfg.enableAllPreloadedLibraries || cfg.cfg.configurePgStatStatements) (lib.concatStrings (map (db:
             |                                                           ^
          264|               (lib.concatMapStringsSep "\n" (ext: /* bash */ ''

[SuperSandro2000]

What I am wondering is why CI didn't catch that 🤔 do you have any idea?

I am definitely looking into adding a check for postgres to hopefully eval deeper into the module.

[ahuston-0]

What I am wondering is why CI didn't catch that 🤔 do you have any idea?

I am definitely looking into adding a check for postgres to hopefully eval deeper into the module.

That I haven't quite figured out yet. I'm wondering if maybe its because the nixosModules.* resolve into just imports and its not actually checking some (all?) of the options at check time? I did do a verbose nix flake check (log here, recommend searching for postgres as its 60k lines) and noticed it doesn't seem to be evaluating anything else when it gets to the nixosModules level.

[SuperSandro2000]

That could be it. So we probably need a test for each module to force it to evaluate.