lib/types: allow custom submoduleWith descriptions

[ncfavier]

Currently the only way to set the description for a submodule type is to use freeformType. This is not ideal as it requires setting a freeform type, and evaluates the submodule config unnecessarily. Instead, add a description argument to submoduleWith.

The motivation for this is to be able to evaluate Home Manager's NixOS module options in nixos-search with minimal evaluation. To get the description for the attrsOf hmModule type, we currently need to evaluate the whole HM module stack in order to know whether there's a freeformType to get the description from, which requires access to the pkgs argument.

I also added a trivial lib.maybe function to simplify the if x != null then x else y pattern. The name is inspired by Haskell (well, fromMaybe is more accurate) but I'm open to better suggestions.

---

We really need to simplify the payload/binOp logic at some point... something like

{
  op = {
    modules = concatLists;
    specialArgs = mergeDisjointAttrs;
    description = mergeEqualValues;
    # ...
  };
}

would be nice, short of a meta-type system.

[roberth]

We really need to simplify the payload/binOp logic at some point... something like

This exists to support type merging, which is a very useful feature.

Personally, I don't like to be wrong, so I prefer to ask questions. Somewhat conversely, I like to answer whereas I don't like to refute. ☮️

would be nice, short of a meta-type system.

We do have a type for types, or did you have something else in mind?

[ncfavier]

This exists to support type merging, which is a very useful feature.

I know! I'm only saying its implementation could be simplified.

We do have a type for types, or did you have something else in mind?

I guess what I mean is that currently optionType is defined using mergeOptionDecls, but the dependency should really be the other way around, with things like description or shorthandOnlyDefinesConfig as suboptions of the option type. But this sounds like a mine field of infinite recursions.

[ncfavier]

boop

[roberth]

Thanks!