The refactoring in fd75dc8765
introduced a mistake in the error message that doesn't show the full
context anymore. E.g. with this module:
options.foo.bar = lib.mkOption {
type = lib.types.submodule {
baz = 10;
};
default = {};
};
You'd get the error
The option `baz' defined in `/home/infinisil/src/nixpkgs/config.nix' does not exist.
instead of the previous
The option `foo.bar.baz' defined in `/home/infinisil/src/nixpkgs/config.nix' does not exist.
This commit undoes this regression
For programs that have a lot of (Nix-representable) configuration options,
a simple way to represent this in a NixOS module is to declare an
option of a type like `attrsOf str`, representing a key-value mapping
which then gets generated into a config file. However with such a type,
there's no way to add type checking for only some key values.
On the other end of the spectrum, one can declare a single separate
option for every key value that the program supports, ending up with a module
with potentially 100s of options. This has the benefit that every value
gets type checked, catching mistakes at evaluation time already. However
the disadvantage is that the module becomes big, becomes coupled to the
program version and takes a lot of effort to write and maintain.
Previously there was a middle ground between these two
extremes: Declare an option of a type like `attrsOf str`, but declare
additional separate options for the values you wish to have type
checked, and assign their values to the `attrsOf str` option. While this
works decently, it has the problem of duplicated options, since now both
the additional options and the `attrsOf str` option can be used to set a
key value. This leads to confusion about what should happen if both are
set, which defaults should apply, and more.
Now with this change, a middle ground becomes available that solves above
problems: The module system now supports setting a freeform type, which
gets used for all definitions that don't have an associated option. This
means that you can now declare all options you wish to have type
checked, while for the rest a freeform type like `attrsOf str` can be
used.
This fundamentally changes how the module evaluation internally
handles definitions without an associated option.
Previously the values of these definitions were discarded and only
the names were propagated. This was fine because that's all that's
needed for optionally checking whether all definitions have an
associated option with _module.check.
However with the upcoming change of supporting freeform modules,
we *do* need the values of these.
With this change, the module evaluation cleanly separates definitions
that match an option, and ones that do not.
The _module option is added as an internal option set, and it messes up
the results of module evaluations, requiring people to manually filter
_modules out.
If people depend on this, they can still use config._module from inside
the modules, exposing _module as an explicitly declared user option. Or
alternatively with the _module attribute now returned by evalModules.
This helps with troubleshooting exceptions in config values, which were hard
to track down for options with many definitions.
The trace will look like:
error: while evaluating the attribute 'config.foo' at undefined position:
[...]
while evaluating the option `foo':
[...]
while evaluating definitions from `/home/user/mymod.nix':
while evaluating 'dischargeProperties' at /home/user/nixpkgs/lib/modules.nix:464:25, called from /home/user/nixpkgs/lib/modules.nix:392:137:
while evaluating the attribute 'value' at /home/user/nixpkgs/lib/modules.nix:277:44:
Value error!
where the `/home/user/mymod.nix` module is
{ lib, ... }: {
options.foo = lib.mkOption {
type = lib.types.lines;
};
config.foo = builtins.throw "Value error!";
}
Without this change, accessing `mergedValue` from `mergeDefinitions` in
case there are no definitions will throw an error like
error: evaluation aborted with the following error message: 'This case should never happen.'
This change makes it throw the appropriate error
error: The option `foo' is used but not defined.
This is fully backwards compatible.
This fixes imports from the store not being possible, which was caused by
https://github.com/NixOS/nixpkgs/pull/76857
E.g. such a case:
imports = [ "${home-manager}/nixos" ];
With this change, disabledModules applies recursively, meaning if you
have a module "foo.nix" with
imports = [ ./bar.nix ];
then setting
disabledModules = [ "foo.nix" ];
will disable both "foo.nix" and "bar.nix", whereas previously only
"foo.nix" would be disabled.
This change along with https://github.com/NixOS/nixpkgs/pull/61570 allows
modules to be fully disabled even when they have some `mkRenamedOption`
imports.
If I understand correctly, the problem isn't so much that you're assigning to
that top-level attribute, but that the assignment to the attribute (or any
child of the attribute) introduces the 'config' object and prevents 'lifting'
all settings to a generated 'config' object.
We don't want to ignore config that can mess up machines. In general
this should always fail evaluation, as you think you are changing
behaviour and don't, which can easily create run-time errors we can
catch early.
Previously mkRemovedOptionModule would only show the replacement
instructions when the removed option was *defined*. With this change, it
also does so when an option is *used*.
This is essential for options that are only intended to be used such as
`security.acme.directory`, whose replacement instructions would never
trigger without this change because almost everybody only uses the
option and isn't defining it.
This allows `apply` functions to return a valid value if they completely
ignore their argument, which is the case for the option renaming
functions like `mkAliasOptionModule`. Therefore this solves issue #63693
The explicit remove helped to uncover some hidden uses of `optionSet`
in NixOps. However it makes life harder for end-users of NixOps - it will
be impossible to deploy 19.03 systems with old NixOps, but there is no
new release of NixOps with `optionSet` fixes.
Also, "deprecation" process isn't well defined. Even that `optionSet` was
declared "deprecated" for many years, it was never announced. Hence, I
leave "deprecation" announce. Then, 3 releases after announce,
we can announce removal of this feature.
This type has to be removed, not `throw`-ed in runtime, because it makes
some perfectly fine code to fail. For example:
```
$ nix-instantiate --eval -E '(import <nixpkgs/lib>).types' --strict
trace: `types.list` is deprecated; use `types.listOf` instead
error: types.optionSet is deprecated; use types.submodule instead
(use '--show-trace' to show detailed location information)
```
This commit changes the `mkAliasOptionModule` function to make sure that
the priority for the aliased option is propagated to the non-aliased
option.
This also affects the `mkRenamedOptionModule` function in a similar
fashion.
This also removes the `mkAliasOptionModuleWithPriority` function, since
its functionality is now subsumed by `mkAliasOptionModule`.
This change was recommended by @nbp:
https://github.com/NixOS/nixpkgs/pull/53397#discussion_r245487432
This commit adds a function `mkAliasOptionModuleWithPriority`. This
function will make an alias to an existing option and copy over the
priority.
This functionality is needed for PRs like #53041. In that case
`nixos-generate-config` added an option to `hardware-configuration.nix`
with `mkDefault`. That option was then changed and an alias created for
the old name.
The end user should be able to set the non-alias option in their
`configuration.nix` and have everything work correctly. Without this
function, the priority for the option won't be copied over correctly
and the end-user will get a message saying they have the same option
set to two different values.
Before this change `mkRenamedOptionModule` would override option defaults
even when the old option name is left unused. For instance
```nix
{
optios = {
services.name.new = mkOption {
default = { one = {}; };
};
};
imports = [
(mkRenamedOptionModule [ "services" "name" "old" ] [ "services" "name" "new" "two" ])
];
config = {};
}
```
would evaluate to
`{ config.services.name.new = { two = {}; }; }`
when you'd expect it to evaluate to
`{ config.services.name.new = { one = {}; }; }`.
The function isn’t used anywhere and `addErrorContext` is an undocumented
builtin.
The builtin is explicitely qualified at its two uses in the module system.
This allows the lib fixed point to be extended with
myLib = lib.extend (self: super: {
foo = "foo";
})
With this it's possible to have the new modified lib attrset available to all
modules when using evalModules
myLib.evalModules {
modules = [ ({ lib, ... }: {
options.bar = lib.mkOption {
default = lib.foo;
};
}) ];
}
=> { config = { bar = "foo"; ... }; options = ...; }
Among other things, this will allow *2nix tools to output plain data
while still being composable with the traditional
callPackage/.override interfaces.
Before:
<x> is not a integer between 0 and 100 (inclusively).
(notice that “a” is wrong, it should be “an”)
Now:
<x> is not of type `integer between 0 and 100 (inclusively)'.
This sounds a bit more formal, but circumvents the grammatical problems.
Multi-word type descriptions are also easier to see.
This does break the API of being able to import any lib file and get
its libs, however I'm not sure people did this.
I made this while exploring being able to swap out docFn with a stub
in #2305, to avoid functor performance problems. I don't know if that
is going to move forward (or if it is a problem or not,) but after
doing all this work figured I'd put it up anyway :)
Two notable advantages to this approach:
1. when a lib inherits another lib's functions, it doesn't
automatically get put in to the scope of lib
2. when a lib implements a new obscure functions, it doesn't
automatically get put in to the scope of lib
Using the test script (later in this commit) I got the following diff
on the API:
+ diff master fixed-lib
11764a11765,11766
> .types.defaultFunctor
> .types.defaultTypeMerge
11774a11777,11778
> .types.isOptionType
> .types.isType
11781a11786
> .types.mkOptionType
11788a11794
> .types.setType
11795a11802
> .types.types
This means that this commit _adds_ to the API, however I can't find a
way to fix these last remaining discrepancies. At least none are
_removed_.
Test script (run with nix-repl in the PATH):
#!/bin/sh
set -eux
repl() {
suff=${1:-}
echo "(import ./lib)$suff" \
| nix-repl 2>&1
}
attrs_to_check() {
repl "${1:-}" \
| tr ';' $'\n' \
| grep "\.\.\." \
| cut -d' ' -f2 \
| sed -e "s/^/${1:-}./" \
| sort
}
summ() {
repl "${1:-}" \
| tr ' ' $'\n' \
| sort \
| uniq
}
deep_summ() {
suff="${1:-}"
depth="${2:-4}"
depth=$((depth - 1))
summ "$suff"
for attr in $(attrs_to_check "$suff" | grep -v "types.types"); do
if [ $depth -eq 0 ]; then
summ "$attr" | sed -e "s/^/$attr./"
else
deep_summ "$attr" "$depth" | sed -e "s/^/$attr./"
fi
done
}
(
cd nixpkgs
#git add .
#git commit -m "Auto-commit, sorry" || true
git checkout fixed-lib
deep_summ > ../fixed-lib
git checkout master
deep_summ > ../master
)
if diff master fixed-lib; then
echo "SHALLOW MATCH!"
fi
(
cd nixpkgs
git checkout fixed-lib
repl .types
)
* lib: introduce imap0, imap1
For historical reasons, imap starts counting at 1 and it's not
consistent with the rest of the lib.
So for now we split imap into imap0 that starts counting at zero and
imap1 that starts counting at 1. And imap is marked as deprecated.
See c71e2d4235 (commitcomment-21873221)
* replace uses of lib.imap
* lib: move imap to deprecated.nix
