This has been planned for some time.
It seems necessary to do so now because after updating inputs to track
NixOS 22.05 channels and releases, checks fail unexpectedly, with the
traces leading back to `bud`. Disabling its modules allows checks to
run successfully.
Rather than figuring out what's going wrong with `bud`, it seems more
appropriate to remove it as planned.
the `with pkgs` statement in `commands` muddies the origin of the
variables used within:
- editors are unable to provide helpful advice about undefined variables
because `with pkgs` acts as a blanket "anything is possible" signifier
- at a cursory glance, it may not be immediately clear that `system` is
coming from `pkgs.system`
- `lib` is coming from `pkgs.lib`, but we can use the `lib` module arg.
- some pkgs within `commands` are still referenced with
`pkgs.<package-name>` unnecessarily
Type-check such configs as modules ("moduleType") rather than as
functions evaluating to attribute sets ("functionTo attrs"). This
ensures that they get invoked with all expected module arguments
("pkgs", "lib", etc.).
Update ./examples/hmOnly so that the testuser config references the pkgs
module arg, thus testing that the switch to moduleType works as
intended.
Closes#119
now that we have per-host custom tests, it does not make sense anymore
to have magic lumpsum tests.
users are encouraged to use the new digga.lib.allProfilesTest standard
test and add it to a host's test like so:
`nixos.<host>.tests = [ digga.lib.allPrefilesTest ];`
In this changeset, the `self` module argument is not actually passed.
Upstream nixpkgs, for unknown reasons does not fail with the usual
error message, presumably becuase this is sumoduleWith type, but instead
bails with an inifinite recursion.
