Nix functions

All functions in Nix are anonymous, and can be specified in any of the following ways.

Single identifier

A function written this way will match any argument.

x: !x

This function negates Nix values of the boolean type.

x: y: x + y

This function takes a number or a string and produces a function which takes a value of the same type and either adds or concatenates them, respectively.

Functions written this way can be partially applied.

  concat = x: y: x + y
  map (concat "foo") [ "bar" "baz" "quux" ]

This evaluates to ["foobar" "foobaz" "fooquux"]. We’ve named our function by binding it to an attribute called concat.

Set pattern

A function written this way will match only a set containing the named attributes.

{ x, y }: x + y

Additional arguments can be allowed with ....

{ x, y, ... }: x + y

This is equivalent to the first definition, but can take any set containing x and y attributes.

Default arguments can be set using ?.

{ x ? "hello", y, ... }: x + y

Thus, only a set containing y is required, but we can accept a set with any other attributes. If x is unset, it will default to "hello".


An as-pattern uses a name to refer to the entire set passed to the function.

  func = args@{ x, y, z, ... }: z + y + x + args.a;
  func { x = 1; y = 2; z = 3; a = 4; }

This evaluates to 10.

We can also equivalently write

{ x, y, z, ... } @ args: z + y + x + args.a