List of Figures

19.1. Staging workflow

Table of Contents

1.1. Overview of Nixpkgs

The Nix Packages collection (Nixpkgs) is a set of thousands of packages for the Nix package manager, released under a permissive MIT/X11 license. Packages are available for several platforms, and can be used with the Nix package manager on most GNU/Linux distributions as well as NixOS.

This manual primarily describes how to write packages for the Nix Packages collection (Nixpkgs). Thus it’s mainly for packagers and developers who want to add packages to Nixpkgs. If you like to learn more about the Nix package manager and the Nix expression language, then you are kindly referred to the Nix manual. The NixOS distribution is documented in the NixOS manual.

1.1. Overview of Nixpkgs

Nix expressions describe how to build packages from source and are collected in the nixpkgs repository. Also included in the collection are Nix expressions for NixOS modules. With these expressions the Nix package manager can build binary packages.

Packages, including the Nix packages collection, are distributed through channels. The collection is distributed for users of Nix on non-NixOS distributions through the channel nixpkgs. Users of NixOS generally use one of the nixos-* channels, e.g. nixos-19.09, which includes all packages and modules for the stable NixOS 19.09. Stable NixOS releases are generally only given security updates. More up to date packages and modules are available via the nixos-unstable channel.

Both nixos-unstable and nixpkgs follow the master branch of the Nixpkgs repository, although both do lag the master branch by generally a couple of days. Updates to a channel are distributed as soon as all tests for that channel pass, e.g. this table shows the status of tests for the nixpkgs channel.

The tests are conducted by a cluster called Hydra, which also builds binary packages from the Nix expressions in Nixpkgs for x86_64-linux, i686-linux and x86_64-darwin. The binaries are made available via a binary cache.

The current Nix expressions of the channels are available in the nixpkgs repository in branches that correspond to the channel names (e.g. nixos-19.09-small).

Chapter 2. Global configuration

Nix comes with certain defaults about what packages can and cannot be installed, based on a package’s metadata. By default, Nix will prevent installation if any of the following criteria are true:

  • The package is thought to be broken, and has had its meta.broken set to true.

  • The package isn’t intended to run on the given system, as none of its meta.platforms match the given system.

  • The package’s meta.license is set to a license which is considered to be unfree.

  • The package has known security vulnerabilities but has not or can not be updated for some reason, and a list of issues has been entered in to the package’s meta.knownVulnerabilities.

Note that all this is checked during evaluation already, and the check includes any package that is evaluated. In particular, all build-time dependencies are checked. nix-env -qa will (attempt to) hide any packages that would be refused.

Each of these criteria can be altered in the nixpkgs configuration.

The nixpkgs configuration for a NixOS system is set in the configuration.nix, as in the following example:

  nixpkgs.config = {
    allowUnfree = true;

However, this does not allow unfree software for individual users. Their configurations are managed separately.

A user’s nixpkgs configuration is stored in a user-specific configuration file located at ~/.config/nixpkgs/config.nix. For example:

  allowUnfree = true;

Note that we are not able to test or build unfree software on Hydra due to policy. Most unfree licenses prohibit us from either executing or distributing the software.

2.1. Installing broken packages

There are two ways to try compiling a package which has been marked as broken.

  • For allowing the build of a broken package once, you can use an environment variable for a single invocation of the nix tools:

  • For permanently allowing broken packages to be built, you may add allowBroken = true; to your user’s configuration file, like this:

      allowBroken = true;

2.2. Installing packages on unsupported systems

There are also two ways to try compiling a package which has been marked as unsupported for the given system.

  • For allowing the build of an unsupported package once, you can use an environment variable for a single invocation of the nix tools:

  • For permanently allowing unsupported packages to be built, you may add allowUnsupportedSystem = true; to your user’s configuration file, like this:

      allowUnsupportedSystem = true;

The difference between a package being unsupported on some system and being broken is admittedly a bit fuzzy. If a program ought to work on a certain platform, but doesn’t, the platform should be included in meta.platforms, but marked as broken with e.g. meta.broken = !hostPlatform.isWindows. Of course, this begs the question of what "ought" means exactly. That is left to the package maintainer.

2.3. Installing unfree packages

There are several ways to tweak how Nix handles a package which has been marked as unfree.

  • To temporarily allow all unfree packages, you can use an environment variable for a single invocation of the nix tools:

  • It is possible to permanently allow individual unfree packages, while still blocking unfree packages by default using the allowUnfreePredicate configuration option in the user configuration file.

    This option is a function which accepts a package as a parameter, and returns a boolean. The following example configuration accepts a package and always returns false:

      allowUnfreePredicate = (pkg: false);

    For a more useful example, try the following. This configuration only allows unfree packages named roon-server and visual studio code:

      allowUnfreePredicate = pkg: builtins.elem (lib.getName pkg) [
  • It is also possible to allow and block licenses that are specifically acceptable or not acceptable, using allowlistedLicenses and blocklistedLicenses, respectively.

    The following example configuration allowlists the licenses amd and wtfpl:

      allowlistedLicenses = with lib.licenses; [ amd wtfpl ];

    The following example configuration blocklists the gpl3Only and agpl3Only licenses:

      blocklistedLicenses = with lib.licenses; [ agpl3Only gpl3Only ];

    Note that allowlistedLicenses only applies to unfree licenses unless allowUnfree is enabled. It is not a generic allowlist for all types of licenses. blocklistedLicenses applies to all licenses.

A complete list of licenses can be found in the file lib/licenses.nix of the nixpkgs tree.

2.4. Installing insecure packages

There are several ways to tweak how Nix handles a package which has been marked as insecure.

  • To temporarily allow all insecure packages, you can use an environment variable for a single invocation of the nix tools:

  • It is possible to permanently allow individual insecure packages, while still blocking other insecure packages by default using the permittedInsecurePackages configuration option in the user configuration file.

    The following example configuration permits the installation of the hypothetically insecure package hello, version 1.2.3:

      permittedInsecurePackages = [
  • It is also possible to create a custom policy around which insecure packages to allow and deny, by overriding the allowInsecurePredicate configuration option.

    The allowInsecurePredicate option is a function which accepts a package and returns a boolean, much like allowUnfreePredicate.

    The following configuration example only allows insecure packages with very short names:

      allowInsecurePredicate = pkg: builtins.stringLength (lib.getName pkg) <= 5;

    Note that permittedInsecurePackages is only checked if allowInsecurePredicate is not specified.

2.5. Modify packages via packageOverrides

You can define a function called packageOverrides in your local ~/.config/nixpkgs/config.nix to override Nix packages. It must be a function that takes pkgs as an argument and returns a modified set of packages.

  packageOverrides = pkgs: rec {
    foo = { ... };

2.6. Declarative Package Management

2.6.1. Build an environment

Using packageOverrides, it is possible to manage packages declaratively. This means that we can list all of our desired packages within a declarative Nix expression. For example, to have aspell, bc, ffmpeg, coreutils, gdb, nixUnstable, emscripten, jq, nox, and silver-searcher, we could use the following in ~/.config/nixpkgs/config.nix:

  packageOverrides = pkgs: with pkgs; {
    myPackages = pkgs.buildEnv {
      name = "my-packages";
      paths = [

To install it into our environment, you can just run nix-env -iA nixpkgs.myPackages. If you want to load the packages to be built from a working copy of nixpkgs you just run nix-env -f. -iA myPackages. To explore what’s been installed, just look through ~/.nix-profile/. You can see that a lot of stuff has been installed. Some of this stuff is useful some of it isn’t. Let’s tell Nixpkgs to only link the stuff that we want:

  packageOverrides = pkgs: with pkgs; {
    myPackages = pkgs.buildEnv {
      name = "my-packages";
      paths = [
      pathsToLink = [ "/share" "/bin" ];

pathsToLink tells Nixpkgs to only link the paths listed which gets rid of the extra stuff in the profile. /bin and /share are good defaults for a user environment, getting rid of the clutter. If you are running on Nix on MacOS, you may want to add another path as well, /Applications, that makes GUI apps available.

2.6.2. Getting documentation

After building that new environment, look through ~/.nix-profile to make sure everything is there that we wanted. Discerning readers will note that some files are missing. Look inside ~/.nix-profile/share/man/man1/ to verify this. There are no man pages for any of the Nix tools! This is because some packages like Nix have multiple outputs for things like documentation (see section 4). Let’s make Nix install those as well.

  packageOverrides = pkgs: with pkgs; {
    myPackages = pkgs.buildEnv {
      name = "my-packages";
      paths = [
      pathsToLink = [ "/share/man" "/share/doc" "/bin" ];
      extraOutputsToInstall = [ "man" "doc" ];

This provides us with some useful documentation for using our packages. However, if we actually want those manpages to be detected by man, we need to set up our environment. This can also be managed within Nix expressions.

  packageOverrides = pkgs: with pkgs; rec {
    myProfile = writeText "my-profile" ''
      export PATH=$HOME/.nix-profile/bin:/nix/var/nix/profiles/default/bin:/sbin:/bin:/usr/sbin:/usr/bin
      export MANPATH=$HOME/.nix-profile/share/man:/nix/var/nix/profiles/default/share/man:/usr/share/man
    myPackages = pkgs.buildEnv {
      name = "my-packages";
      paths = [
        (runCommand "profile" {} ''
          mkdir -p $out/etc/profile.d
          cp ${myProfile} $out/etc/profile.d/
      pathsToLink = [ "/share/man" "/share/doc" "/bin" "/etc" ];
      extraOutputsToInstall = [ "man" "doc" ];

For this to work fully, you must also have this script sourced when you are logged in. Try adding something like this to your ~/.profile file:

if [ -d $HOME/.nix-profile/etc/profile.d ]; then
  for i in $HOME/.nix-profile/etc/profile.d/*.sh; do
    if [ -r $i ]; then
      . $i

Now just run source $HOME/.profile and you can starting loading man pages from your environment.

2.6.3. GNU info setup

Configuring GNU info is a little bit trickier than man pages. To work correctly, info needs a database to be generated. This can be done with some small modifications to our environment scripts.

  packageOverrides = pkgs: with pkgs; rec {
    myProfile = writeText "my-profile" ''
      export PATH=$HOME/.nix-profile/bin:/nix/var/nix/profiles/default/bin:/sbin:/bin:/usr/sbin:/usr/bin
      export MANPATH=$HOME/.nix-profile/share/man:/nix/var/nix/profiles/default/share/man:/usr/share/man
      export INFOPATH=$HOME/.nix-profile/share/info:/nix/var/nix/profiles/default/share/info:/usr/share/info
    myPackages = pkgs.buildEnv {
      name = "my-packages";
      paths = [
        (runCommand "profile" {} ''
          mkdir -p $out/etc/profile.d
          cp ${myProfile} $out/etc/profile.d/
      pathsToLink = [ "/share/man" "/share/doc" "/share/info" "/bin" "/etc" ];
      extraOutputsToInstall = [ "man" "doc" "info" ];
      postBuild = ''
        if [ -x $out/bin/install-info -a -w $out/share/info ]; then
          shopt -s nullglob
          for i in $out/share/info/*.info $out/share/info/*.info.gz; do
              $out/bin/install-info $i $out/share/info/dir

postBuild tells Nixpkgs to run a command after building the environment. In this case, install-info adds the installed info pages to dir which is GNU info’s default root node. Note that texinfoInteractive is added to the environment to give the install-info command.

Chapter 3. Overlays

This chapter describes how to extend and change Nixpkgs using overlays. Overlays are used to add layers in the fixed-point used by Nixpkgs to compose the set of all packages.

Nixpkgs can be configured with a list of overlays, which are applied in order. This means that the order of the overlays can be significant if multiple layers override the same package.

3.1. Installing overlays

The list of overlays can be set either explicitly in a Nix expression, or through <nixpkgs-overlays> or user configuration files.

3.1.1. Set overlays in NixOS or Nix expressions

On a NixOS system the value of the nixpkgs.overlays option, if present, is passed to the system Nixpkgs directly as an argument. Note that this does not affect the overlays for non-NixOS operations (e.g. nix-env), which are looked up independently.

The list of overlays can be passed explicitly when importing nixpkgs, for example import <nixpkgs> { overlays = [ overlay1 overlay2 ]; }.

NOTE: DO NOT USE THIS in nixpkgs. Further overlays can be added by calling the pkgs.extend or pkgs.appendOverlays, although it is often preferable to avoid these functions, because they recompute the Nixpkgs fixpoint, which is somewhat expensive to do.

3.1.2. Install overlays via configuration lookup

The list of overlays is determined as follows.

  1. First, if an overlays argument to the Nixpkgs function itself is given, then that is used and no path lookup will be performed.

  2. Otherwise, if the Nix path entry <nixpkgs-overlays> exists, we look for overlays at that path, as described below.

    See the section on NIX_PATH in the Nix manual for more details on how to set a value for <nixpkgs-overlays>.

  3. If one of ~/.config/nixpkgs/overlays.nix and ~/.config/nixpkgs/overlays/ exists, then we look for overlays at that path, as described below. It is an error if both exist.

If we are looking for overlays at a path, then there are two cases:

  • If the path is a file, then the file is imported as a Nix expression and used as the list of overlays.

  • If the path is a directory, then we take the content of the directory, order it lexicographically, and attempt to interpret each as an overlay by:

    • Importing the file, if it is a .nix file.

    • Importing a top-level default.nix file, if it is a directory.

Because overlays that are set in NixOS configuration do not affect non-NixOS operations such as nix-env, the overlays.nix option provides a convenient way to use the same overlays for a NixOS system configuration and user configuration: the same file can be used as overlays.nix and imported as the value of nixpkgs.overlays.

3.2. Defining overlays

Overlays are Nix functions which accept two arguments, conventionally called self and super, and return a set of packages. For example, the following is a valid overlay.

self: super:

  boost = super.boost.override {
    python = self.python3;
  rr = super.callPackage ./pkgs/rr {
    stdenv = self.stdenv_32bit;

The first argument (self) corresponds to the final package set. You should use this set for the dependencies of all packages specified in your overlay. For example, all the dependencies of rr in the example above come from self, as well as the overridden dependencies used in the boost override.

The second argument (super) corresponds to the result of the evaluation of the previous stages of Nixpkgs. It does not contain any of the packages added by the current overlay, nor any of the following overlays. This set should be used either to refer to packages you wish to override, or to access functions defined in Nixpkgs. For example, the original recipe of boost in the above example, comes from super, as well as the callPackage function.

The value returned by this function should be a set similar to pkgs/top-level/all-packages.nix, containing overridden and/or new packages.

Overlays are similar to other methods for customizing Nixpkgs, in particular the packageOverrides attribute described in Section 2.5, “Modify packages via packageOverrides. Indeed, packageOverrides acts as an overlay with only the super argument. It is therefore appropriate for basic use, but overlays are more powerful and easier to distribute.

3.3. Using overlays to configure alternatives

Certain software packages have different implementations of the same interface. Other distributions have functionality to switch between these. For example, Debian provides DebianAlternatives. Nixpkgs has what we call alternatives, which are configured through overlays.


In Nixpkgs, we have multiple implementations of the BLAS/LAPACK numerical linear algebra interfaces. They are:

  • OpenBLAS

    The Nixpkgs attribute is openblas for ILP64 (integer width = 64 bits) and openblasCompat for LP64 (integer width = 32 bits). openblasCompat is the default.

  • LAPACK reference (also provides BLAS)

    The Nixpkgs attribute is lapack-reference.

  • Intel MKL (only works on the x86_64 architecture, unfree)

    The Nixpkgs attribute is mkl.

  • BLIS

    BLIS, available through the attribute blis, is a framework for linear algebra kernels. In addition, it implements the BLAS interface.

  • AMD BLIS/LIBFLAME (optimized for modern AMD x86_64 CPUs)

    The AMD fork of the BLIS library, with attribute amd-blis, extends BLIS with optimizations for modern AMD CPUs. The changes are usually submitted to the upstream BLIS project after some time. However, AMD BLIS typically provides some performance improvements on AMD Zen CPUs. The complementary AMD LIBFLAME library, with attribute amd-libflame, provides a LAPACK implementation.

Introduced in PR #83888, we are able to override the blas and lapack packages to use different implementations, through the blasProvider and lapackProvider argument. This can be used to select a different provider. BLAS providers will have symlinks in $out/lib/ and $out/lib/ to their respective BLAS libraries. Likewise, LAPACK providers will have symlinks in $out/lib/ and $out/lib/ to their respective LAPACK libraries. For example, Intel MKL is both a BLAS and LAPACK provider. An overlay can be created to use Intel MKL that looks like:

self: super:

  blas = super.blas.override {
    blasProvider = self.mkl;

  lapack = super.lapack.override {
    lapackProvider = self.mkl;

This overlay uses Intel’s MKL library for both BLAS and LAPACK interfaces. Note that the same can be accomplished at runtime using LD_LIBRARY_PATH of and For instance:

$ LD_LIBRARY_PATH=$(nix-build -A mkl)/lib${LD_LIBRARY_PATH:+:}$LD_LIBRARY_PATH nix-shell -p octave --run octave

Intel MKL requires an openmp implementation when running with multiple processors. By default, mkl will use Intel’s iomp implementation if no other is specified, but this is a runtime-only dependency and binary compatible with the LLVM implementation. To use that one instead, Intel recommends users set it with LD_PRELOAD. Note that mkl is only available on x86_64-linux and x86_64-darwin. Moreover, Hydra is not building and distributing pre-compiled binaries using it.

For BLAS/LAPACK switching to work correctly, all packages must depend on blas or lapack. This ensures that only one BLAS/LAPACK library is used at one time. There are two versions of BLAS/LAPACK currently in the wild, LP64 (integer size = 32 bits) and ILP64 (integer size = 64 bits). Some software needs special flags or patches to work with ILP64. You can check if ILP64 is used in Nixpkgs with blas.isILP64 and lapack.isILP64. Some software does NOT work with ILP64, and derivations need to specify an assertion to prevent this. You can prevent ILP64 from being used with the following:

{ stdenv, blas, lapack, ... }:

assert (!blas.isILP64) && (!lapack.isILP64);

stdenv.mkDerivation {

3.3.2. Switching the MPI implementation

All programs that are built with MPI support use the generic attribute mpi as an input. At the moment Nixpkgs natively provides two different MPI implementations:

  • Open MPI (default), attribute name openmpi

  • MPICH, attribute name mpich

  • MVAPICH, attribute name mvapich

To provide MPI enabled applications that use MPICH, instead of the default Open MPI, simply use the following overlay:

self: super:

  mpi = self.mpich;

Chapter 4. Overriding

Sometimes one wants to override parts of nixpkgs, e.g. derivation attributes, the results of derivations.

These functions are used to make changes to packages, returning only single packages. Overlays, on the other hand, can be used to combine the overridden packages across the entire package set of Nixpkgs.

4.1. <pkg>.override

The function override is usually available for all the derivations in the nixpkgs expression (pkgs).

It is used to override the arguments passed to a function.

Example usages: { arg1 = val1; arg2 = val2; ... }
import pkgs.path { overlays = [ (self: super: {
  foo = { barSupport = true ; };
mypkg = pkgs.callPackage ./mypkg.nix {
  mydep = pkgs.mydep.override { ... };

In the first example, is the result of a function call with some default arguments, usually a derivation. Using will call the same function with the given new arguments.

4.2. <pkg>.overrideAttrs

The function overrideAttrs allows overriding the attribute set passed to a stdenv.mkDerivation call, producing a new derivation based on the original one. This function is available on all derivations produced by the stdenv.mkDerivation function, which is most packages in the nixpkgs expression pkgs.

Example usage:

helloWithDebug = pkgs.hello.overrideAttrs (oldAttrs: rec {
  separateDebugInfo = true;

In the above example, the separateDebugInfo attribute is overridden to be true, thus building debug info for helloWithDebug, while all other attributes will be retained from the original hello package.

The argument oldAttrs is conventionally used to refer to the attr set originally passed to stdenv.mkDerivation.

Note: Note that separateDebugInfo is processed only by the stdenv.mkDerivation function, not the generated, raw Nix derivation. Thus, using overrideDerivation will not work in this case, as it overrides only the attributes of the final derivation. It is for this reason that overrideAttrs should be preferred in (almost) all cases to overrideDerivation, i.e. to allow using stdenv.mkDerivation to process input arguments, as well as the fact that it is easier to use (you can use the same attribute names you see in your Nix code, instead of the ones generated (e.g. buildInputs vs nativeBuildInputs), and it involves less typing).

4.3. <pkg>.overrideDerivation

Warning: You should prefer overrideAttrs in almost all cases, see its documentation for the reasons why. overrideDerivation is not deprecated and will continue to work, but is less nice to use and does not have as many abilities as overrideAttrs.
Warning: Do not use this function in Nixpkgs as it evaluates a Derivation before modifying it, which breaks package abstraction and removes error-checking of function arguments. In addition, this evaluation-per-function application incurs a performance penalty, which can become a problem if many overrides are used. It is only intended for ad-hoc customisation, such as in ~/.config/nixpkgs/config.nix.

The function overrideDerivation creates a new derivation based on an existing one by overriding the original’s attributes with the attribute set produced by the specified function. This function is available on all derivations defined using the makeOverridable function. Most standard derivation-producing functions, such as stdenv.mkDerivation, are defined using this function, which means most packages in the nixpkgs expression, pkgs, have this function.

Example usage:

mySed = pkgs.gnused.overrideDerivation (oldAttrs: {
  name = "sed-4.2.2-pre";
  src = fetchurl {
    url =;
    sha256 = "11nq06d131y4wmf3drm0yk502d2xc6n5qy82cg88rb9nqd2lj41k";
  patches = [];

In the above example, the name, src, and patches of the derivation will be overridden, while all other attributes will be retained from the original derivation.

The argument oldAttrs is used to refer to the attribute set of the original derivation.

Note: A package’s attributes are evaluated before being modified by the overrideDerivation function. For example, the name attribute reference in url = "mirror://gnu/hello/${name}.tar.gz"; is filled-in before the overrideDerivation function modifies the attribute set. This means that overriding the name attribute, in this example, will not change the value of the url attribute. Instead, we need to override both the name and url attributes.

4.4. lib.makeOverridable

The function lib.makeOverridable is used to make the result of a function easily customizable. This utility only makes sense for functions that accept an argument set and return an attribute set.

Example usage:

f = { a, b }: { result = a+b; };
c = lib.makeOverridable f { a = 1; b = 2; };

The variable c is the value of the f function applied with some default arguments. Hence the value of c.result is 3, in this example.

The variable c however also has some additional functions, like c.override which can be used to override the default arguments. In this example the value of (c.override { a = 4; }).result is 6.

Chapter 5. Functions reference

The nixpkgs repository has several utility functions to manipulate Nix expressions.

5.1. Nixpkgs Library Functions

Nixpkgs provides a standard library at pkgs.lib, or through import <nixpkgs/lib>.

5.1.1. Assert functions lib.asserts.assertMsg

assertMsg :: Bool -> String -> Bool

Located at lib/asserts.nix:21 in <nixpkgs>.

Print a trace message if pred is false.

Intended to be used to augment asserts with helpful error messages.


Condition under which the msg should not be printed.


Message to print.

Example 5.1. Printing when the predicate is false

assert lib.asserts.assertMsg ("foo" == "bar") "foo is not bar, silly"
stderr> trace: foo is not bar, silly
stderr> assert failed lib.asserts.assertOneOf

assertOneOf :: String -> String -> StringList -> Bool

Located at lib/asserts.nix:38 in <nixpkgs>.

Specialized asserts.assertMsg for checking if val is one of the elements of xs. Useful for checking enums.


The name of the variable the user entered val into, for inclusion in the error message.


The value of what the user provided, to be compared against the values in xs.


The list of valid values.

Example 5.2. Ensuring a user provided a possible value

let sslLibrary = "bearssl";
in lib.asserts.assertOneOf "sslLibrary" sslLibrary [ "openssl" "libressl" ];
=> false
stderr> trace: sslLibrary must be one of "openssl", "libressl", but is: "bearssl"

5.1.2. Attribute-Set Functions lib.attrset.attrByPath

attrByPath :: [String] -> Any -> AttrSet -> Any

Located at lib/attrsets.nix:24 in <nixpkgs>.

Return an attribute from within nested attribute sets.


A list of strings representing the path through the nested attribute set set.


Default value if attrPath does not resolve to an existing value.


The nested attributeset to select values from.

Example 5.3. Extracting a value from a nested attribute set

let set = { a = { b = 3; }; };
in lib.attrsets.attrByPath [ "a" "b" ] 0 set
=> 3

Example 5.4. No value at the path, instead using the default

lib.attrsets.attrByPath [ "a" "b" ] 0 {}
=> 0 lib.attrsets.hasAttrByPath

hasAttrByPath :: [String] -> AttrSet -> Bool

Located at lib/attrsets.nix:42 in <nixpkgs>.

Determine if an attribute exists within a nested attribute set.


A list of strings representing the path through the nested attribute set set.


The nested attributeset to check.

Example 5.5. A nested value does exist inside a set

  [ "a" "b" "c" "d" ]
  { a = { b = { c = { d = 123; }; }; }; }
=> true lib.attrsets.setAttrByPath

setAttrByPath :: [String] -> Any -> AttrSet

Located at lib/attrsets.nix:57 in <nixpkgs>.

Create a new attribute set with value set at the nested attribute location specified in attrPath.


A list of strings representing the path through the nested attribute set.


The value to set at the location described by attrPath.

Example 5.6. Creating a new nested attribute set

lib.attrsets.setAttrByPath [ "a" "b" ] 3
=> { a = { b = 3; }; } lib.attrsets.getAttrFromPath

getAttrFromPath :: [String] -> AttrSet -> Value

Located at lib/attrsets.nix:76 in <nixpkgs>.

Like Section, “lib.attrset.attrByPath except without a default, and it will throw if the value doesn't exist.


A list of strings representing the path through the nested attribute set set.


The nested attribute set to find the value in.

Example 5.7. Succesfully getting a value from an attribute set

lib.attrsets.getAttrFromPath [ "a" "b" ] { a = { b = 3; }; }
=> 3

Example 5.8. Throwing after failing to get a value from an attribute set

lib.attrsets.getAttrFromPath [ "x" "y" ] { }
=> error: cannot find attribute `x.y' lib.attrsets.attrVals

attrVals :: [String] -> AttrSet -> [Any]

Located at lib/attrsets.nix:87 in <nixpkgs>.

Return the specified attributes from a set. All values must exist.


The list of attributes to fetch from set. Each attribute name must exist on the attrbitue set.


The set to get attribute values from.

Example 5.9. Getting several values from an attribute set

lib.attrsets.attrVals [ "a" "b" "c" ] { a = 1; b = 2; c = 3; }
=> [ 1 2 3 ]

Example 5.10. Getting missing values from an attribute set

lib.attrsets.attrVals [ "d" ] { }
error: attribute 'd' missing lib.attrsets.attrValues

attrValues :: AttrSet -> [Any]

Located at lib/attrsets.nix:97 in <nixpkgs>.

Get all the attribute values from an attribute set.

Provides a backwards-compatible interface of builtins.attrValues for Nix version older than 1.8.


The attribute set.

Example 5.11. 

lib.attrsets.attrValues { a = 1; b = 2; c = 3; }
=> [ 1 2 3 ] lib.attrsets.catAttrs

catAttrs :: String -> [AttrSet] -> [Any]

Located at lib/attrsets.nix:116 in <nixpkgs>.

Collect each attribute named `attr' from the list of attribute sets, sets. Sets that don't contain the named attribute are ignored.

Provides a backwards-compatible interface of builtins.catAttrs for Nix version older than 1.9.


Attribute name to select from each attribute set in sets.


The list of attribute sets to select attr from.

Example 5.12. Collect an attribute from a list of attribute sets.

Attribute sets which don't have the attribute are ignored.

catAttrs "a" [{a = 1;} {b = 0;} {a = 2;}]
=> [ 1 2 ] lib.attrsets.filterAttrs

filterAttrs :: (String -> Any -> Bool) -> AttrSet -> AttrSet

Located at lib/attrsets.nix:127 in <nixpkgs>.

Filter an attribute set by removing all attributes for which the given predicate return false.


String -> Any -> Bool

Predicate which returns true to include an attribute, or returns false to exclude it.


The attribute's name


The attribute's value

Returns true to include the attribute, false to exclude the attribute.


The attribute set to filter

Example 5.13. Filtering an attributeset

filterAttrs (n: v: n == "foo") { foo = 1; bar = 2; }
=> { foo = 1; } lib.attrsets.filterAttrsRecursive

filterAttrsRecursive :: (String -> Any -> Bool) -> AttrSet -> AttrSet

Located at lib/attrsets.nix:138 in <nixpkgs>.

Filter an attribute set recursively by removing all attributes for which the given predicate return false.


String -> Any -> Bool

Predicate which returns true to include an attribute, or returns false to exclude it.


The attribute's name


The attribute's value

Returns true to include the attribute, false to exclude the attribute.


The attribute set to filter

Example 5.14. Recursively filtering an attribute set

  (n: v: v != null)
    levelA = {
      example = "hi";
      levelB = {
        hello = "there";
        this-one-is-present = {
          this-is-excluded = null;
      this-one-is-also-excluded = null;
    also-excluded = null;
=> {
     levelA = {
       example = "hi";
       levelB = {
         hello = "there";
         this-one-is-present = { };
   } lib.attrsets.foldAttrs

foldAttrs :: (Any -> Any -> Any) -> Any -> [AttrSets] -> Any

Located at lib/attrsets.nix:157 in <nixpkgs>.

Apply fold function to values grouped by key.


Any -> Any -> Any

Given a value val and a collector col, combine the two.


An attribute's value


The result of previous op calls with other values and nul.


The null-value, the starting value.


A list of attribute sets to fold together by key.

Example 5.15. Combining an attribute of lists in to one attribute set

  (n: a: [n] ++ a) []
    { a = 2; b = 7; }
    { a = 3; }
    { b = 6; }
=> { a = [ 2 3 ]; b = [ 7 6 ]; } lib.attrsets.collect

collect :: (Any -> Bool) -> AttrSet -> [Any]

Located at lib/attrsets.nix:181 in <nixpkgs>.

Recursively collect sets that verify a given predicate named pred from the set attrs. The recursion stops when pred returns true.


Any -> Bool

Given an attribute's value, determine if recursion should stop.


The attribute set value.


The attribute set to recursively collect.

Example 5.16. Collecting all lists from an attribute set

lib.attrsets.collect isList { a = { b = ["b"]; }; c = [1]; }
=> [["b"] [1]]

Example 5.17. Collecting all attribute-sets which contain the outPath attribute name.

collect (x: x ? outPath)
  { a = { outPath = "a/"; }; b = { outPath = "b/"; }; }
=> [{ outPath = "a/"; } { outPath = "b/"; }] lib.attrsets.nameValuePair

nameValuePair :: String -> Any -> AttrSet

Located at lib/attrsets.nix:215 in <nixpkgs>.

Utility function that creates a {name, value} pair as expected by builtins.listToAttrs.


The attribute name.


The attribute value.

Example 5.18. Creating a name value pair

nameValuePair "some" 6
=> { name = "some"; value = 6; } lib.attrsets.mapAttrs

Located at lib/attrsets.nix:228 in <nixpkgs>.

Apply a function to each element in an attribute set, creating a new attribute set.

Provides a backwards-compatible interface of builtins.mapAttrs for Nix version older than 2.1.


String -> Any -> Any

Given an attribute's name and value, return a new value.


The name of the attribute.


The attribute's value.

Example 5.19. Modifying each value of an attribute set

  (name: value: name + "-" + value)
  { x = "foo"; y = "bar"; }
=> { x = "x-foo"; y = "y-bar"; } lib.attrsets.mapAttrs'

mapAttrs' :: (String -> Any -> { name = String; value = Any }) -> AttrSet -> AttrSet

Located at lib/attrsets.nix:242 in <nixpkgs>.

Like mapAttrs, but allows the name of each attribute to be changed in addition to the value. The applied function should return both the new name and value as a nameValuePair.


String -> Any -> { name = String; value = Any }

Given an attribute's name and value, return a new name value pair.


The name of the attribute.


The attribute's value.


The attribute set to map over.

Example 5.20. Change the name and value of each attribute of an attribute set

lib.attrsets.mapAttrs' (name: value: lib.attrsets.nameValuePair ("foo_" + name) ("bar-" + value))
   { x = "a"; y = "b"; }
=> { foo_x = "bar-a"; foo_y = "bar-b"; } lib.attrsets.mapAttrsToList

mapAttrsToList :: (String -> Any -> Any) -> AttrSet -> [Any]

Located at lib/attrsets.nix:258 in <nixpkgs>.

Call fn for each attribute in the given set and return the result in a list.


String -> Any -> Any

Given an attribute's name and value, return a new value.


The name of the attribute.


The attribute's value.


The attribute set to map over.

Example 5.21. Combine attribute values and names in to a list

lib.attrsets.mapAttrsToList (name: value: "${name}=${value}")
   { x = "a"; y = "b"; }
=> [ "x=a" "y=b" ] lib.attrsets.mapAttrsRecursive

mapAttrsRecursive :: ([String] > Any -> Any) -> AttrSet -> AttrSet

Located at lib/attrsets.nix:275 in <nixpkgs>.

Like mapAttrs, except that it recursively applies itself to attribute sets. Also, the first argument of the argument function is a list of the names of the containing attributes.


[ String ] -> Any -> Any

Given a list of attribute names and value, return a new value.


The list of attribute names to this value.

For example, the name_path for the example string in the attribute set { foo = { bar = "example"; }; } is [ "foo" "bar" ].


The attribute's value.


The attribute set to recursively map over.

Example 5.22. A contrived example of using lib.attrsets.mapAttrsRecursive

  (path: value: concatStringsSep "-" (path ++ [value]))
    n = {
      a = "A";
      m = {
        b = "B";
        c = "C";
    d = "D";
=> {
     n = {
       a = "n-a-A";
       m = {
         b = "n-m-b-B";
         c = "n-m-c-C";
     d = "d-D";
   } lib.attrsets.mapAttrsRecursiveCond

mapAttrsRecursiveCond :: (AttrSet -> Bool) -> ([ String ] -> Any -> Any) -> AttrSet -> AttrSet

Located at lib/attrsets.nix:296 in <nixpkgs>.

Like mapAttrsRecursive, but it takes an additional predicate function that tells it whether to recursive into an attribute set. If it returns false, mapAttrsRecursiveCond does not recurse, but does apply the map function. It is returns true, it does recurse, and does not apply the map function.


(AttrSet -> Bool)

Determine if mapAttrsRecursive should recurse deeper in to the attribute set.


An attribute set.


[ String ] -> Any -> Any

Given a list of attribute names and value, return a new value.


The list of attribute names to this value.

For example, the name_path for the example string in the attribute set { foo = { bar = "example"; }; } is [ "foo" "bar" ].


The attribute's value.


The attribute set to recursively map over.

Example 5.23. Only convert attribute values to JSON if the containing attribute set is marked for recursion

  ({ recurse ? false, ... }: recurse)
  (name: value: builtins.toJSON value)
    dorecur = {
      recurse = true;
      hello = "there";
    dontrecur = {
      converted-to- = "json";
=> {
     dorecur = {
       hello = "\"there\"";
       recurse = "true";
     dontrecur = "{\"converted-to\":\"json\"}";
   } lib.attrsets.genAttrs

genAttrs :: [ String ] -> (String -> Any) -> AttrSet

Located at lib/attrsets.nix:316 in <nixpkgs>.

Generate an attribute set by mapping a function over a list of attribute names.


Names of values in the resulting attribute set.


String -> Any

Takes the name of the attribute and return the attribute's value.


The name of the attribute to generate a value for.

Example 5.24. Generate an attrset based on names only

lib.attrsets.genAttrs [ "foo" "bar" ] (name: "x_${name}")
=> { foo = "x_foo"; bar = "x_bar"; } lib.attrsets.isDerivation

isDerivation :: Any -> Bool

Located at lib/attrsets.nix:330 in <nixpkgs>.

Check whether the argument is a derivation. Any set with { type = "derivation"; } counts as a derivation.


The value which is possibly a derivation.

Example 5.25. A package is a derivation

lib.attrsets.isDerivation (import <nixpkgs> {}).ruby
=> true

Example 5.26. Anything else is not a derivation

lib.attrsets.isDerivation "foobar"
=> false lib.attrsets.toDerivation

toDerivation :: Path -> Derivation

Located at lib/attrsets.nix:333 in <nixpkgs>.

Converts a store path to a fake derivation.


A store path to convert to a derivation. lib.attrsets.optionalAttrs

optionalAttrs :: Bool -> AttrSet

Located at lib/attrsets.nix:356 in <nixpkgs>.

Conditionally return an attribute set or an empty attribute set.


Condition under which the as attribute set is returned.


The attribute set to return if cond is true.

Example 5.27. Return the provided attribute set when cond is true

lib.attrsets.optionalAttrs true { my = "set"; }
=> { my = "set"; }

Example 5.28. Return an empty attribute set when cond is false

lib.attrsets.optionalAttrs false { my = "set"; }
=> { } lib.attrsets.zipAttrsWithNames

zipAttrsWithNames :: [ String ] -> (String -> [ Any ] -> Any) -> [ AttrSet ] -> AttrSet

Located at lib/attrsets.nix:366 in <nixpkgs>.

Merge sets of attributes and use the function f to merge attribute values where the attribute name is in names.


A list of attribute names to zip.


(String -> [ Any ] -> Any

Accepts an attribute name, all the values, and returns a combined value.


The name of the attribute each value came from.


A list of values collected from the list of attribute sets.


A list of attribute sets to zip together.

Example 5.29. Summing a list of attribute sets of numbers

  [ "a" "b" ]
  (name: vals: "${name} ${toString (builtins.foldl' (a: b: a + b) 0 vals)}")
    { a = 1; b = 1; c = 1; }
    { a = 10; }
    { b = 100; }
    { c = 1000; }
=> { a = "a 11"; b = "b 101"; } lib.attrsets.zipAttrsWith

zipAttrsWith :: (String -> [ Any ] -> Any) -> [ AttrSet ] -> AttrSet

Located at lib/attrsets.nix:381 in <nixpkgs>.

Merge sets of attributes and use the function f to merge attribute values. Similar to Section, “lib.attrsets.zipAttrsWithNames where all key names are passed for names.


(String -> [ Any ] -> Any

Accepts an attribute name, all the values, and returns a combined value.


The name of the attribute each value came from.


A list of values collected from the list of attribute sets.


A list of attribute sets to zip together.

Example 5.30. Summing a list of attribute sets of numbers

  (name: vals: "${name} ${toString (builtins.foldl' (a: b: a + b) 0 vals)}")
    { a = 1; b = 1; c = 1; }
    { a = 10; }
    { b = 100; }
    { c = 1000; }
=> { a = "a 11"; b = "b 101"; c = "c 1001"; } lib.attrsets.zipAttrs

zipAttrsWith :: [ AttrSet ] -> AttrSet

Located at lib/attrsets.nix:388 in <nixpkgs>.

Merge sets of attributes and combine each attribute value in to a list. Similar to Section, “lib.attrsets.zipAttrsWith where the merge function returns a list of all values.


A list of attribute sets to zip together.

Example 5.31. Combining a list of attribute sets

    { a = 1; b = 1; c = 1; }
    { a = 10; }
    { b = 100; }
    { c = 1000; }
=> { a = [ 1 10 ]; b = [ 1 100 ]; c = [ 1 1000 ]; } lib.attrsets.recursiveUpdateUntil

recursiveUpdateUntil :: ( [ String ] -> AttrSet -> AttrSet -> Bool ) -> AttrSet -> AttrSet -> AttrSet

Located at lib/attrsets.nix:418 in <nixpkgs>.

Does the same as the update operator // except that attributes are merged until the given predicate is verified. The predicate should accept 3 arguments which are the path to reach the attribute, a part of the first attribute set and a part of the second attribute set. When the predicate is verified, the value of the first attribute set is replaced by the value of the second attribute set.


[ String ] -> AttrSet -> AttrSet -> Bool


The path to the values in the left and right hand sides.


The left hand side value.


The right hand side value.


The left hand attribute set of the merge.


The right hand attribute set of the merge.

Example 5.32. Recursively merging two attribute sets

lib.attrsets.recursiveUpdateUntil (path: l: r: path == ["foo"])
    # first attribute set = 1;
    foo.baz = 2;
    bar = 3;
    #second attribute set = 1;
    foo.quz = 2;
    baz = 4;
=> { = 1; # 'foo.*' from the second set
  foo.quz = 2; #
  bar = 3;     # 'bar' from the first set
  baz = 4;     # 'baz' from the second set
} lib.attrsets.recursiveUpdate

recursiveUpdate :: AttrSet -> AttrSet -> AttrSet

Located at lib/attrsets.nix:449 in <nixpkgs>.

A recursive variant of the update operator //. The recursion stops when one of the attribute values is not an attribute set, in which case the right hand side value takes precedence over the left hand side value.


The left hand attribute set of the merge.


The right hand attribute set of the merge.

Example 5.33. Recursively merging two attribute sets

    boot.loader.grub.enable = true;
    boot.loader.grub.device = "/dev/hda";
    boot.loader.grub.device = "";
=> {
  boot.loader.grub.enable = true;
  boot.loader.grub.device = "";
} lib.attrsets.recurseIntoAttrs

recurseIntoAttrs :: AttrSet -> AttrSet

Located at lib/attrsets.nix:508 in <nixpkgs>.

Make various Nix tools consider the contents of the resulting attribute set when looking for what to build, find, etc.

This function only affects a single attribute set; it does not apply itself recursively for nested attribute sets.


An attribute set to scan for derivations.

Example 5.34. Making Nix look inside an attribute set

{ pkgs ? import <nixpkgs> {} }:
  myTools = pkgs.lib.recurseIntoAttrs {
    inherit (pkgs) hello figlet;
} lib.attrsets.cartesianProductOfSets

cartesianProductOfSets :: AttrSet -> [ AttrSet ]

Located at lib/attrsets.nix:200 in <nixpkgs>.

Return the cartesian product of attribute set value combinations.


An attribute set with attributes that carry lists of values.

Example 5.35. Creating the cartesian product of a list of attribute values

cartesianProductOfSets { a = [ 1 2 ]; b = [ 10 20 ]; }
=> [
     { a = 1; b = 10; }
     { a = 1; b = 20; }
     { a = 2; b = 10; }
     { a = 2; b = 20; }

5.1.3. String manipulation functions lib.strings.concatStrings

concatStrings :: [string] -> string

Concatenate a list of strings.

Example 5.36. lib.strings.concatStrings usage example

concatStrings ["foo" "bar"]
=> "foobar"

Located at lib/strings.nix:43 in <nixpkgs>. lib.strings.concatMapStrings

concatMapStrings :: (a -> string) -> [a] -> string

Map a function over a list and concatenate the resulting strings.


Function argument


Function argument

Example 5.37. lib.strings.concatMapStrings usage example

concatMapStrings (x: "a" + x) ["foo" "bar"]
=> "afooabar"

Located at lib/strings.nix:53 in <nixpkgs>. lib.strings.concatImapStrings

concatImapStrings :: (int -> a -> string) -> [a] -> string

Like `concatMapStrings` except that the f functions also gets the position as a parameter.


Function argument


Function argument

Example 5.38. lib.strings.concatImapStrings usage example

concatImapStrings (pos: x: "${toString pos}-${x}") ["foo" "bar"]
=> "1-foo2-bar"

Located at lib/strings.nix:64 in <nixpkgs>. lib.strings.intersperse

intersperse :: a -> [a] -> [a]

Place an element between each element of a list


Separator to add between elements


Input list

Example 5.39. lib.strings.intersperse usage example

intersperse "/" ["usr" "local" "bin"]
=> ["usr" "/" "local" "/" "bin"].

Located at lib/strings.nix:74 in <nixpkgs>. lib.strings.concatStringsSep

concatStringsSep :: string -> [string] -> string

Concatenate a list of strings with a separator between each element

Example 5.40. lib.strings.concatStringsSep usage example

concatStringsSep "/" ["usr" "local" "bin"]
=> "usr/local/bin"

Located at lib/strings.nix:91 in <nixpkgs>. lib.strings.concatMapStringsSep

concatMapStringsSep :: string -> (a -> string) -> [a] -> string

Maps a function over a list of strings and then concatenates the result with the specified separator interspersed between elements.


Separator to add between elements


Function to map over the list


List of input strings

Example 5.41. lib.strings.concatMapStringsSep usage example

concatMapStringsSep "-" (x: toUpper x)  ["foo" "bar" "baz"]

Located at lib/strings.nix:104 in <nixpkgs>. lib.strings.concatImapStringsSep

concatIMapStringsSep :: string -> (int -> a -> string) -> [a] -> string

Same as `concatMapStringsSep`, but the mapping function additionally receives the position of its argument.


Separator to add between elements


Function that receives elements and their positions


List of input strings

Example 5.42. lib.strings.concatImapStringsSep usage example

concatImapStringsSep "-" (pos: x: toString (x / pos)) [ 6 6 6 ]
=> "6-3-2"

Located at lib/strings.nix:121 in <nixpkgs>. lib.strings.makeSearchPath

makeSearchPath :: string -> [string] -> string

Construct a Unix-style, colon-separated search path consisting of the given `subDir` appended to each of the given paths.


Directory name to append


List of base paths

Example 5.43. lib.strings.makeSearchPath usage example

makeSearchPath "bin" ["/root" "/usr" "/usr/local"]
=> "/root/bin:/usr/bin:/usr/local/bin"
makeSearchPath "bin" [""]
=> "/bin"

Located at lib/strings.nix:140 in <nixpkgs>. lib.strings.makeSearchPathOutput

string -> string -> [package] -> string

Construct a Unix-style search path by appending the given `subDir` to the specified `output` of each of the packages. If no output by the given name is found, fallback to `.out` and then to the default.


Package output to use


Directory name to append


List of packages

Example 5.44. lib.strings.makeSearchPathOutput usage example

makeSearchPathOutput "dev" "bin" [ pkgs.openssl pkgs.zlib ]
=> "/nix/store/9rz8gxhzf8sw4kf2j2f1grr49w8zx5vj-openssl-1.0.1r-dev/bin:/nix/store/wwh7mhwh269sfjkm6k5665b5kgp7jrk2-zlib-1.2.8/bin"

Located at lib/strings.nix:158 in <nixpkgs>. lib.strings.makeLibraryPath

Construct a library search path (such as RPATH) containing the libraries for a set of packages

Example 5.45. lib.strings.makeLibraryPath usage example

makeLibraryPath [ "/usr" "/usr/local" ]
=> "/usr/lib:/usr/local/lib"
pkgs = import <nixpkgs> { }
makeLibraryPath [ pkgs.openssl pkgs.zlib ]
=> "/nix/store/9rz8gxhzf8sw4kf2j2f1grr49w8zx5vj-openssl-1.0.1r/lib:/nix/store/wwh7mhwh269sfjkm6k5665b5kgp7jrk2-zlib-1.2.8/lib"

Located at lib/strings.nix:176 in <nixpkgs>. lib.strings.makeBinPath

Construct a binary search path (such as $PATH) containing the binaries for a set of packages.

Example 5.46. lib.strings.makeBinPath usage example

makeBinPath ["/root" "/usr" "/usr/local"]
=> "/root/bin:/usr/bin:/usr/local/bin"

Located at lib/strings.nix:185 in <nixpkgs>. lib.strings.optionalString

optionalString :: bool -> string -> string

Depending on the boolean `cond', return either the given string or the empty string. Useful to concatenate against a bigger string.




String to return if condition is true

Example 5.47. lib.strings.optionalString usage example

optionalString true "some-string"
=> "some-string"
optionalString false "some-string"
=> ""

Located at lib/strings.nix:198 in <nixpkgs>. lib.strings.hasPrefix

hasPrefix :: string -> string -> bool

Determine whether a string has given prefix.


Prefix to check for


Input string

Example 5.48. lib.strings.hasPrefix usage example

hasPrefix "foo" "foobar"
=> true
hasPrefix "foo" "barfoo"
=> false

Located at lib/strings.nix:214 in <nixpkgs>. lib.strings.hasSuffix

hasSuffix :: string -> string -> bool

Determine whether a string has given suffix.


Suffix to check for


Input string

Example 5.49. lib.strings.hasSuffix usage example

hasSuffix "foo" "foobar"
=> false
hasSuffix "foo" "barfoo"
=> true

Located at lib/strings.nix:230 in <nixpkgs>. lib.strings.hasInfix

hasInfix :: string -> string -> bool

Determine whether a string contains the given infix


Function argument


Function argument

Example 5.50. lib.strings.hasInfix usage example

hasInfix "bc" "abcd"
=> true
hasInfix "ab" "abcd"
=> true
hasInfix "cd" "abcd"
=> true
hasInfix "foo" "abcd"
=> false

Located at lib/strings.nix:255 in <nixpkgs>. lib.strings.stringToCharacters

stringToCharacters :: string -> [string]

Convert a string to a list of characters (i.e. singleton strings). This allows you to, e.g., map a function over each character. However, note that this will likely be horribly inefficient; Nix is not a general purpose programming language. Complex string manipulations should, if appropriate, be done in a derivation. Also note that Nix treats strings as a list of bytes and thus doesn't handle unicode.


Function argument

Example 5.51. lib.strings.stringToCharacters usage example

stringToCharacters ""
=> [ ]
stringToCharacters "abc"
=> [ "a" "b" "c" ]
stringToCharacters "💩"
=> [ "�" "�" "�" "�" ]

Located at lib/strings.nix:279 in <nixpkgs>. lib.strings.stringAsChars

stringAsChars :: (string -> string) -> string -> string

Manipulate a string character by character and replace them by strings before concatenating the results.


Function to map over each individual character


Input string

Example 5.52. lib.strings.stringAsChars usage example

stringAsChars (x: if x == "a" then "i" else x) "nax"
=> "nix"

Located at lib/strings.nix:291 in <nixpkgs>. lib.strings.escape

escape :: [string] -> string -> string

Escape occurrence of the elements of `list` in `string` by prefixing it with a backslash.


Function argument

Example 5.53. lib.strings.escape usage example

escape ["(" ")"] "(foo)"
=> "\\(foo\\)"

Located at lib/strings.nix:308 in <nixpkgs>. lib.strings.escapeShellArg

escapeShellArg :: string -> string

Quote string to be used safely within the Bourne shell.


Function argument

Example 5.54. lib.strings.escapeShellArg usage example

escapeShellArg "esc'ape\nme"
=> "'esc'\\''ape\nme'"

Located at lib/strings.nix:318 in <nixpkgs>. lib.strings.escapeShellArgs

escapeShellArgs :: [string] -> string

Quote all arguments to be safely passed to the Bourne shell.

Example 5.55. lib.strings.escapeShellArgs usage example

escapeShellArgs ["one" "two three" "four'five"]
=> "'one' 'two three' 'four'\\''five'"

Located at lib/strings.nix:328 in <nixpkgs>. lib.strings.escapeNixString

string -> string

Turn a string into a Nix expression representing that string


Function argument

Example 5.56. lib.strings.escapeNixString usage example

escapeNixString "hello\${}\n"
=> "\"hello\\\${}\\n\""

Located at lib/strings.nix:338 in <nixpkgs>. lib.strings.escapeRegex

string -> string

Turn a string into an exact regular expression

Example 5.57. lib.strings.escapeRegex usage example

escapeRegex "[^a-z]*"
=> "\\[\\^a-z]\\*"

Located at lib/strings.nix:348 in <nixpkgs>. lib.strings.escapeNixIdentifier

string -> string

Quotes a string if it can't be used as an identifier directly.


Function argument

Example 5.58. lib.strings.escapeNixIdentifier usage example

escapeNixIdentifier "hello"
=> "hello"
escapeNixIdentifier "0abc"
=> "\"0abc\""

Located at lib/strings.nix:360 in <nixpkgs>. lib.strings.escapeXML

string -> string

Escapes a string such that it is safe to include verbatim in an XML document.

Example 5.59. lib.strings.escapeXML usage example

escapeXML ''"test" 'test' < & >''
=> "&quot;test&quot; &apos;test&apos; &lt; &amp; &gt;"

Located at lib/strings.nix:374 in <nixpkgs>. lib.strings.toLower

toLower :: string -> string

Converts an ASCII string to lower-case.

Example 5.60. lib.strings.toLower usage example

toLower "HOME"
=> "home"

Located at lib/strings.nix:404 in <nixpkgs>. lib.strings.toUpper

toUpper :: string -> string

Converts an ASCII string to upper-case.

Example 5.61. lib.strings.toUpper usage example

toUpper "home"
=> "HOME"

Located at lib/strings.nix:414 in <nixpkgs>. lib.strings.addContextFrom

Appends string context from another string. This is an implementation detail of Nix.

Strings in Nix carry an invisible `context` which is a list of strings representing store paths. If the string is later used in a derivation attribute, the derivation will properly populate the inputDrvs and inputSrcs.


Function argument


Function argument

Example 5.62. lib.strings.addContextFrom usage example

pkgs = import <nixpkgs> { };
addContextFrom pkgs.coreutils "bar"
=> "bar"

Located at lib/strings.nix:429 in <nixpkgs>. lib.strings.splitString

Cut a string with a separator and produces a list of strings which were separated by this separator.


Function argument


Function argument

Example 5.63. lib.strings.splitString usage example

splitString "." ""
=> [ "foo" "bar" "baz" ]
splitString "/" "/usr/local/bin"
=> [ "" "usr" "local" "bin" ]

Located at lib/strings.nix:440 in <nixpkgs>. lib.strings.removePrefix

string -> string -> string

Return a string without the specified prefix, if the prefix matches.


Prefix to remove if it matches


Input string

Example 5.64. lib.strings.removePrefix usage example

removePrefix "foo." ""
=> "bar.baz"
removePrefix "xxx" ""
=> ""

Located at lib/strings.nix:458 in <nixpkgs>. lib.strings.removeSuffix

string -> string -> string

Return a string without the specified suffix, if the suffix matches.


Suffix to remove if it matches


Input string

Example 5.65. lib.strings.removeSuffix usage example

removeSuffix "front" "homefront"
=> "home"
removeSuffix "xxx" "homefront"
=> "homefront"

Located at lib/strings.nix:482 in <nixpkgs>. lib.strings.versionOlder

Return true if string v1 denotes a version older than v2.


Function argument


Function argument

Example 5.66. lib.strings.versionOlder usage example

versionOlder "1.1" "1.2"
=> true
versionOlder "1.1" "1.1"
=> false

Located at lib/strings.nix:504 in <nixpkgs>. lib.strings.versionAtLeast

Return true if string v1 denotes a version equal to or newer than v2.


Function argument


Function argument

Example 5.67. lib.strings.versionAtLeast usage example

versionAtLeast "1.1" "1.0"
=> true
versionAtLeast "1.1" "1.1"
=> true
versionAtLeast "1.1" "1.2"
=> false

Located at lib/strings.nix:516 in <nixpkgs>. lib.strings.getName

This function takes an argument that's either a derivation or a derivation's "name" attribute and extracts the name part from that argument.


Function argument

Example 5.68. lib.strings.getName usage example

getName "youtube-dl-2016.01.01"
=> "youtube-dl"
=> "youtube-dl"

Located at lib/strings.nix:528 in <nixpkgs>. lib.strings.getVersion

This function takes an argument that's either a derivation or a derivation's "name" attribute and extracts the version part from that argument.


Function argument

Example 5.69. lib.strings.getVersion usage example

getVersion "youtube-dl-2016.01.01"
=> "2016.01.01"
=> "2016.01.01"

Located at lib/strings.nix:545 in <nixpkgs>. lib.strings.nameFromURL

Extract name with version from URL. Ask for separator which is supposed to start extension.


Function argument


Function argument

Example 5.70. lib.strings.nameFromURL usage example

nameFromURL "" "-"
=> "nix"
nameFromURL "" "_"
=> "nix-1.7-x86"

Located at lib/strings.nix:561 in <nixpkgs>. lib.strings.enableFeature

Create an --{enable,disable}-<feat> string that can be passed to standard GNU Autoconf scripts.


Function argument


Function argument

Example 5.71. lib.strings.enableFeature usage example

enableFeature true "shared"
=> "--enable-shared"
enableFeature false "shared"
=> "--disable-shared"

Located at lib/strings.nix:577 in <nixpkgs>. lib.strings.enableFeatureAs

Create an --{enable-<feat>=<value>,disable-<feat>} string that can be passed to standard GNU Autoconf scripts.


Function argument


Function argument


Function argument

Example 5.72. lib.strings.enableFeatureAs usage example

enableFeatureAs true "shared" "foo"
=> "--enable-shared=foo"
enableFeatureAs false "shared" (throw "ignored")
=> "--disable-shared"

Located at lib/strings.nix:590 in <nixpkgs>. lib.strings.withFeature

Create an --{with,without}-<feat> string that can be passed to standard GNU Autoconf scripts.


Function argument


Function argument

Example 5.73. lib.strings.withFeature usage example

withFeature true "shared"
=> "--with-shared"
withFeature false "shared"
=> "--without-shared"

Located at lib/strings.nix:601 in <nixpkgs>. lib.strings.withFeatureAs

Create an --{with-<feat>=<value>,without-<feat>} string that can be passed to standard GNU Autoconf scripts.


Function argument


Function argument


Function argument

Example 5.74. lib.strings.withFeatureAs usage example

withFeatureAs true "shared" "foo"
=> "--with-shared=foo"
withFeatureAs false "shared" (throw "ignored")
=> "--without-shared"

Located at lib/strings.nix:614 in <nixpkgs>. lib.strings.fixedWidthString

fixedWidthString :: int -> string -> string -> string

Create a fixed width string with additional prefix to match required width.

This function will fail if the input string is longer than the requested length.


Function argument


Function argument


Function argument

Example 5.75. lib.strings.fixedWidthString usage example

fixedWidthString 5 "0" (toString 15)
=> "00015"

Located at lib/strings.nix:628 in <nixpkgs>. lib.strings.fixedWidthNumber

Format a number adding leading zeroes up to fixed width.


Function argument


Function argument

Example 5.76. lib.strings.fixedWidthNumber usage example

fixedWidthNumber 5 15
=> "00015"

Located at lib/strings.nix:645 in <nixpkgs>. lib.strings.floatToString

Convert a float to a string, but emit a warning when precision is lost during the conversion


Function argument

Example 5.77. lib.strings.floatToString usage example

floatToString 0.000001
=> "0.000001"
floatToString 0.0000001
=> trace: warning: Imprecise conversion from float to string 0.000000

Located at lib/strings.nix:657 in <nixpkgs>. lib.strings.isCoercibleToString

Check whether a value can be coerced to a string


Function argument

Located at lib/strings.nix:664 in <nixpkgs>. lib.strings.isStorePath

Check whether a value is a store path.


Function argument

Example 5.78. lib.strings.isStorePath usage example

isStorePath "/nix/store/d945ibfx9x185xf04b890y4f9g3cbb63-python-2.7.11/bin/python"
=> false
isStorePath "/nix/store/d945ibfx9x185xf04b890y4f9g3cbb63-python-2.7.11"
=> true
isStorePath pkgs.python
=> true
isStorePath [] || isStorePath 42 || isStorePath {} || …
=> false

Located at lib/strings.nix:682 in <nixpkgs>. lib.strings.toInt

string -> int

Parse a string as an int.


Function argument

Example 5.79. lib.strings.toInt usage example

toInt "1337"
=> 1337
toInt "-4"
=> -4
toInt "3.14"
=> error: floating point JSON numbers are not supported

Located at lib/strings.nix:703 in <nixpkgs>. lib.strings.readPathsFromFile

Read a list of paths from `file`, relative to the `rootPath`. Lines beginning with `#` are treated as comments and ignored. Whitespace is significant.

NOTE: This function is not performant and should be avoided.

Example 5.80. lib.strings.readPathsFromFile usage example

readPathsFromFile /prefix
=> [ "/prefix/dlopen-resolv.patch" "/prefix/tzdir.patch"
"/prefix/dlopen-libXcursor.patch" "/prefix/dlopen-openssl.patch"
"/prefix/dlopen-dbus.patch" "/prefix/xdg-config-dirs.patch"
"/prefix/compose-search-path.patch" ]

Located at lib/strings.nix:724 in <nixpkgs>. lib.strings.fileContents

fileContents :: path -> string

Read the contents of a file removing the trailing \n


Function argument

Example 5.81. lib.strings.fileContents usage example

$ echo "1.0" > ./version

fileContents ./version
=> "1.0"

Located at lib/strings.nix:744 in <nixpkgs>. lib.strings.sanitizeDerivationName

sanitizeDerivationName :: String -> String

Creates a valid derivation name from a potentially invalid one.


Function argument

Example 5.82. lib.strings.sanitizeDerivationName usage example

sanitizeDerivationName "../ # foo"
=> ""
sanitizeDerivationName ""
=> "unknown"
sanitizeDerivationName pkgs.hello
=> "-nix-store-2g75chlbpxlrqn15zlby2dfh8hr9qwbk-hello-2.10"

Located at lib/strings.nix:759 in <nixpkgs>.

5.1.4. Miscellaneous functions

id :: a -> a

The identity function For when you need a function that does “nothing”.


The value to return

Located at lib/trivial.nix:12 in <nixpkgs>. lib.trivial.const

const :: a -> b -> a

The constant function

Ignores the second argument. If called with only one argument, constructs a function that always returns a static value.


Value to return


Value to ignore

Example 5.83. lib.trivial.const usage example

let f = const 5; in f 10
=> 5

Located at lib/trivial.nix:26 in <nixpkgs>. lib.trivial.pipe

pipe :: a -> [<functions>] -> <return type of last function>

Pipes a value through a list of functions, left to right.


Function argument


Function argument

Example 5.84. lib.trivial.pipe usage example

pipe 2 [
(x: x + 2)  # 2 + 2 = 4
(x: x * 2)  # 4 * 2 = 8
=> 8

# ideal to do text transformations
pipe [ "a/b" "a/c" ] [

# create the cp command
(map (file: ''cp "${src}/${file}" $out\n''))

# concatenate all commands into one string

# make that string into a nix derivation
(pkgs.runCommand "copy-to-out" {})

=> <drv which copies all files to $out>

The output type of each function has to be the input type
of the next function, and the last function returns the
final value.

Located at lib/trivial.nix:61 in <nixpkgs>. lib.trivial.concat

note please don’t add a function like `compose = flip pipe`. This would confuse users, because the order of the functions in the list is not clear. With pipe, it’s obvious that it goes first-to-last. With `compose`, not so much.


Function argument


Function argument

Located at lib/trivial.nix:80 in <nixpkgs>. lib.trivial.or

boolean “or”


Function argument


Function argument

Located at lib/trivial.nix:83 in <nixpkgs>. lib.trivial.and

boolean “and”


Function argument


Function argument

Located at lib/trivial.nix:86 in <nixpkgs>. lib.trivial.bitAnd

bitwise “and”

Located at lib/trivial.nix:89 in <nixpkgs>. lib.trivial.bitOr

bitwise “or”

Located at lib/trivial.nix:94 in <nixpkgs>. lib.trivial.bitXor

bitwise “xor”

Located at lib/trivial.nix:99 in <nixpkgs>. lib.trivial.bitNot

bitwise “not”

Located at lib/trivial.nix:104 in <nixpkgs>. lib.trivial.boolToString

boolToString :: bool -> string

Convert a boolean to a string.

This function uses the strings "true" and "false" to represent boolean values. Calling `toString` on a bool instead returns "1" and "" (sic!).


Function argument

Located at lib/trivial.nix:114 in <nixpkgs>. lib.trivial.mergeAttrs

Merge two attribute sets shallowly, right side trumps left

mergeAttrs :: attrs -> attrs -> attrs


Left attribute set


Right attribute set (higher precedence for equal keys)

Example 5.85. lib.trivial.mergeAttrs usage example

mergeAttrs { a = 1; b = 2; } { b = 3; c = 4; }
=> { a = 1; b = 3; c = 4; }

Located at lib/trivial.nix:124 in <nixpkgs>. lib.trivial.flip

flip :: (a -> b -> c) -> (b -> a -> c)

Flip the order of the arguments of a binary function.


Function argument


Function argument


Function argument

Example 5.86. lib.trivial.flip usage example

flip concat [1] [2]
=> [ 2 1 ]

Located at lib/trivial.nix:138 in <nixpkgs>. lib.trivial.mapNullable

Apply function if the supplied argument is non-null.


Function to call


Argument to check for null before passing it to `f`

Example 5.87. lib.trivial.mapNullable usage example

mapNullable (x: x+1) null
=> null
mapNullable (x: x+1) 22
=> 23

Located at lib/trivial.nix:148 in <nixpkgs>. lib.trivial.version

Returns the current full nixpkgs version number.

Located at lib/trivial.nix:164 in <nixpkgs>. lib.trivial.release

Returns the current nixpkgs release number as string.

Located at lib/trivial.nix:167 in <nixpkgs>. lib.trivial.codeName

Returns the current nixpkgs release code name.

On each release the first letter is bumped and a new animal is chosen starting with that new letter.

Located at lib/trivial.nix:174 in <nixpkgs>. lib.trivial.versionSuffix

Returns the current nixpkgs version suffix as string.

Located at lib/trivial.nix:177 in <nixpkgs>. lib.trivial.revisionWithDefault

revisionWithDefault :: string -> string

Attempts to return the the current revision of nixpkgs and returns the supplied default value otherwise.


Default value to return if revision can not be determined

Located at lib/trivial.nix:188 in <nixpkgs>. lib.trivial.inNixShell

inNixShell :: bool

Determine whether the function is being called from inside a Nix shell.

Located at lib/trivial.nix:206 in <nixpkgs>. lib.trivial.min

Return minimum of two numbers.


Function argument


Function argument

Located at lib/trivial.nix:212 in <nixpkgs>. lib.trivial.max

Return maximum of two numbers.


Function argument


Function argument

Located at lib/trivial.nix:215 in <nixpkgs>. lib.trivial.mod

Integer modulus


Function argument


Function argument

Example 5.88. lib.trivial.mod usage example

mod 11 10
=> 1
mod 1 10
=> 1

Located at lib/trivial.nix:225 in <nixpkgs>.

C-style comparisons

a < b, compare a b => -1 a == b, compare a b => 0 a > b, compare a b => 1


Function argument


Function argument

Located at lib/trivial.nix:236 in <nixpkgs>. lib.trivial.splitByAndCompare

(a -> bool) -> (a -> a -> int) -> (a -> a -> int) -> (a -> a -> int)

Split type into two subtypes by predicate `p`, take all elements of the first subtype to be less than all the elements of the second subtype, compare elements of a single subtype with `yes` and `no` respectively.




Comparison function if predicate holds for both values


Comparison function if predicate holds for neither value


First value to compare


Second value to compare

Example 5.89. lib.trivial.splitByAndCompare usage example

let cmp = splitByAndCompare (hasPrefix "foo") compare compare; in

cmp "a" "z" => -1
cmp "fooa" "fooz" => -1

cmp "f" "a" => 1
cmp "fooa" "a" => -1
# while
compare "fooa" "a" => 1

Located at lib/trivial.nix:261 in <nixpkgs>. lib.trivial.importJSON

Reads a JSON file.

Type :: path -> any


Function argument

Located at lib/trivial.nix:281 in <nixpkgs>. lib.trivial.importTOML

Reads a TOML file.

Type :: path -> any


Function argument

Located at lib/trivial.nix:288 in <nixpkgs>. lib.trivial.warn

string -> a -> a

Print a warning before returning the second argument. This function behaves like `builtins.trace`, but requires a string message and formats it as a warning, including the `warning: ` prefix.

To get a call stack trace and abort evaluation, set the environment variable `NIX_ABORT_ON_WARN=true` and set the Nix options `--option pure-eval false --show-trace`

Located at lib/trivial.nix:316 in <nixpkgs>. lib.trivial.warnIf

bool -> string -> a -> a

Like warn, but only warn when the first argument is `true`.


Function argument


Function argument

Located at lib/trivial.nix:326 in <nixpkgs>. lib.trivial.setFunctionArgs

Add metadata about expected function arguments to a function. The metadata should match the format given by builtins.functionArgs, i.e. a set from expected argument to a bool representing whether that argument has a default or not. setFunctionArgs : (a → b) → Map String Bool → (a → b)

This function is necessary because you can't dynamically create a function of the { a, b ? foo, ... }: format, but some facilities like callPackage expect to be able to query expected arguments.


Function argument


Function argument

Located at lib/trivial.nix:344 in <nixpkgs>. lib.trivial.functionArgs

Extract the expected function arguments from a function. This works both with nix-native { a, b ? foo, ... }: style functions and functions with args set with 'setFunctionArgs'. It has the same return type and semantics as builtins.functionArgs. setFunctionArgs : (a → b) → Map String Bool.


Function argument

Located at lib/trivial.nix:356 in <nixpkgs>. lib.trivial.isFunction

Check whether something is a function or something annotated with function args.


Function argument

Located at lib/trivial.nix:364 in <nixpkgs>. lib.trivial.toHexString

Convert the given positive integer to a string of its hexadecimal representation. For example:

toHexString 0 => "0"

toHexString 16 => "10"

toHexString 250 => "FA"


Function argument

Located at lib/trivial.nix:376 in <nixpkgs>. lib.trivial.toBaseDigits

`toBaseDigits base i` converts the positive integer i to a list of its digits in the given base. For example:

toBaseDigits 10 123 => [ 1 2 3 ]

toBaseDigits 2 6 => [ 1 1 0 ]

toBaseDigits 16 250 => [ 15 10 ]


Function argument


Function argument

Located at lib/trivial.nix:402 in <nixpkgs>.

5.1.5. List manipulation functions lib.lists.singleton

singleton :: a -> [a]

Create a list consisting of a single element. `singleton x` is sometimes more convenient with respect to indentation than `[x]` when x spans multiple lines.


Function argument

Example 5.90. lib.lists.singleton usage example

singleton "foo"
=> [ "foo" ]

Located at lib/lists.nix:22 in <nixpkgs>. lib.lists.forEach

forEach :: [a] -> (a -> b) -> [b]

Apply the function to each element in the list. Same as `map`, but arguments flipped.


Function argument


Function argument

Example 5.91. lib.lists.forEach usage example

forEach [ 1 2 ] (x:
toString x
=> [ "1" "2" ]

Located at lib/lists.nix:35 in <nixpkgs>. lib.lists.foldr

foldr :: (a -> b -> b) -> b -> [a] -> b

“right fold” a binary function `op` between successive elements of `list` with `nul' as the starting value, i.e., `foldr op nul [x_1 x_2 ... x_n] == op x_1 (op x_2 ... (op x_n nul))`.


Function argument


Function argument


Function argument

Example 5.92. lib.lists.foldr usage example

concat = foldr (a: b: a + b) "z"
concat [ "a" "b" "c" ]
=> "abcz"
# different types
strange = foldr (int: str: toString (int + 1) + str) "a"
strange [ 1 2 3 4 ]
=> "2345a"

Located at lib/lists.nix:52 in <nixpkgs>. lib.lists.fold

`fold` is an alias of `foldr` for historic reasons

Located at lib/lists.nix:63 in <nixpkgs>. lib.lists.foldl

foldl :: (b -> a -> b) -> b -> [a] -> b

“left fold”, like `foldr`, but from the left: `foldl op nul [x_1 x_2 ... x_n] == op (... (op (op nul x_1) x_2) ... x_n)`.


Function argument


Function argument


Function argument

Example 5.93. lib.lists.foldl usage example

lconcat = foldl (a: b: a + b) "z"
lconcat [ "a" "b" "c" ]
=> "zabc"
# different types
lstrange = foldl (str: int: str + toString (int + 1)) "a"
lstrange [ 1 2 3 4 ]
=> "a2345"

Located at lib/lists.nix:80 in <nixpkgs>. lib.lists.foldl'

foldl' :: (b -> a -> b) -> b -> [a] -> b

Strict version of `foldl`.

The difference is that evaluation is forced upon access. Usually used with small whole results (in contrast with lazily-generated list or large lists where only a part is consumed.)

Located at lib/lists.nix:96 in <nixpkgs>. lib.lists.imap0

imap0 :: (int -> a -> b) -> [a] -> [b]

Map with index starting from 0


Function argument


Function argument

Example 5.94. lib.lists.imap0 usage example

imap0 (i: v: "${v}-${toString i}") ["a" "b"]
=> [ "a-0" "b-1" ]

Located at lib/lists.nix:106 in <nixpkgs>. lib.lists.imap1

imap1 :: (int -> a -> b) -> [a] -> [b]

Map with index starting from 1


Function argument


Function argument

Example 5.95. lib.lists.imap1 usage example

imap1 (i: v: "${v}-${toString i}") ["a" "b"]
=> [ "a-1" "b-2" ]

Located at lib/lists.nix:116 in <nixpkgs>. lib.lists.concatMap

concatMap :: (a -> [b]) -> [a] -> [b]

Map and concatenate the result.

Example 5.96. lib.lists.concatMap usage example

concatMap (x: [x] ++ ["z"]) ["a" "b"]
=> [ "a" "z" "b" "z" ]

Located at lib/lists.nix:126 in <nixpkgs>. lib.lists.flatten

Flatten the argument into a single list; that is, nested lists are spliced into the top-level lists.


Function argument

Example 5.97. lib.lists.flatten usage example

flatten [1 [2 [3] 4] 5]
=> [1 2 3 4 5]
flatten 1
=> [1]

Located at lib/lists.nix:137 in <nixpkgs>. lib.lists.remove

remove :: a -> [a] -> [a]

Remove elements equal to 'e' from a list. Useful for buildInputs.


Element to remove from the list

Example 5.98. lib.lists.remove usage example

remove 3 [ 1 3 4 3 ]
=> [ 1 4 ]

Located at lib/lists.nix:150 in <nixpkgs>. lib.lists.findSingle

findSingle :: (a -> bool) -> a -> a -> [a] -> a

Find the sole element in the list matching the specified predicate, returns `default` if no such element exists, or `multiple` if there are multiple matching elements.




Default value to return if element was not found.


Default value to return if more than one element was found


Input list

Example 5.99. lib.lists.findSingle usage example

findSingle (x: x == 3) "none" "multiple" [ 1 3 3 ]
=> "multiple"
findSingle (x: x == 3) "none" "multiple" [ 1 3 ]
=> 3
findSingle (x: x == 3) "none" "multiple" [ 1 9 ]
=> "none"

Located at lib/lists.nix:168 in <nixpkgs>. lib.lists.findFirst

findFirst :: (a -> bool) -> a -> [a] -> a

Find the first element in the list matching the specified predicate or return `default` if no such element exists.




Default value to return


Input list

Example 5.100. lib.lists.findFirst usage example

findFirst (x: x > 3) 7 [ 1 6 4 ]
=> 6
findFirst (x: x > 9) 7 [ 1 6 4 ]
=> 7

Located at lib/lists.nix:193 in <nixpkgs>. lib.lists.any

any :: (a -> bool) -> [a] -> bool

Return true if function `pred` returns true for at least one element of `list`.

Example 5.101. lib.lists.any usage example

any isString [ 1 "a" { } ]
=> true
any isString [ 1 { } ]
=> false

Located at lib/lists.nix:214 in <nixpkgs>. lib.lists.all

all :: (a -> bool) -> [a] -> bool

Return true if function `pred` returns true for all elements of `list`.

Example 5.102. lib.lists.all usage example

all (x: x < 3) [ 1 2 ]
=> true
all (x: x < 3) [ 1 2 3 ]
=> false

Located at lib/lists.nix:227 in <nixpkgs>. lib.lists.count

count :: (a -> bool) -> [a] -> int

Count how many elements of `list` match the supplied predicate function.



Example 5.103. lib.lists.count usage example

count (x: x == 3) [ 3 2 3 4 6 ]
=> 2

Located at lib/lists.nix:238 in <nixpkgs>. lib.lists.optional

optional :: bool -> a -> [a]

Return a singleton list or an empty list, depending on a boolean value. Useful when building lists with optional elements (e.g. `++ optional (system == "i686-linux") firefox').


Function argument


Function argument

Example 5.104. lib.lists.optional usage example

optional true "foo"
=> [ "foo" ]
optional false "foo"
=> [ ]

Located at lib/lists.nix:254 in <nixpkgs>. lib.lists.optionals

optionals :: bool -> [a] -> [a]

Return a list or an empty list, depending on a boolean value.




List to return if condition is true

Example 5.105. lib.lists.optionals usage example

optionals true [ 2 3 ]
=> [ 2 3 ]
optionals false [ 2 3 ]
=> [ ]

Located at lib/lists.nix:266 in <nixpkgs>. lib.lists.toList

If argument is a list, return it; else, wrap it in a singleton list. If you're using this, you should almost certainly reconsider if there isn't a more "well-typed" approach.


Function argument

Example 5.106. lib.lists.toList usage example

toList [ 1 2 ]
=> [ 1 2 ]
toList "hi"
=> [ "hi "]

Located at lib/lists.nix:283 in <nixpkgs>. lib.lists.range

range :: int -> int -> [int]

Return a list of integers from `first' up to and including `last'.


First integer in the range


Last integer in the range

Example 5.107. lib.lists.range usage example

range 2 4
=> [ 2 3 4 ]
range 3 2
=> [ ]

Located at lib/lists.nix:295 in <nixpkgs>. lib.lists.partition

(a -> bool) -> [a] -> { right :: [a], wrong :: [a] }

Splits the elements of a list in two lists, `right` and `wrong`, depending on the evaluation of a predicate.

Example 5.108. lib.lists.partition usage example

partition (x: x > 2) [ 5 1 2 3 4 ]
=> { right = [ 5 3 4 ]; wrong = [ 1 2 ]; }

Located at lib/lists.nix:314 in <nixpkgs>. lib.lists.groupBy'

Splits the elements of a list into many lists, using the return value of a predicate. Predicate should return a string which becomes keys of attrset `groupBy' returns.

`groupBy'` allows to customise the combining function and initial value


Function argument


Function argument


Function argument


Function argument

Example 5.109. lib.lists.groupBy' usage example

groupBy (x: boolToString (x > 2)) [ 5 1 2 3 4 ]
=> { true = [ 5 3 4 ]; false = [ 1 2 ]; }
groupBy (x: [ {name = "icewm"; script = "icewm &";}
{name = "xfce";  script = "xfce4-session &";}
{name = "icewm"; script = "icewmbg &";}
{name = "mate";  script = "gnome-session &";}
=> { icewm = [ { name = "icewm"; script = "icewm &"; }
{ name = "icewm"; script = "icewmbg &"; } ];
mate  = [ { name = "mate";  script = "gnome-session &"; } ];
xfce  = [ { name = "xfce";  script = "xfce4-session &"; } ];

groupBy' builtins.add 0 (x: boolToString (x > 2)) [ 5 1 2 3 4 ]
=> { true = 12; false = 3; }

Located at lib/lists.nix:343 in <nixpkgs>. lib.lists.zipListsWith

zipListsWith :: (a -> b -> c) -> [a] -> [b] -> [c]

Merges two lists of the same size together. If the sizes aren't the same the merging stops at the shortest. How both lists are merged is defined by the first argument.


Function to zip elements of both lists


First list


Second list

Example 5.110. lib.lists.zipListsWith usage example

zipListsWith (a: b: a + b) ["h" "l"] ["e" "o"]
=> ["he" "lo"]

Located at lib/lists.nix:363 in <nixpkgs>. lib.lists.zipLists

zipLists :: [a] -> [b] -> [{ fst :: a, snd :: b}]

Merges two lists of the same size together. If the sizes aren't the same the merging stops at the shortest.

Example 5.111. lib.lists.zipLists usage example

zipLists [ 1 2 ] [ "a" "b" ]
=> [ { fst = 1; snd = "a"; } { fst = 2; snd = "b"; } ]

Located at lib/lists.nix:382 in <nixpkgs>. lib.lists.reverseList

reverseList :: [a] -> [a]

Reverse the order of the elements of a list.


Function argument

Example 5.112. lib.lists.reverseList usage example

reverseList [ "b" "o" "j" ]
=> [ "j" "o" "b" ]

Located at lib/lists.nix:393 in <nixpkgs>. lib.lists.listDfs

Depth-First Search (DFS) for lists `list != []`.

`before a b == true` means that `b` depends on `a` (there's an edge from `b` to `a`).


Function argument


Function argument


Function argument

Example 5.113. lib.lists.listDfs usage example

listDfs true hasPrefix [ "/home/user" "other" "/" "/home" ]
== { minimal = "/";                  # minimal element
visited = [ "/home/user" ];     # seen elements (in reverse order)
rest    = [ "/home" "other" ];  # everything else

listDfs true hasPrefix [ "/home/user" "other" "/" "/home" "/" ]
== { cycle   = "/";                  # cycle encountered at this element
loops   = [ "/" ];              # and continues to these elements
visited = [ "/" "/home/user" ]; # elements leading to the cycle (in reverse order)
rest    = [ "/home" "other" ];  # everything else

Located at lib/lists.nix:415 in <nixpkgs>. lib.lists.toposort

Sort a list based on a partial ordering using DFS. This implementation is O(N^2), if your ordering is linear, use `sort` instead.

`before a b == true` means that `b` should be after `a` in the result.


Function argument


Function argument

Example 5.114. lib.lists.toposort usage example

toposort hasPrefix [ "/home/user" "other" "/" "/home" ]
== { result = [ "/" "/home" "/home/user" "other" ]; }

toposort hasPrefix [ "/home/user" "other" "/" "/home" "/" ]
== { cycle = [ "/home/user" "/" "/" ]; # path leading to a cycle
loops = [ "/" ]; }                # loops back to these elements

toposort hasPrefix [ "other" "/home/user" "/home" "/" ]
== { result = [ "other" "/" "/home" "/home/user" ]; }

toposort (a: b: a < b) [ 3 2 1 ] == { result = [ 1 2 3 ]; }

Located at lib/lists.nix:454 in <nixpkgs>. lib.lists.sort

Sort a list based on a comparator function which compares two elements and returns true if the first argument is strictly below the second argument. The returned list is sorted in an increasing order. The implementation does a quick-sort.

Example 5.115. lib.lists.sort usage example

sort (a: b: a < b) [ 5 3 7 ]
=> [ 3 5 7 ]

Located at lib/lists.nix:482 in <nixpkgs>. lib.lists.compareLists

Compare two lists element-by-element.


Function argument


Function argument


Function argument

Example 5.116. lib.lists.compareLists usage example

compareLists compare [] []
=> 0
compareLists compare [] [ "a" ]
=> -1
compareLists compare [ "a" ] []
=> 1
compareLists compare [ "a" "b" ] [ "a" "c" ]
=> 1

Located at lib/lists.nix:511 in <nixpkgs>. lib.lists.naturalSort

Sort list using "Natural sorting". Numeric portions of strings are sorted in numeric order.


Function argument

Example 5.117. lib.lists.naturalSort usage example

naturalSort ["disk11" "disk8" "disk100" "disk9"]
=> ["disk8" "disk9" "disk11" "disk100"]
naturalSort ["" "" ""]
=> ["" "" ""]
naturalSort ["v0.2" "v0.15" "v0.0.9"]
=> [ "v0.0.9" "v0.2" "v0.15" ]

Located at lib/lists.nix:534 in <nixpkgs>. lib.lists.take

take :: int -> [a] -> [a]

Return the first (at most) N elements of a list.


Number of elements to take

Example 5.118. lib.lists.take usage example

take 2 [ "a" "b" "c" "d" ]
=> [ "a" "b" ]
take 2 [ ]
=> [ ]

Located at lib/lists.nix:552 in <nixpkgs>. lib.lists.drop

drop :: int -> [a] -> [a]

Remove the first (at most) N elements of a list.


Number of elements to drop


Input list

Example 5.119. lib.lists.drop usage example

drop 2 [ "a" "b" "c" "d" ]
=> [ "c" "d" ]
drop 2 [ ]
=> [ ]

Located at lib/lists.nix:566 in <nixpkgs>. lib.lists.sublist

sublist :: int -> int -> [a] -> [a]

Return a list consisting of at most `count` elements of `list`, starting at index `start`.


Index at which to start the sublist


Number of elements to take


Input list

Example 5.120. lib.lists.sublist usage example

sublist 1 3 [ "a" "b" "c" "d" "e" ]
=> [ "b" "c" "d" ]
sublist 1 3 [ ]
=> [ ]

Located at lib/lists.nix:583 in <nixpkgs>. lib.lists.last

last :: [a] -> a

Return the last element of a list.

This function throws an error if the list is empty.


Function argument

Example 5.121. lib.lists.last usage example

last [ 1 2 3 ]
=> 3

Located at lib/lists.nix:607 in <nixpkgs>. lib.lists.init

init :: [a] -> [a]

Return all elements but the last.

This function throws an error if the list is empty.


Function argument

Example 5.122. lib.lists.init usage example

init [ 1 2 3 ]
=> [ 1 2 ]

Located at lib/lists.nix:621 in <nixpkgs>. lib.lists.crossLists

Return the image of the cross product of some lists by a function.

Example 5.123. lib.lists.crossLists usage example

crossLists (x:y: "${toString x}${toString y}") [[1 2] [3 4]]
=> [ "13" "14" "23" "24" ]

Located at lib/lists.nix:632 in <nixpkgs>. lib.lists.unique

unique :: [a] -> [a]

Remove duplicate elements from the list. O(n^2) complexity.

Example 5.124. lib.lists.unique usage example

unique [ 3 2 3 4 ]
=> [ 3 2 4 ]

Located at lib/lists.nix:645 in <nixpkgs>. lib.lists.intersectLists

Intersects list 'e' and another list. O(nm) complexity.


Function argument

Example 5.125. lib.lists.intersectLists usage example

intersectLists [ 1 2 3 ] [ 6 3 2 ]
=> [ 3 2 ]

Located at lib/lists.nix:653 in <nixpkgs>. lib.lists.subtractLists

Subtracts list 'e' from another list. O(nm) complexity.


Function argument

Example 5.126. lib.lists.subtractLists usage example

subtractLists [ 3 2 ] [ 1 2 3 4 5 3 ]
=> [ 1 4 5 ]

Located at lib/lists.nix:661 in <nixpkgs>. lib.lists.mutuallyExclusive

Test if two lists have no common element. It should be slightly more efficient than (intersectLists a b == [])


Function argument


Function argument

Located at lib/lists.nix:666 in <nixpkgs>.

5.1.6. Debugging functions lib.debug.traceIf

traceIf :: bool -> string -> a -> a

Conditionally trace the supplied message, based on a predicate.


Predicate to check


Message that should be traced


Value to return

Example 5.127. lib.debug.traceIf usage example

traceIf true "hello" 3
trace: hello
=> 3

Located at lib/debug.nix:51 in <nixpkgs>. lib.debug.traceValFn

traceValFn :: (a -> b) -> a -> a

Trace the supplied value after applying a function to it, and return the original value.


Function to apply


Value to trace and return

Example 5.128. lib.debug.traceValFn usage example

traceValFn (v: "mystring ${v}") "foo"
trace: mystring foo
=> "foo"

Located at lib/debug.nix:69 in <nixpkgs>. lib.debug.traceVal

traceVal :: a -> a

Trace the supplied value and return it.

Example 5.129. lib.debug.traceVal usage example

traceVal 42
# trace: 42
=> 42

Located at lib/debug.nix:84 in <nixpkgs>. lib.debug.traceSeq

traceSeq :: a -> b -> b

`builtins.trace`, but the value is `builtins.deepSeq`ed first.


The value to trace


The value to return

Example 5.130. lib.debug.traceSeq usage example

trace { a.b.c = 3; } null
trace: { a = <CODE>; }
=> null
traceSeq { a.b.c = 3; } null
trace: { a = { b = { c = 3; }; }; }
=> null

Located at lib/debug.nix:98 in <nixpkgs>. lib.debug.traceSeqN

Like `traceSeq`, but only evaluate down to depth n. This is very useful because lots of `traceSeq` usages lead to an infinite recursion.


Function argument


Function argument


Function argument

Example 5.131. lib.debug.traceSeqN usage example

traceSeqN 2 { a.b.c = 3; } null
trace: { a = { b = {…}; }; }
=> null

Located at lib/debug.nix:113 in <nixpkgs>. lib.debug.traceValSeqFn

A combination of `traceVal` and `traceSeq` that applies a provided function to the value to be traced after `deepSeq`ing it.


Function to apply


Value to trace

Located at lib/debug.nix:130 in <nixpkgs>. lib.debug.traceValSeq

A combination of `traceVal` and `traceSeq`.

Located at lib/debug.nix:137 in <nixpkgs>. lib.debug.traceValSeqNFn

A combination of `traceVal` and `traceSeqN` that applies a provided function to the value to be traced.


Function to apply


Function argument


Value to trace

Located at lib/debug.nix:141 in <nixpkgs>. lib.debug.traceValSeqN

A combination of `traceVal` and `traceSeqN`.

Located at lib/debug.nix:149 in <nixpkgs>. lib.debug.traceFnSeqN

Trace the input and output of a function `f` named `name`, both down to `depth`.

This is useful for adding around a function call, to see the before/after of values as they are transformed.


Function argument


Function argument


Function argument


Function argument

Example 5.132. lib.debug.traceFnSeqN usage example

traceFnSeqN 2 "id" (x: x) { a.b.c = 3; }
trace: { fn = "id"; from = { a.b = {…}; }; to = { a.b = {…}; }; }
=> { a.b.c = 3; }

Located at lib/debug.nix:162 in <nixpkgs>. lib.debug.runTests

Evaluate a set of tests. A test is an attribute set `{expr, expected}`, denoting an expression and its expected result. The result is a list of failed tests, each represented as `{name, expected, actual}`, denoting the attribute name of the failing test and its expected and actual results.

Used for regression testing of the functions in lib; see tests.nix for an example. Only tests having names starting with "test" are run.

Add attr { tests = ["testName"]; } to run these tests only.


Tests to run

Located at lib/debug.nix:188 in <nixpkgs>. lib.debug.testAllTrue

Create a test assuming that list elements are `true`.


Function argument

Example 5.133. lib.debug.testAllTrue usage example

{ testX = allTrue [ true ]; }

Located at lib/debug.nix:204 in <nixpkgs>.

5.1.7. NixOS / nixpkgs option handling lib.options.isOption

isOption :: a -> bool

Returns true when the given argument is an option

Example 5.134. lib.options.isOption usage example

isOption 1             // => false
isOption (mkOption {}) // => true

Located at lib/options.nix:49 in <nixpkgs>. lib.options.mkOption

Creates an Option attribute set. mkOption accepts an attribute set with the following keys:

All keys default to `null` when not given.


Structured function argument


Default value used when no definition is given in the configuration.


Textual representation of the default, for the manual.


Example value used in the manual.


String describing the option.


Related packages used in the manual (see `genRelatedPackages` in ../nixos/lib/make-options-doc/default.nix).


Option type, providing type-checking and value merging.


Function that converts the option value to something else.


Whether the option is for NixOS developers only.


Whether the option shows up in the manual. Default: true. Use false to hide the option and any sub-options from submodules. Use "shallow" to hide only sub-options.


Whether the option can be set only once


Deprecated, used by types.optionSet.

Example 5.135. lib.options.mkOption usage example

mkOption { }  // => { _type = "option"; }
mkOption { default = "foo"; } // => { _type = "option"; default = "foo"; }

Located at lib/options.nix:59 in <nixpkgs>. lib.options.mkEnableOption

Creates an Option attribute set for a boolean value option i.e an option to be toggled on or off:


Name for the created option

Example 5.136. lib.options.mkEnableOption usage example

mkEnableOption "foo"
=> { _type = "option"; default = false; description = "Whether to enable foo."; example = true; type = { ... }; }

Located at lib/options.nix:93 in <nixpkgs>. lib.options.mkSinkUndeclaredOptions

This option accepts anything, but it does not produce any result.

This is useful for sharing a module across different module sets without having to implement similar features as long as the values of the options are not accessed.


Function argument

Located at lib/options.nix:107 in <nixpkgs>. lib.options.mergeEqualOption

"Merge" option definitions by checking that they all have the same value.


Function argument


Function argument

Located at lib/options.nix:138 in <nixpkgs>. lib.options.getValues

getValues :: [ { value :: a } ] -> [a]

Extracts values of all "value" keys of the given list.

Example 5.137. lib.options.getValues usage example

getValues [ { value = 1; } { value = 2; } ] // => [ 1 2 ]
getValues [ ]                               // => [ ]

Located at lib/options.nix:158 in <nixpkgs>. lib.options.getFiles

getFiles :: [ { file :: a } ] -> [a]

Extracts values of all "file" keys of the given list

Example 5.138. lib.options.getFiles usage example

getFiles [ { file = "file1"; } { file = "file2"; } ] // => [ "file1" "file2" ]
getFiles [ ]                                         // => [ ]

Located at lib/options.nix:168 in <nixpkgs>. lib.options.scrubOptionValue

This function recursively removes all derivation attributes from `x` except for the `name` attribute.

This is to make the generation of `options.xml` much more efficient: the XML representation of derivations is very large (on the order of megabytes) and is not actually used by the manual generator.


Function argument

Located at lib/options.nix:211 in <nixpkgs>. lib.options.literalExpression

For use in the `defaultText` and `example` option attributes. Causes the given string to be rendered verbatim in the documentation as Nix code. This is necessary for complex values, e.g. functions, or values that depend on other values or packages.


Function argument

Located at lib/options.nix:224 in <nixpkgs>. lib.options.literalDocBook

For use in the `defaultText` and `example` option attributes. Causes the given DocBook text to be inserted verbatim in the documentation, for when a `literalExpression` would be too hard to read.


Function argument

Located at lib/options.nix:235 in <nixpkgs>. lib.options.showOption

Convert an option, described as a list of the option parts in to a safe, human readable version.


Function argument

Example 5.139. lib.options.showOption usage example

(showOption ["foo" "bar" "baz"]) == ""
(showOption ["foo" "bar.baz" "tux"]) == ""

Placeholders will not be quoted as they are not actual values:
(showOption ["foo" "*" "bar"]) == "foo.*.bar"
(showOption ["foo" "<name>" "bar"]) == "foo.<name>.bar"

Unlike attributes, options can also start with numbers:
(showOption ["windowManager" "2bwm" "enable"]) == "windowManager.2bwm.enable"

Located at lib/options.nix:255 in <nixpkgs>.

5.1.8. Source filtering functions lib.sources.trace

sources.trace :: sourceLike -> Source

Add logging to a source, for troubleshooting the filtering behavior.


Source to debug. The returned source will behave like this source, but also log its filter invocations.

Located at lib/sources.nix:246 in <nixpkgs>. lib.sources.sourceFilesBySuffices

sourceLike -> [String] -> Source

Get all files ending with the specified suffices from the given source directory or its descendants, omitting files that do not match any suffix. The result of the example below will include files like `./dir/module.c` and `./dir/subdir/doc.xml` if present.


Path or source containing the files to be returned


A list of file suffix strings

Example 5.140. lib.sources.sourceFilesBySuffices usage example

sourceFilesBySuffices ./. [ ".xml" ".c" ]

Located at lib/sources.nix:246 in <nixpkgs>.

5.2. Generators

Generators are functions that create file formats from nix data structures, e. g. for configuration files. There are generators available for: INI, JSON and YAML

All generators follow a similar call interface: generatorName configFunctions data, where configFunctions is an attrset of user-defined functions that format nested parts of the content. They each have common defaults, so often they do not need to be set manually. An example is mkSectionName ? (name: libStr.escape [ "[" "]" ] name) from the INI generator. It receives the name of a section and sanitizes it. The default mkSectionName escapes [ and ] with a backslash.

Generators can be fine-tuned to produce exactly the file format required by your application/service. One example is an INI-file format which uses : as separator, the strings "yes"/"no" as boolean values and requires all string values to be quoted:

with lib;
  customToINI = generators.toINI {
    # specifies how to format a key/value pair
    mkKeyValue = generators.mkKeyValueDefault {
      # specifies the generated string for a subset of nix values
      mkValueString = v:
             if v == true then ''"yes"''
        else if v == false then ''"no"''
        else if isString v then ''"${v}"''
        # and delegats all other values to the default generator
        else generators.mkValueStringDefault {} v;
    } ":";

# the INI file can now be given as plain old nix values
in customToINI {
  main = {
    pushinfo = true;
    autopush = false;
    host = "localhost";
    port = 42;
  mergetool = {
    merge = "diff3";

This will produce the following INI file as nix string:


Note: Nix store paths can be converted to strings by enclosing a derivation attribute like so: "${drv}".

Detailed documentation for each generator can be found in lib/generators.nix.

5.3. Debugging Nix Expressions

Nix is a unityped, dynamic language, this means every value can potentially appear anywhere. Since it is also non-strict, evaluation order and what ultimately is evaluated might surprise you. Therefore it is important to be able to debug nix expressions.

In the lib/debug.nix file you will find a number of functions that help (pretty-)printing values while evaluation is running. You can even specify how deep these values should be printed recursively, and transform them on the fly. Please consult the docstrings in lib/debug.nix for usage information.

5.4. prefer-remote-fetch overlay

prefer-remote-fetch is an overlay that download sources on remote builder. This is useful when the evaluating machine has a slow upload while the builder can fetch faster directly from the source. To use it, put the following snippet as a new overlay:

self: super:
  (super.prefer-remote-fetch self super)

A full configuration example for that sets the overlay up for your own account, could look like this

$ mkdir ~/.config/nixpkgs/overlays/
$ cat > ~/.config/nixpkgs/overlays/prefer-remote-fetch.nix <<EOF
  self: super: super.prefer-remote-fetch self super

5.5. pkgs.nix-gitignore

pkgs.nix-gitignore is a function that acts similarly to builtins.filterSource but also allows filtering with the help of the gitignore format.

5.5.1. Usage

pkgs.nix-gitignore exports a number of functions, but you'll most likely need either gitignoreSource or gitignoreSourcePure. As their first argument, they both accept either 1. a file with gitignore lines or 2. a string with gitignore lines, or 3. a list of either of the two. They will be concatenated into a single big string.

{ pkgs ? import <nixpkgs> {} }:

 nix-gitignore.gitignoreSource [] ./source
     # Simplest version

 nix-gitignore.gitignoreSource "supplemental-ignores\n" ./source
     # This one reads the ./source/.gitignore and concats the auxiliary ignores

 nix-gitignore.gitignoreSourcePure "ignore-this\nignore-that\n" ./source
     # Use this string as gitignore, don't read ./source/.gitignore.

 nix-gitignore.gitignoreSourcePure ["ignore-this\nignore-that\n", ~/.gitignore] ./source
     # It also accepts a list (of strings and paths) that will be concatenated
     # once the paths are turned to strings via readFile.

These functions are derived from the Filter functions by setting the first filter argument to (_: _: true):

gitignoreSourcePure = gitignoreFilterSourcePure (_: _: true);
gitignoreSource = gitignoreFilterSource (_: _: true);

Those filter functions accept the same arguments the builtins.filterSource function would pass to its filters, thus fn: gitignoreFilterSourcePure fn "" should be extensionally equivalent to filterSource. The file is blacklisted if it's blacklisted by either your filter or the gitignoreFilter.

If you want to make your own filter from scratch, you may use

gitignoreFilter = ign: root: filterPattern (gitignoreToPatterns ign) root;

5.5.2. gitignore files in subdirectories

If you wish to use a filter that would search for .gitignore files in subdirectories, just like git does by default, use this function:

gitignoreFilterRecursiveSource = filter: patterns: root:
# OR
gitignoreRecursiveSource = gitignoreFilterSourcePure (_: _: true);

Chapter 6. The Standard Environment

The standard build environment in the Nix Packages collection provides an environment for building Unix packages that does a lot of common build tasks automatically. In fact, for Unix packages that use the standard ./configure; make; make install build interface, you don’t need to write a build script at all; the standard environment does everything automatically. If stdenv doesn’t do what you need automatically, you can easily customise or override the various build phases.

6.1. Using stdenv

To build a package with the standard environment, you use the function stdenv.mkDerivation, instead of the primitive built-in function derivation, e.g.

stdenv.mkDerivation {
  name = "libfoo-1.2.3";
  src = fetchurl {
    url = "";
    sha256 = "0x2g1jqygyr5wiwg4ma1nd7w4ydpy82z9gkcv8vh2v8dn3y58v5m";

(stdenv needs to be in scope, so if you write this in a separate Nix expression from pkgs/all-packages.nix, you need to pass it as a function argument.) Specifying a name and a src is the absolute minimum Nix requires. For convenience, you can also use pname and version attributes and mkDerivation will automatically set name to "${pname}-${version}" by default. Since RFC 0035, this is preferred for packages in Nixpkgs, as it allows us to reuse the version easily:

stdenv.mkDerivation rec {
  pname = "libfoo";
  version = "1.2.3";
  src = fetchurl {
    url = "${version}.tar.bz2";
    sha256 = "0x2g1jqygyr5wiwg4ma1nd7w4ydpy82z9gkcv8vh2v8dn3y58v5m";

Many packages have dependencies that are not provided in the standard environment. It’s usually sufficient to specify those dependencies in the buildInputs attribute:

stdenv.mkDerivation {
  name = "libfoo-1.2.3";
  buildInputs = [libbar perl ncurses];

This attribute ensures that the bin subdirectories of these packages appear in the PATH environment variable during the build, that their include subdirectories are searched by the C compiler, and so on. (See Section 6.7, “Package setup hooks” for details.)

Often it is necessary to override or modify some aspect of the build. To make this easier, the standard environment breaks the package build into a number of phases, all of which can be overridden or modified individually: unpacking the sources, applying patches, configuring, building, and installing. (There are some others; see Section 6.5, “Phases”.) For instance, a package that doesn’t supply a makefile but instead has to be compiled manually could be handled like this:

stdenv.mkDerivation {
  name = "fnord-4.5";
  buildPhase = ''
    gcc foo.c -o foo
  installPhase = ''
    mkdir -p $out/bin
    cp foo $out/bin

(Note the use of ''-style string literals, which are very convenient for large multi-line script fragments because they don’t need escaping of " and \, and because indentation is intelligently removed.)

There are many other attributes to customise the build. These are listed in Section 6.4, “Attributes”.

While the standard environment provides a generic builder, you can still supply your own build script:

stdenv.mkDerivation {
  name = "libfoo-1.2.3";
  builder = ./;

where the builder can do anything it wants, but typically starts with

source $stdenv/setup

to let stdenv set up the environment (e.g., process the buildInputs). If you want, you can still use stdenv’s generic builder:

source $stdenv/setup

buildPhase() {
  echo "... this is my custom build phase ..."
  gcc foo.c -o foo

installPhase() {
  mkdir -p $out/bin
  cp foo $out/bin


6.2. Tools provided by stdenv

The standard environment provides the following packages:

  • The GNU C Compiler, configured with C and C++ support.

  • GNU coreutils (contains a few dozen standard Unix commands).

  • GNU findutils (contains find).

  • GNU diffutils (contains diff, cmp).

  • GNU sed.

  • GNU grep.

  • GNU awk.

  • GNU tar.

  • gzip, bzip2 and xz.

  • GNU Make.

  • Bash. This is the shell used for all builders in the Nix Packages collection. Not using /bin/sh removes a large source of portability problems.

  • The patch command.

On Linux, stdenv also includes the patchelf utility.

6.3. Specifying dependencies

As described in the Nix manual, almost any *.drv store path in a derivation’s attribute set will induce a dependency on that derivation. mkDerivation, however, takes a few attributes intended to include all the dependencies of a package. This is done both for structure and consistency, but also so that certain other setup can take place. For example, certain dependencies need their bin directories added to the PATH. That is built-in, but other setup is done via a pluggable mechanism that works in conjunction with these dependency attributes. See Section 6.7, “Package setup hooks” for details.

Dependencies can be broken down along three axes: their host and target platforms relative to the new derivation’s, and whether they are propagated. The platform distinctions are motivated by cross compilation; see Chapter 9, Cross-compilation for exactly what each platform means. [1] But even if one is not cross compiling, the platforms imply whether or not the dependency is needed at run-time or build-time, a concept that makes perfect sense outside of cross compilation. By default, the run-time/build-time distinction is just a hint for mental clarity, but with strictDeps set it is mostly enforced even in the native case.

The extension of PATH with dependencies, alluded to above, proceeds according to the relative platforms alone. The process is carried out only for dependencies whose host platform matches the new derivation’s build platform i.e. dependencies which run on the platform where the new derivation will be built. [2] For each dependency <dep> of those dependencies, dep/bin, if present, is added to the PATH environment variable.

A dependency is said to be propagated when some of its other-transitive (non-immediate) downstream dependencies also need it as an immediate dependency. [3]

It is important to note that dependencies are not necessarily propagated as the same sort of dependency that they were before, but rather as the corresponding sort so that the platform rules still line up. To determine the exact rules for dependency propagation, we start by assigning to each dependency a couple of ternary numbers (-1 for build, 0 for host, and 1 for target), representing how respectively its host and target platforms are offset from the depending derivation’s platforms. The following table summarize the different combinations that can be obtained:

host → target attribute name offset
build --> build depsBuildBuild -1, -1
build --> host nativeBuildInputs -1, 0
build --> target depsBuildTarget -1, 1
host --> host depsHostHost 0, 0
host --> target buildInputs 0, 1
target --> target depsTargetTarget 1, 1

Algorithmically, we traverse propagated inputs, accumulating every propagated dependency’s propagated dependencies and adjusting them to account for the “shift in perspective” described by the current dependency’s platform offsets. This results is sort of a transitive closure of the dependency relation, with the offsets being approximately summed when two dependency links are combined. We also prune transitive dependencies whose combined offsets go out-of-bounds, which can be viewed as a filter over that transitive closure removing dependencies that are blatantly absurd.

We can define the process precisely with Natural Deduction using the inference rules. This probably seems a bit obtuse, but so is the bash code that actually implements it! [4] They’re confusing in very different ways so… hopefully if something doesn’t make sense in one presentation, it will in the other!

let mapOffset(h, t, i) = i + (if i <= 0 then h else t - 1)

propagated-dep(h0, t0, A, B)
propagated-dep(h1, t1, B, C)
h0 + h1 in {-1, 0, 1}
h0 + t1 in {-1, 0, 1}
-------------------------------------- Transitive property
propagated-dep(mapOffset(h0, t0, h1),
               mapOffset(h0, t0, t1),
               A, C)
let mapOffset(h, t, i) = i + (if i <= 0 then h else t - 1)

dep(h0, _, A, B)
propagated-dep(h1, t1, B, C)
h0 + h1 in {-1, 0, 1}
h0 + t1 in {-1, 0, -1}
----------------------------- Take immediate dependencies' propagated dependencies
propagated-dep(mapOffset(h0, t0, h1),
               mapOffset(h0, t0, t1),
               A, C)
propagated-dep(h, t, A, B)
----------------------------- Propagated dependencies count as dependencies
dep(h, t, A, B)

Some explanation of this monstrosity is in order. In the common case, the target offset of a dependency is the successor to the target offset: t = h + 1. That means that:

let f(h, t, i) = i + (if i <= 0 then h else t - 1)
let f(h, h + 1, i) = i + (if i <= 0 then h else (h + 1) - 1)
let f(h, h + 1, i) = i + (if i <= 0 then h else h)
let f(h, h + 1, i) = i + h

This is where “sum-like” comes in from above: We can just sum all of the host offsets to get the host offset of the transitive dependency. The target offset is the transitive dependency is simply the host offset + 1, just as it was with the dependencies composed to make this transitive one; it can be ignored as it doesn’t add any new information.

Because of the bounds checks, the uncommon cases are h = t and h + 2 = t. In the former case, the motivation for mapOffset is that since its host and target platforms are the same, no transitive dependency of it should be able to “discover” an offset greater than its reduced target offsets. mapOffset effectively “squashes” all its transitive dependencies’ offsets so that none will ever be greater than the target offset of the original h = t package. In the other case, h + 1 is skipped over between the host and target offsets. Instead of squashing the offsets, we need to “rip” them apart so no transitive dependencies’ offset is that one.

Overall, the unifying theme here is that propagation shouldn’t be introducing transitive dependencies involving platforms the depending package is unaware of. [One can imagine the dependending package asking for dependencies with the platforms it knows about; other platforms it doesn’t know how to ask for. The platform description in that scenario is a kind of unforagable capability.] The offset bounds checking and definition of mapOffset together ensure that this is the case. Discovering a new offset is discovering a new platform, and since those platforms weren’t in the derivation “spec” of the needing package, they cannot be relevant. From a capability perspective, we can imagine that the host and target platforms of a package are the capabilities a package requires, and the depending package must provide the capability to the dependency.

6.3.1. Variables specifying dependencies depsBuildBuild

A list of dependencies whose host and target platforms are the new derivation’s build platform. These are programs and libraries used at build time that produce programs and libraries also used at build time. If the dependency doesn’t care about the target platform (i.e. isn’t a compiler or similar tool), put it in nativeBuildInputs instead. The most common use of this, the default C compiler for this role. That example crops up more than one might think in old commonly used C libraries.

Since these packages are able to be run at build-time, they are always added to the PATH, as described above. But since these packages are only guaranteed to be able to run then, they shouldn’t persist as run-time dependencies. This isn’t currently enforced, but could be in the future. nativeBuildInputs

A list of dependencies whose host platform is the new derivation’s build platform, and target platform is the new derivation’s host platform. These are programs and libraries used at build-time that, if they are a compiler or similar tool, produce code to run at run-time—i.e. tools used to build the new derivation. If the dependency doesn’t care about the target platform (i.e. isn’t a compiler or similar tool), put it here, rather than in depsBuildBuild or depsBuildTarget. This could be called depsBuildHost but nativeBuildInputs is used for historical continuity.

Since these packages are able to be run at build-time, they are added to the PATH, as described above. But since these packages are only guaranteed to be able to run then, they shouldn’t persist as run-time dependencies. This isn’t currently enforced, but could be in the future. depsBuildTarget

A list of dependencies whose host platform is the new derivation’s build platform, and target platform is the new derivation’s target platform. These are programs used at build time that produce code to run with code produced by the depending package. Most commonly, these are tools used to build the runtime or standard library that the currently-being-built compiler will inject into any code it compiles. In many cases, the currently-being-built-compiler is itself employed for that task, but when that compiler won’t run (i.e. its build and host platform differ) this is not possible. Other times, the compiler relies on some other tool, like binutils, that is always built separately so that the dependency is unconditional.

This is a somewhat confusing concept to wrap one’s head around, and for good reason. As the only dependency type where the platform offsets, -1 and 1, are not adjacent integers, it requires thinking of a bootstrapping stage two away from the current one. It and its use-case go hand in hand and are both considered poor form: try to not need this sort of dependency, and try to avoid building standard libraries and runtimes in the same derivation as the compiler produces code using them. Instead strive to build those like a normal library, using the newly-built compiler just as a normal library would. In short, do not use this attribute unless you are packaging a compiler and are sure it is needed.

Since these packages are able to run at build time, they are added to the PATH, as described above. But since these packages are only guaranteed to be able to run then, they shouldn’t persist as run-time dependencies. This isn’t currently enforced, but could be in the future. depsHostHost

A list of dependencies whose host and target platforms match the new derivation’s host platform. In practice, this would usually be tools used by compilers for macros or a metaprogramming system, or libraries used by the macros or metaprogramming code itself. It’s always preferable to use a depsBuildBuild dependency in the derivation being built over a depsHostHost on the tool doing the building for this purpose. buildInputs

A list of dependencies whose host platform and target platform match the new derivation’s. This would be called depsHostTarget but for historical continuity. If the dependency doesn’t care about the target platform (i.e. isn’t a compiler or similar tool), put it here, rather than in depsBuildBuild.

These are often programs and libraries used by the new derivation at run-time, but that isn’t always the case. For example, the machine code in a statically-linked library is only used at run-time, but the derivation containing the library is only needed at build-time. Even in the dynamic case, the library may also be needed at build-time to appease the linker. depsTargetTarget

A list of dependencies whose host platform matches the new derivation’s target platform. These are packages that run on the target platform, e.g. the standard library or run-time deps of standard library that a compiler insists on knowing about. It’s poor form in almost all cases for a package to depend on another from a future stage [future stage corresponding to positive offset]. Do not use this attribute unless you are packaging a compiler and are sure it is needed. depsBuildBuildPropagated

The propagated equivalent of depsBuildBuild. This perhaps never ought to be used, but it is included for consistency [see below for the others]. propagatedNativeBuildInputs

The propagated equivalent of nativeBuildInputs. This would be called depsBuildHostPropagated but for historical continuity. For example, if package Y has propagatedNativeBuildInputs = [X], and package Z has buildInputs = [Y], then package Z will be built as if it included package X in its nativeBuildInputs. If instead, package Z has nativeBuildInputs = [Y], then Z will be built as if it included X in the depsBuildBuild of package Z, because of the sum of the two -1 host offsets. depsBuildTargetPropagated

The propagated equivalent of depsBuildTarget. This is prefixed for the same reason of alerting potential users. depsHostHostPropagated

The propagated equivalent of depsHostHost. propagatedBuildInputs

The propagated equivalent of buildInputs. This would be called depsHostTargetPropagated but for historical continuity. depsTargetTargetPropagated

The propagated equivalent of depsTargetTarget. This is prefixed for the same reason of alerting potential users.

6.4. Attributes

6.4.1. Variables affecting stdenv initialisation NIX_DEBUG

A natural number indicating how much information to log. If set to 1 or higher, stdenv will print moderate debugging information during the build. In particular, the gcc and ld wrapper scripts will print out the complete command line passed to the wrapped tools. If set to 6 or higher, the stdenv setup script will be run with set -x tracing. If set to 7 or higher, the gcc and ld wrapper scripts will also be run with set -x tracing.

6.4.2. Attributes affecting build properties enableParallelBuilding

If set to true, stdenv will pass specific flags to make and other build tools to enable parallel building with up to build-cores workers.

Unless set to false, some build systems with good support for parallel building including cmake, meson, and qmake will set it to true.

6.4.3. Special variables passthru

This is an attribute set which can be filled with arbitrary values. For example:

passthru = {
  foo = "bar";
  baz = {
    value1 = 4;
    value2 = 5;

Values inside it are not passed to the builder, so you can change them without triggering a rebuild. However, they can be accessed outside of a derivation directly, as if they were set inside a derivation itself, e.g. hello.baz.value1. We don’t specify any usage or schema of passthru - it is meant for values that would be useful outside the derivation in other parts of a Nix expression (e.g. in other derivations). An example would be to convey some specific dependency of your derivation which contains a program with plugins support. Later, others who make derivations with plugins can use passed-through dependency to ensure that their plugin would be binary-compatible with built program. passthru.updateScript

A script to be run by maintainers/scripts/update.nix when the package is matched. It needs to be an executable file, either on the file system:

passthru.updateScript = ./;

or inside the expression itself:

passthru.updateScript = writeScript "update-zoom-us" ''
  #!/usr/bin/env nix-shell
  #!nix-shell -i bash -p curl pcre common-updater-scripts

  set -eu -o pipefail

  version="$(curl -sI | grep -Fi 'Location:' | pcregrep -o1 '/(([0-9]\.?)+)/')"
  update-source-version zoom-us "$version"

The attribute can also contain a list, a script followed by arguments to be passed to it:

passthru.updateScript = [ ../../ pname "--requested-release=unstable" ];

The script will be run with UPDATE_NIX_ATTR_PATH environment variable set to the attribute path it is supposed to update.

Note: The script will be usually run from the root of the Nixpkgs repository but you should not rely on that. Also note that the update scripts will be run in parallel by default; you should avoid running git commit or any other commands that cannot handle that.

For information about how to run the updates, execute nix-shell maintainers/scripts/update.nix.

6.5. Phases

The generic builder has a number of phases. Package builds are split into phases to make it easier to override specific parts of the build (e.g., unpacking the sources or installing the binaries). Furthermore, it allows a nicer presentation of build logs in the Nix build farm.

Each phase can be overridden in its entirety either by setting the environment variable namePhase to a string containing some shell commands to be executed, or by redefining the shell function namePhase. The former is convenient to override a phase from the derivation, while the latter is convenient from a build script. However, typically one only wants to add some commands to a phase, e.g. by defining postInstall or preFixup, as skipping some of the default actions may have unexpected consequences. The default script for each phase is defined in the file pkgs/stdenv/generic/

6.5.1. Controlling phases

There are a number of variables that control what phases are executed and in what order: Variables affecting phase control phases

Specifies the phases. You can change the order in which phases are executed, or add new phases, by setting this variable. If it’s not set, the default value is used, which is $prePhases unpackPhase patchPhase $preConfigurePhases configurePhase $preBuildPhases buildPhase checkPhase $preInstallPhases installPhase fixupPhase installCheckPhase $preDistPhases distPhase $postPhases.

Usually, if you just want to add a few phases, it’s more convenient to set one of the variables below (such as preInstallPhases), as you then don’t specify all the normal phases. prePhases

Additional phases executed before any of the default phases. preConfigurePhases

Additional phases executed just before the configure phase. preBuildPhases

Additional phases executed just before the build phase. preInstallPhases

Additional phases executed just before the install phase. preFixupPhases

Additional phases executed just before the fixup phase. preDistPhases

Additional phases executed just before the distribution phase. postPhases

Additional phases executed after any of the default phases.

6.5.2. The unpack phase

The unpack phase is responsible for unpacking the source code of the package. The default implementation of unpackPhase unpacks the source files listed in the src environment variable to the current directory. It supports the following files by default: Tar files

These can optionally be compressed using gzip (.tar.gz, .tgz or .tar.Z), bzip2 (.tar.bz2, .tbz2 or .tbz) or xz (.tar.xz, .tar.lzma or .txz). Zip files

Zip files are unpacked using unzip. However, unzip is not in the standard environment, so you should add it to nativeBuildInputs yourself. Directories in the Nix store

These are simply copied to the current directory. The hash part of the file name is stripped, e.g. /nix/store/1wydxgby13cz...-my-sources would be copied to my-sources.

Additional file types can be supported by setting the unpackCmd variable (see below). Variables controlling the unpack phase srcs / src

The list of source files or directories to be unpacked or copied. One of these must be set. Note that if you use srcs, you should also set sourceRoot or setSourceRoot. sourceRoot

After running unpackPhase, the generic builder changes the current directory to the directory created by unpacking the sources. If there are multiple source directories, you should set sourceRoot to the name of the intended directory. Set sourceRoot = "."; if you use srcs and control the unpack phase yourself. setSourceRoot

Alternatively to setting sourceRoot, you can set setSourceRoot to a shell command to be evaluated by the unpack phase after the sources have been unpacked. This command must set sourceRoot. preUnpack

Hook executed at the start of the unpack phase. postUnpack

Hook executed at the end of the unpack phase. dontUnpack

Set to true to skip the unpack phase. dontMakeSourcesWritable

If set to 1, the unpacked sources are not made writable. By default, they are made writable to prevent problems with read-only sources. For example, copied store directories would be read-only without this. unpackCmd

The unpack phase evaluates the string $unpackCmd for any unrecognised file. The path to the current source file is contained in the curSrc variable.

6.5.3. The patch phase

The patch phase applies the list of patches defined in the patches variable. Variables controlling the patch phase dontPatch

Set to true to skip the patch phase. patches

The list of patches. They must be in the format accepted by the patch command, and may optionally be compressed using gzip (.gz), bzip2 (.bz2) or xz (.xz). patchFlags

Flags to be passed to patch. If not set, the argument -p1 is used, which causes the leading directory component to be stripped from the file names in each patch. prePatch

Hook executed at the start of the patch phase. postPatch

Hook executed at the end of the patch phase.

6.5.4. The configure phase

The configure phase prepares the source tree for building. The default configurePhase runs ./configure (typically an Autoconf-generated script) if it exists. Variables controlling the configure phase configureScript

The name of the configure script. It defaults to ./configure if it exists; otherwise, the configure phase is skipped. This can actually be a command (like perl ./ configureFlags

A list of strings passed as additional arguments to the configure script. dontConfigure

Set to true to skip the configure phase. configureFlagsArray

A shell array containing additional arguments passed to the configure script. You must use this instead of configureFlags if the arguments contain spaces. dontAddPrefix

By default, the flag --prefix=$prefix is added to the configure flags. If this is undesirable, set this variable to true. prefix

The prefix under which the package must be installed, passed via the --prefix option to the configure script. It defaults to $out. prefixKey

The key to use when specifying the prefix. By default, this is set to --prefix= as that is used by the majority of packages. dontAddStaticConfigureFlags

By default, when building statically, stdenv will try to add build system appropriate configure flags to try to enable static builds.

If this is undesirable, set this variable to true. dontAddDisableDepTrack

By default, the flag --disable-dependency-tracking is added to the configure flags to speed up Automake-based builds. If this is undesirable, set this variable to true. dontFixLibtool

By default, the configure phase applies some special hackery to all files called before running the configure script in order to improve the purity of Libtool-based packages [5] . If this is undesirable, set this variable to true. dontDisableStatic

By default, when the configure script has --enable-static, the option --disable-static is added to the configure flags.

If this is undesirable, set this variable to true. It is automatically set to true when building statically, for example through pkgsStatic. configurePlatforms

By default, when cross compiling, the configure script has --build=... and --host=... passed. Packages can instead pass [ "build" "host" "target" ] or a subset to control exactly which platform flags are passed. Compilers and other tools can use this to also pass the target platform. [6] preConfigure

Hook executed at the start of the configure phase. postConfigure

Hook executed at the end of the configure phase.

6.5.5. The build phase

The build phase is responsible for actually building the package (e.g. compiling it). The default buildPhase simply calls make if a file named Makefile, makefile or GNUmakefile exists in the current directory (or the makefile is explicitly set); otherwise it does nothing. Variables controlling the build phase dontBuild

Set to true to skip the build phase. makefile

The file name of the Makefile. makeFlags

A list of strings passed as additional flags to make. These flags are also used by the default install and check phase. For setting make flags specific to the build phase, use buildFlags (see below).

makeFlags = [ "PREFIX=$(out)" ];
Note: The flags are quoted in bash, but environment variables can be specified by using the make syntax. makeFlagsArray

A shell array containing additional arguments passed to make. You must use this instead of makeFlags if the arguments contain spaces, e.g.

preBuild = ''
  makeFlagsArray+=(CFLAGS="-O0 -g" LDFLAGS="-lfoo -lbar")

Note that shell arrays cannot be passed through environment variables, so you cannot set makeFlagsArray in a derivation attribute (because those are passed through environment variables): you have to define them in shell code. buildFlags / buildFlagsArray

A list of strings passed as additional flags to make. Like makeFlags and makeFlagsArray, but only used by the build phase. preBuild

Hook executed at the start of the build phase. postBuild

Hook executed at the end of the build phase.

You can set flags for make through the makeFlags variable.

Before and after running make, the hooks preBuild and postBuild are called, respectively.

6.5.6. The check phase

The check phase checks whether the package was built correctly by running its test suite. The default checkPhase calls make check, but only if the doCheck variable is enabled. Variables controlling the check phase doCheck

Controls whether the check phase is executed. By default it is skipped, but if doCheck is set to true, the check phase is usually executed. Thus you should set

doCheck = true;

in the derivation to enable checks. The exception is cross compilation. Cross compiled builds never run tests, no matter how doCheck is set, as the newly-built program won’t run on the platform used to build it. makeFlags / makeFlagsArray / makefile

See the build phase for details. checkTarget

The make target that runs the tests. Defaults to check. checkFlags / checkFlagsArray

A list of strings passed as additional flags to make. Like makeFlags and makeFlagsArray, but only used by the check phase. checkInputs

A list of dependencies used by the phase. This gets included in nativeBuildInputs when doCheck is set. preCheck

Hook executed at the start of the check phase. postCheck

Hook executed at the end of the check phase.

6.5.7. The install phase

The install phase is responsible for installing the package in the Nix store under out. The default installPhase creates the directory $out and calls make install. Variables controlling the install phase dontInstall

Set to true to skip the install phase. makeFlags / makeFlagsArray / makefile

See the build phase for details. installTargets

The make targets that perform the installation. Defaults to install. Example:

installTargets = "install-bin install-doc"; installFlags / installFlagsArray

A list of strings passed as additional flags to make. Like makeFlags and makeFlagsArray, but only used by the install phase. preInstall

Hook executed at the start of the install phase. postInstall

Hook executed at the end of the install phase.

6.5.8. The fixup phase

The fixup phase performs some (Nix-specific) post-processing actions on the files installed under $out by the install phase. The default fixupPhase does the following:

  • It moves the man/, doc/ and info/ subdirectories of $out to share/.

  • It strips libraries and executables of debug information.

  • On Linux, it applies the patchelf command to ELF executables and libraries to remove unused directories from the RPATH in order to prevent unnecessary runtime dependencies.

  • It rewrites the interpreter paths of shell scripts to paths found in PATH. E.g., /usr/bin/perl will be rewritten to /nix/store/some-perl/bin/perl found in PATH. Variables controlling the fixup phase dontFixup

Set to true to skip the fixup phase. dontStrip

If set, libraries and executables are not stripped. By default, they are. dontStripHost

Like dontStrip, but only affects the strip command targetting the package’s host platform. Useful when supporting cross compilation, but otherwise feel free to ignore. dontStripTarget

Like dontStrip, but only affects the strip command targetting the packages’ target platform. Useful when supporting cross compilation, but otherwise feel free to ignore. dontMoveSbin

If set, files in $out/sbin are not moved to $out/bin. By default, they are. stripAllList

List of directories to search for libraries and executables from which all symbols should be stripped. By default, it’s empty. Stripping all symbols is risky, since it may remove not just debug symbols but also ELF information necessary for normal execution. stripAllFlags

Flags passed to the strip command applied to the files in the directories listed in stripAllList. Defaults to -s (i.e. --strip-all). stripDebugList

List of directories to search for libraries and executables from which only debugging-related symbols should be stripped. It defaults to lib lib32 lib64 libexec bin sbin. stripDebugFlags

Flags passed to the strip command applied to the files in the directories listed in stripDebugList. Defaults to -S (i.e. --strip-debug). dontPatchELF

If set, the patchelf command is not used to remove unnecessary RPATH entries. Only applies to Linux. dontPatchShebangs

If set, scripts starting with #! do not have their interpreter paths rewritten to paths in the Nix store. dontPruneLibtoolFiles

If set, libtool .la files associated with shared libraries won’t have their dependency_libs field cleared. forceShare

The list of directories that must be moved from $out to $out/share. Defaults to man doc info. setupHook

A package can export a setup hook by setting this variable. The setup hook, if defined, is copied to $out/nix-support/setup-hook. Environment variables are then substituted in it using substituteAll. preFixup

Hook executed at the start of the fixup phase. postFixup

Hook executed at the end of the fixup phase. separateDebugInfo

If set to true, the standard environment will enable debug information in C/C++ builds. After installation, the debug information will be separated from the executables and stored in the output named debug. (This output is enabled automatically; you don’t need to set the outputs attribute explicitly.) To be precise, the debug information is stored in debug/lib/debug/.build-id/XX/YYYY…, where <XXYYYY…> is the <build ID> of the binary — a SHA-1 hash of the contents of the binary. Debuggers like GDB use the build ID to look up the separated debug information.

For example, with GDB, you can add

set debug-file-directory ~/.nix-profile/lib/debug

to ~/.gdbinit. GDB will then be able to find debug information installed via nix-env -i.

6.5.9. The installCheck phase

The installCheck phase checks whether the package was installed correctly by running its test suite against the installed directories. The default installCheck calls make installcheck.

It is often better to add tests that are not part of the source distribution to passthru.tests (see Section 7.1.12, “tests). This avoids adding overhead to every build and enables us to run them independently. Variables controlling the installCheck phase doInstallCheck

Controls whether the installCheck phase is executed. By default it is skipped, but if doInstallCheck is set to true, the installCheck phase is usually executed. Thus you should set

doInstallCheck = true;

in the derivation to enable install checks. The exception is cross compilation. Cross compiled builds never run tests, no matter how doInstallCheck is set, as the newly-built program won’t run on the platform used to build it. installCheckTarget

The make target that runs the install tests. Defaults to installcheck. installCheckFlags / installCheckFlagsArray

A list of strings passed as additional flags to make. Like makeFlags and makeFlagsArray, but only used by the installCheck phase. installCheckInputs

A list of dependencies used by the phase. This gets included in nativeBuildInputs when doInstallCheck is set. preInstallCheck

Hook executed at the start of the installCheck phase. postInstallCheck

Hook executed at the end of the installCheck phase.

6.5.10. The distribution phase

The distribution phase is intended to produce a source distribution of the package. The default distPhase first calls make dist, then it copies the resulting source tarballs to $out/tarballs/. This phase is only executed if the attribute doDist is set. Variables controlling the distribution phase distTarget

The make target that produces the distribution. Defaults to dist. distFlags / distFlagsArray

Additional flags passed to make. tarballs

The names of the source distribution files to be copied to $out/tarballs/. It can contain shell wildcards. The default is *.tar.gz. dontCopyDist

If set, no files are copied to $out/tarballs/. preDist

Hook executed at the start of the distribution phase. postDist

Hook executed at the end of the distribution phase.

6.6. Shell functions

The standard environment provides a number of useful functions.

6.6.1. makeWrapper <executable> <wrapperfile> <args>

Constructs a wrapper for a program with various possible arguments. For example:

# adds `FOOBAR=baz` to `$out/bin/foo`’s environment
makeWrapper $out/bin/foo $wrapperfile --set FOOBAR baz

# prefixes the binary paths of `hello` and `git`
# Be advised that paths often should be patched in directly
# (via string replacements or in `configurePhase`).
makeWrapper $out/bin/foo $wrapperfile --prefix PATH : ${lib.makeBinPath [ hello git ]}

There’s many more kinds of arguments, they are documented in nixpkgs/pkgs/build-support/setup-hooks/

wrapProgram is a convenience function you probably want to use most of the time.

6.6.2. substitute <infile> <outfile> <subs>

Performs string substitution on the contents of <infile>, writing the result to <outfile>. The substitutions in <subs> are of the following form: --replace <s1> <s2>

Replace every occurrence of the string <s1> by <s2>. --subst-var <varName>

Replace every occurrence of @varName@ by the contents of the environment variable <varName>. This is useful for generating files from templates, using @...@ in the template as placeholders. --subst-var-by <varName> <s>

Replace every occurrence of @varName@ by the string <s>.


substitute ./ ./foo.out \
    --replace /usr/bin/bar $bar/bin/bar \
    --replace "a string containing spaces" "some other text" \
    --subst-var someVar

6.6.3. substituteInPlace <file> <subs>

Like substitute, but performs the substitutions in place on the file <file>.

6.6.4. substituteAll <infile> <outfile>

Replaces every occurrence of @varName@, where <varName> is any environment variable, in <infile>, writing the result to <outfile>. For instance, if <infile> has the contents

#! @bash@/bin/sh
echo @foo@

and the environment contains bash=/nix/store/bmwp0q28cf21...-bash-3.2-p39 and coreutils=/nix/store/68afga4khv0w...-coreutils-6.12, but does not contain the variable foo, then the output will be

#! /nix/store/bmwp0q28cf21...-bash-3.2-p39/bin/sh
echo @foo@

That is, no substitution is performed for undefined variables.

Environment variables that start with an uppercase letter or an underscore are filtered out, to prevent global variables (like HOME) or private variables (like __ETC_PROFILE_DONE) from accidentally getting substituted. The variables also have to be valid bash names, as defined in the bash manpage (alphanumeric or _, must not start with a number).

6.6.5. substituteAllInPlace <file>

Like substituteAll, but performs the substitutions in place on the file <file>.

6.6.6. stripHash <path>

Strips the directory and hash part of a store path, outputting the name part to stdout. For example:

# prints coreutils-8.24
stripHash "/nix/store/9s9r019176g7cvn2nvcw41gsp862y6b4-coreutils-8.24"

If you wish to store the result in another variable, then the following idiom may be useful:

someVar=$(stripHash $name)

6.6.7. wrapProgram <executable> <makeWrapperArgs>

Convenience function for makeWrapper that automatically creates a sane wrapper file. It takes all the same arguments as makeWrapper, except for --argv0.

It cannot be applied multiple times, since it will overwrite the wrapper file.

6.7. Package setup hooks

Nix itself considers a build-time dependency as merely something that should previously be built and accessible at build time—packages themselves are on their own to perform any additional setup. In most cases, that is fine, and the downstream derivation can deal with its own dependencies. But for a few common tasks, that would result in almost every package doing the same sort of setup work—depending not on the package itself, but entirely on which dependencies were used.

In order to alleviate this burden, the setup hook mechanism was written, where any package can include a shell script that [by convention rather than enforcement by Nix], any downstream reverse-dependency will source as part of its build process. That allows the downstream dependency to merely specify its dependencies, and lets those dependencies effectively initialize themselves. No boilerplate mirroring the list of dependencies is needed.

The setup hook mechanism is a bit of a sledgehammer though: a powerful feature with a broad and indiscriminate area of effect. The combination of its power and implicit use may be expedient, but isn’t without costs. Nix itself is unchanged, but the spirit of added dependencies being effect-free is violated even if the letter isn’t. For example, if a derivation path is mentioned more than once, Nix itself doesn’t care and simply makes sure the dependency derivation is already built just the same—depending is just needing something to exist, and needing is idempotent. However, a dependency specified twice will have its setup hook run twice, and that could easily change the build environment (though a well-written setup hook will therefore strive to be idempotent so this is in fact not observable). More broadly, setup hooks are anti-modular in that multiple dependencies, whether the same or different, should not interfere and yet their setup hooks may well do so.

The most typical use of the setup hook is actually to add other hooks which are then run (i.e. after all the setup hooks) on each dependency. For example, the C compiler wrapper’s setup hook feeds itself flags for each dependency that contains relevant libraries and headers. This is done by defining a bash function, and appending its name to one of envBuildBuildHooks, envBuildHostHooks, envBuildTargetHooks, envHostHostHooks, envHostTargetHooks, or envTargetTargetHooks. These 6 bash variables correspond to the 6 sorts of dependencies by platform (there’s 12 total but we ignore the propagated/non-propagated axis).

Packages adding a hook should not hard code a specific hook, but rather choose a variable relative to how they are included. Returning to the C compiler wrapper example, if the wrapper itself is an n dependency, then it only wants to accumulate flags from n + 1 dependencies, as only those ones match the compiler’s target platform. The hostOffset variable is defined with the current dependency’s host offset targetOffset with its target offset, before its setup hook is sourced. Additionally, since most environment hooks don’t care about the target platform, that means the setup hook can append to the right bash array by doing something like

addEnvHooks "$hostOffset" myBashFunction

The existence of setups hooks has long been documented and packages inside Nixpkgs are free to use this mechanism. Other packages, however, should not rely on these mechanisms not changing between Nixpkgs versions. Because of the existing issues with this system, there’s little benefit from mandating it be stable for any period of time.

First, let’s cover some setup hooks that are part of Nixpkgs default stdenv. This means that they are run for every package built using stdenv.mkDerivation. Some of these are platform specific, so they may run on Linux but not Darwin or vice-versa.


This setup hook moves any installed documentation to the /share subdirectory directory. This includes the man, doc and info directories. This is needed for legacy programs that do not know how to use the share subdirectory.


This setup hook compresses any man pages that have been installed. The compression is done using the gzip program. This helps to reduce the installed size of packages.


This runs the strip command on installed binaries and libraries. This removes unnecessary information like debug symbols when they are not needed. This also helps to reduce the installed size of packages.


This setup hook patches installed scripts to use the full path to the shebang interpreter. A shebang interpreter is the first commented line of a script telling the operating system which program will run the script (e.g #!/bin/bash). In Nix, we want an exact path to that interpreter to be used. This often replaces /bin/sh with a path in the Nix store.


This verifies that no references are left from the install binaries to the directory used to build those binaries. This ensures that the binaries do not need things outside the Nix store. This is currently supported in Linux only.


This setup hook adds configure flags that tell packages to install files into any one of the proper outputs listed in outputs. This behavior can be turned off by setting setOutputFlags to false in the derivation environment. See Chapter 8, Multiple-output packages for more information.


This setup hook moves any binaries installed in the sbin/ subdirectory into bin/. In addition, a link is provided from sbin/ to bin/ for compatibility.


This setup hook moves any libraries installed in the lib64/ subdirectory into lib/. In addition, a link is provided from lib64/ to lib/ for compatibility.


This setup hook moves any systemd user units installed in the lib/ subdirectory into share/. In addition, a link is provided from share/ to lib/ for compatibility. This is needed for systemd to find user services when installed into the user profile.


This sets SOURCE_DATE_EPOCH to the modification time of the most recent file.

6.7.11. Bintools Wrapper

The Bintools Wrapper wraps the binary utilities for a bunch of miscellaneous purposes. These are GNU Binutils when targetting Linux, and a mix of cctools and GNU binutils for Darwin. [The “Bintools” name is supposed to be a compromise between “Binutils” and “cctools” not denoting any specific implementation.] Specifically, the underlying bintools package, and a C standard library (glibc or Darwin’s libSystem, just for the dynamic loader) are all fed in, and dependency finding, hardening (see below), and purity checks for each are handled by the Bintools Wrapper. Packages typically depend on CC Wrapper, which in turn (at run time) depends on the Bintools Wrapper.

The Bintools Wrapper was only just recently split off from CC Wrapper, so the division of labor is still being worked out. For example, it shouldn’t care about the C standard library, but just take a derivation with the dynamic loader (which happens to be the glibc on linux). Dependency finding however is a task both wrappers will continue to need to share, and probably the most important to understand. It is currently accomplished by collecting directories of host-platform dependencies (i.e. buildInputs and nativeBuildInputs) in environment variables. The Bintools Wrapper’s setup hook causes any lib and lib64 subdirectories to be added to NIX_LDFLAGS. Since the CC Wrapper and the Bintools Wrapper use the same strategy, most of the Bintools Wrapper code is sparsely commented and refers to the CC Wrapper. But the CC Wrapper’s code, by contrast, has quite lengthy comments. The Bintools Wrapper merely cites those, rather than repeating them, to avoid falling out of sync.

A final task of the setup hook is defining a number of standard environment variables to tell build systems which executables fulfill which purpose. They are defined to just be the base name of the tools, under the assumption that the Bintools Wrapper’s binaries will be on the path. Firstly, this helps poorly-written packages, e.g. ones that look for just gcc when CC isn’t defined yet clang is to be used. Secondly, this helps packages not get confused when cross-compiling, in which case multiple Bintools Wrappers may simultaneously be in use. [7] BUILD_- and TARGET_-prefixed versions of the normal environment variable are defined for additional Bintools Wrappers, properly disambiguating them.

A problem with this final task is that the Bintools Wrapper is honest and defines LD as ld. Most packages, however, firstly use the C compiler for linking, secondly use LD anyways, defining it as the C compiler, and thirdly, only so define LD when it is undefined as a fallback. This triple-threat means Bintools Wrapper will break those packages, as LD is already defined as the actual linker which the package won’t override yet doesn’t want to use. The workaround is to define, just for the problematic package, LD as the C compiler. A good way to do this would be preConfigure = "LD=$CC".

6.7.12. CC Wrapper

The CC Wrapper wraps a C toolchain for a bunch of miscellaneous purposes. Specifically, a C compiler (GCC or Clang), wrapped binary tools, and a C standard library (glibc or Darwin’s libSystem, just for the dynamic loader) are all fed in, and dependency finding, hardening (see below), and purity checks for each are handled by the CC Wrapper. Packages typically depend on the CC Wrapper, which in turn (at run-time) depends on the Bintools Wrapper.

Dependency finding is undoubtedly the main task of the CC Wrapper. This works just like the Bintools Wrapper, except that any include subdirectory of any relevant dependency is added to NIX_CFLAGS_COMPILE. The setup hook itself contains some lengthy comments describing the exact convoluted mechanism by which this is accomplished.

Similarly, the CC Wrapper follows the Bintools Wrapper in defining standard environment variables with the names of the tools it wraps, for the same reasons described above. Importantly, while it includes a cc symlink to the c compiler for portability, the CC will be defined using the compiler’s “real name” (i.e. gcc or clang). This helps lousy build systems that inspect on the name of the compiler rather than run it.

Here are some more packages that provide a setup hook. Since the list of hooks is extensible, this is not an exhaustive list. The mechanism is only to be used as a last resort, so it might cover most uses.

6.7.13. Perl

Adds the lib/site_perl subdirectory of each build input to the PERL5LIB environment variable. For instance, if buildInputs contains Perl, then the lib/site_perl subdirectory of each input is added to the PERL5LIB environment variable.

6.7.14. Python

Adds the lib/${python.libPrefix}/site-packages subdirectory of each build input to the PYTHONPATH environment variable.

6.7.15. pkg-config

Adds the lib/pkgconfig and share/pkgconfig subdirectories of each build input to the PKG_CONFIG_PATH environment variable.

6.7.16. Automake

Adds the share/aclocal subdirectory of each build input to the ACLOCAL_PATH environment variable.

6.7.17. Autoconf

The autoreconfHook derivation adds autoreconfPhase, which runs autoreconf, libtoolize and automake, essentially preparing the configure script in autotools-based builds. Most autotools-based packages come with the configure script pre-generated, but this hook is necessary for a few packages and when you need to patch the package’s configure scripts.

6.7.18. libxml2

Adds every file named catalog.xml found under the xml/dtd and xml/xsl subdirectories of each build input to the XML_CATALOG_FILES environment variable.

6.7.19. teTeX / TeX Live

Adds the share/texmf-nix subdirectory of each build input to the TEXINPUTS environment variable.

6.7.20. Qt 4

Sets the QTDIR environment variable to Qt’s path.

6.7.21. gdk-pixbuf

Exports GDK_PIXBUF_MODULE_FILE environment variable to the builder. Add librsvg package to buildInputs to get svg support. See also the setup hook description in GNOME platform docs.

6.7.22. GHC

Creates a temporary package database and registers every Haskell build input in it (TODO: how?).

6.7.23. GNOME platform

Hooks related to GNOME platform and related libraries like GLib, GTK and GStreamer are described in Section 15.10, “GNOME”.

6.7.24. autoPatchelfHook

This is a special setup hook which helps in packaging proprietary software in that it automatically tries to find missing shared library dependencies of ELF files based on the given buildInputs and nativeBuildInputs.

You can also specify a runtimeDependencies variable which lists dependencies to be unconditionally added to rpath of all executables. This is useful for programs that use dlopen 3 to load libraries at runtime.

In certain situations you may want to run the main command (autoPatchelf) of the setup hook on a file or a set of directories instead of unconditionally patching all outputs. This can be done by setting the dontAutoPatchelf environment variable to a non-empty value.

By default autoPatchelf will fail as soon as any ELF file requires a dependency which cannot be resolved via the given build inputs. In some situations you might prefer to just leave missing dependencies unpatched and continue to patch the rest. This can be achieved by setting the autoPatchelfIgnoreMissingDeps environment variable to a non-empty value.

The autoPatchelf command also recognizes a --no-recurse command line flag, which prevents it from recursing into subdirectories.

6.7.25. breakpointHook

This hook will make a build pause instead of stopping when a failure happens. It prevents nix from cleaning up the build environment immediately and allows the user to attach to a build environment using the cntr command. Upon build error it will print instructions on how to use cntr, which can be used to enter the environment for debugging. Installing cntr and running the command will provide shell access to the build sandbox of failed build. At /var/lib/cntr the sandboxed filesystem is mounted. All commands and files of the system are still accessible within the shell. To execute commands from the sandbox use the cntr exec subcommand. cntr is only supported on Linux-based platforms. To use it first add cntr to your environment.systemPackages on NixOS or alternatively to the root user on non-NixOS systems. Then in the package that is supposed to be inspected, add breakpointHook to nativeBuildInputs.

nativeBuildInputs = [ breakpointHook ];

When a build failure happens there will be an instruction printed that shows how to attach with cntr to the build sandbox.

Note: This won’t work with remote builds as the build environment is on a different machine and can’t be accessed by cntr. Remote builds can be turned off by setting --option builders '' for nix-build or --builders '' for nix build.

6.7.26. installShellFiles

This hook helps with installing manpages and shell completion files. It exposes 2 shell functions installManPage and installShellCompletion that can be used from your postInstall hook.

The installManPage function takes one or more paths to manpages to install. The manpages must have a section suffix, and may optionally be compressed (with .gz suffix). This function will place them into the correct directory.

The installShellCompletion function takes one or more paths to shell completion files. By default it will autodetect the shell type from the completion file extension, but you may also specify it by passing one of --bash, --fish, or --zsh. These flags apply to all paths listed after them (up until another shell flag is given). Each path may also have a custom installation name provided by providing a flag --name NAME before the path. If this flag is not provided, zsh completions will be renamed automatically such that foobar.zsh becomes _foobar. A root name may be provided for all paths using the flag --cmd NAME; this synthesizes the appropriate name depending on the shell (e.g. --cmd foo will synthesize the name foo.bash for bash and _foo for zsh). The path may also be a fifo or named fd (such as produced by <(cmd)), in which case the shell and name must be provided.

nativeBuildInputs = [ installShellFiles ];
postInstall = ''
  installManPage doc/foobar.1 doc/barfoo.3
  # explicit behavior
  installShellCompletion --bash --name foobar.bash share/completions.bash
  installShellCompletion --fish --name share/
  installShellCompletion --zsh --name _foobar share/completions.zsh
  # implicit behavior
  installShellCompletion share/completions/foobar.{bash,fish,zsh}
  # using named fd
  installShellCompletion --cmd foobar \
    --bash <($out/bin/foobar --bash-completion) \
    --fish <($out/bin/foobar --fish-completion) \
    --zsh <($out/bin/foobar --zsh-completion)

6.7.27. libiconv, libintl

A few libraries automatically add to NIX_LDFLAGS their library, making their symbols automatically available to the linker. This includes libiconv and libintl (gettext). This is done to provide compatibility between GNU Linux, where libiconv and libintl are bundled in, and other systems where that might not be the case. Sometimes, this behavior is not desired. To disable this behavior, set dontAddExtraLibs.

6.7.28. validatePkgConfig

The validatePkgConfig hook validates all pkg-config (.pc) files in a package. This helps catching some common errors in pkg-config files, such as undefined variables.

6.7.29. cmake

Overrides the default configure phase to run the CMake command. By default, we use the Make generator of CMake. In addition, dependencies are added automatically to CMAKE_PREFIX_PATH so that packages are correctly detected by CMake. Some additional flags are passed in to give similar behavior to configure-based packages. You can disable this hook’s behavior by setting configurePhase to a custom value, or by setting dontUseCmakeConfigure. cmakeFlags controls flags passed only to CMake. By default, parallel building is enabled as CMake supports parallel building almost everywhere. When Ninja is also in use, CMake will detect that and use the ninja generator.

6.7.30. xcbuildHook

Overrides the build and install phases to run the xcbuild command. This hook is needed when a project only comes with build files for the XCode build system. You can disable this behavior by setting buildPhase and configurePhase to a custom value. xcbuildFlags controls flags passed only to xcbuild.

6.7.31. Meson

Overrides the configure phase to run meson to generate Ninja files. To run these files, you should accompany Meson with ninja. By default, enableParallelBuilding is enabled as Meson supports parallel building almost everywhere. Variables controlling Meson mesonFlags

Controls the flags passed to meson. mesonBuildType

Which --buildtype to pass to Meson. We default to plain. mesonAutoFeatures

What value to set -Dauto_features= to. We default to enabled. mesonWrapMode

What value to set -Dwrap_mode= to. We default to nodownload as we disallow network access. dontUseMesonConfigure

Disables using Meson’s configurePhase.

6.7.32. ninja

Overrides the build, install, and check phase to run ninja instead of make. You can disable this behavior with the dontUseNinjaBuild, dontUseNinjaInstall, and dontUseNinjaCheck, respectively. Parallel building is enabled by default in Ninja.

6.7.33. unzip

This setup hook will allow you to unzip .zip files specified in $src. There are many similar packages like unrar, undmg, etc.

6.7.34. wafHook

Overrides the configure, build, and install phases. This will run the “waf” script used by many projects. If wafPath (default ./waf) doesn’t exist, it will copy the version of waf available in Nixpkgs. wafFlags can be used to pass flags to the waf script.

6.7.35. scons

Overrides the build, install, and check phases. This uses the scons build system as a replacement for make. scons does not provide a configure phase, so everything is managed at build and install time.

6.8. Purity in Nixpkgs

Measures taken to prevent dependencies on packages outside the store, and what you can do to prevent them.

GCC doesn’t search in locations such as /usr/include. In fact, attempts to add such directories through the -I flag are filtered out. Likewise, the linker (from GNU binutils) doesn’t search in standard locations such as /usr/lib. Programs built on Linux are linked against a GNU C Library that likewise doesn’t search in the default system locations.

6.9. Hardening in Nixpkgs

There are flags available to harden packages at compile or link-time. These can be toggled using the stdenv.mkDerivation parameters hardeningDisable and hardeningEnable.

Both parameters take a list of flags as strings. The special "all" flag can be passed to hardeningDisable to turn off all hardening. These flags can also be used as environment variables for testing or development purposes.

For more in-depth information on these hardening flags and hardening in general, refer to the Debian Wiki, Ubuntu Wiki, Gentoo Wiki, and the Arch Wiki.

6.9.1. Hardening flags enabled by default

The following flags are enabled by default and might require disabling with hardeningDisable if the program to package is incompatible. format

Adds the -Wformat -Wformat-security -Werror=format-security compiler options. At present, this warns about calls to printf and scanf functions where the format string is not a string literal and there are no format arguments, as in printf(foo);. This may be a security hole if the format string came from untrusted input and contains %n.

This needs to be turned off or fixed for errors similar to:

/tmp/nix-build-zynaddsubfx-2.5.2.drv-0/zynaddsubfx-2.5.2/src/UI/guimain.cpp:571:28: error: format not a string literal and no format arguments [-Werror=format-security]
cc1plus: some warnings being treated as errors stackprotector

Adds the -fstack-protector-strong --param ssp-buffer-size=4 compiler options. This adds safety checks against stack overwrites rendering many potential code injection attacks into aborting situations. In the best case this turns code injection vulnerabilities into denial of service or into non-issues (depending on the application).

This needs to be turned off or fixed for errors similar to:

bin/blib.a(bios_console.o): In function `bios_handle_cup':
/tmp/nix-build-ipxe-20141124-5cbdc41.drv-0/ipxe-5cbdc41/src/arch/i386/firmware/pcbios/bios_console.c:86: undefined reference to `__stack_chk_fail' fortify

Adds the -O2 -D_FORTIFY_SOURCE=2 compiler options. During code generation the compiler knows a great deal of information about buffer sizes (where possible), and attempts to replace insecure unlimited length buffer function calls with length-limited ones. This is especially useful for old, crufty code. Additionally, format strings in writable memory that contain %n are blocked. If an application depends on such a format string, it will need to be worked around.

Additionally, some warnings are enabled which might trigger build failures if compiler warnings are treated as errors in the package build. In this case, set NIX_CFLAGS_COMPILE to -Wno-error=warning-type.

This needs to be turned off or fixed for errors similar to:

malloc.c:404:15: error: return type is an incomplete type
malloc.c:410:19: error: storage size of 'ms' isn't known

strdup.h:22:1: error: expected identifier or '(' before '__extension__'

strsep.c:65:23: error: register name not specified for 'delim'

installwatch.c:3751:5: error: conflicting types for '__open_2'

fcntl2.h:50:4: error: call to '__open_missing_mode' declared with attribute error: open with O_CREAT or O_TMPFILE in second argument needs 3 arguments pic

Adds the -fPIC compiler options. This options adds support for position independent code in shared libraries and thus making ASLR possible.

Most notably, the Linux kernel, kernel modules and other code not running in an operating system environment like boot loaders won’t build with PIC enabled. The compiler will is most cases complain that PIC is not supported for a specific build.

This needs to be turned off or fixed for assembler errors similar to:

ccbLfRgg.s: Assembler messages:
ccbLfRgg.s:33: Error: missing or invalid displacement expression `private_key_len@GOTOFF' strictoverflow

Signed integer overflow is undefined behaviour according to the C standard. If it happens, it is an error in the program as it should check for overflow before it can happen, not afterwards. GCC provides built-in functions to perform arithmetic with overflow checking, which are correct and faster than any custom implementation. As a workaround, the option -fno-strict-overflow makes gcc behave as if signed integer overflows were defined.

This flag should not trigger any build or runtime errors. relro

Adds the -z relro linker option. During program load, several ELF memory sections need to be written to by the linker, but can be turned read-only before turning over control to the program. This prevents some GOT (and .dtors) overwrite attacks, but at least the part of the GOT used by the dynamic linker (.got.plt) is still vulnerable.

This flag can break dynamic shared object loading. For instance, the module systems of Xorg and OpenCV are incompatible with this flag. In almost all cases the bindnow flag must also be disabled and incompatible programs typically fail with similar errors at runtime. bindnow

Adds the -z bindnow linker option. During program load, all dynamic symbols are resolved, allowing for the complete GOT to be marked read-only (due to relro). This prevents GOT overwrite attacks. For very large applications, this can incur some performance loss during initial load while symbols are resolved, but this shouldn’t be an issue for daemons.

This flag can break dynamic shared object loading. For instance, the module systems of Xorg and PHP are incompatible with this flag. Programs incompatible with this flag often fail at runtime due to missing symbols, like: undefined symbol: vgaHWFreeHWRec

6.9.2. Hardening flags disabled by default

The following flags are disabled by default and should be enabled with hardeningEnable for packages that take untrusted input like network services. pie

This flag is disabled by default for normal glibc based NixOS package builds, but enabled by default for musl based package builds.

Adds the -fPIE compiler and -pie linker options. Position Independent Executables are needed to take advantage of Address Space Layout Randomization, supported by modern kernel versions. While ASLR can already be enforced for data areas in the stack and heap (brk and mmap), the code areas must be compiled as position-independent. Shared libraries already do this with the pic flag, so they gain ASLR automatically, but binary .text regions need to be build with pie to gain ASLR. When this happens, ROP attacks are much harder since there are no static locations to bounce off of during a memory corruption attack.

Static libraries need to be compiled with -fPIE so that executables can link them in with the -pie linker option. If the libraries lack -fPIE, you will get the error recompile with -fPIE.

[1] The build platform is ignored because it is a mere implementation detail of the package satisfying the dependency: As a general programming principle, dependencies are always specified as interfaces, not concrete implementation.

[2] Currently, this means for native builds all dependencies are put on the PATH. But in the future that may not be the case for sake of matching cross: the platforms would be assumed to be unique for native and cross builds alike, so only the depsBuild* and nativeBuildInputs would be added to the PATH.

[3] Nix itself already takes a package’s transitive dependencies into account, but this propagation ensures nixpkgs-specific infrastructure like setup hooks (mentioned above) also are run as if the propagated dependency.

[4] The findInputs function, currently residing in pkgs/stdenv/generic/, implements the propagation logic.

[5] It clears the sys_lib_*search_path variables in the Libtool script to prevent Libtool from using libraries in /usr/lib and such.

[6] Eventually these will be passed building natively as well, to improve determinism: build-time guessing, as is done today, is a risk of impurity.

[7] Each wrapper targets a single platform, so if binaries for multiple platforms are needed, the underlying binaries must be wrapped multiple times. As this is a property of the wrapper itself, the multiple wrappings are needed whether or not the same underlying binaries can target multiple platforms.

Chapter 7. Meta-attributes

Nix packages can declare meta-attributes that contain information about a package such as a description, its homepage, its license, and so on. For instance, the GNU Hello package has a meta declaration like this:

meta = with lib; {
  description = "A program that produces a familiar, friendly greeting";
  longDescription = ''
    GNU Hello is a program that prints "Hello, world!" when you run it.
    It is fully customizable.
  homepage = "";
  license = licenses.gpl3Plus;
  maintainers = [ maintainers.eelco ];
  platforms = platforms.all;

Meta-attributes are not passed to the builder of the package. Thus, a change to a meta-attribute doesn’t trigger a recompilation of the package. The value of a meta-attribute must be a string.

The meta-attributes of a package can be queried from the command-line using nix-env:

$ nix-env -qa hello --json
    "hello": {
        "meta": {
            "description": "A program that produces a familiar, friendly greeting",
            "homepage": "",
            "license": {
                "fullName": "GNU General Public License version 3 or later",
                "shortName": "GPLv3+",
                "url": ""
            "longDescription": "GNU Hello is a program that prints \"Hello, world!\" when you run it.\nIt is fully customizable.\n",
            "maintainers": [
                "Ludovic Court\u00e8s <>"
            "platforms": [
            "position": "/home/user/dev/nixpkgs/pkgs/applications/misc/hello/default.nix:14"
        "name": "hello-2.9",
        "system": "x86_64-linux"

nix-env knows about the description field specifically:

$ nix-env -qa hello --description
hello-2.3  A program that produces a familiar, friendly greeting

7.1. Standard meta-attributes

It is expected that each meta-attribute is one of the following:

7.1.1. description

A short (one-line) description of the package. This is shown by nix-env -q --description and also on the Nixpkgs release pages.

Don’t include a period at the end. Don’t include newline characters. Capitalise the first character. For brevity, don’t repeat the name of package — just describe what it does.

Wrong: "libpng is a library that allows you to decode PNG images."

Right: "A library for decoding PNG images"

7.1.2. longDescription

An arbitrarily long description of the package.

7.1.3. branch

Release branch. Used to specify that a package is not going to receive updates that are not in this branch; for example, Linux kernel 3.0 is supposed to be updated to 3.0.X, not 3.1.

7.1.4. homepage

The package’s homepage. Example:

7.1.5. downloadPage

The page where a link to the current version can be found. Example:

7.1.6. changelog

A link or a list of links to the location of Changelog for a package. A link may use expansion to refer to the correct changelog version. Example: "${version}"

7.1.7. license

The license, or licenses, for the package. One from the attribute set defined in nixpkgs/lib/licenses.nix. At this moment using both a list of licenses and a single license is valid. If the license field is in the form of a list representation, then it means that parts of the package are licensed differently. Each license should preferably be referenced by their attribute. The non-list attribute value can also be a space delimited string representation of the contained attribute shortNames or spdxIds. The following are all valid examples:

  • Single license referenced by attribute (preferred) lib.licenses.gpl3Only.

  • Single license referenced by its attribute shortName (frowned upon) "gpl3Only".

  • Single license referenced by its attribute spdxId (frowned upon) "GPL-3.0-only".

  • Multiple licenses referenced by attribute (preferred) with lib.licenses; [ asl20 free ofl ].

  • Multiple licenses referenced as a space delimited string of attribute shortNames (frowned upon) "asl20 free ofl".

For details, see Licenses.

7.1.8. maintainers

A list of the maintainers of this Nix expression. Maintainers are defined in nixpkgs/maintainers/maintainer-list.nix. There is no restriction to becoming a maintainer, just add yourself to that list in a separate commit titled “maintainers: add alice”, and reference maintainers with maintainers = with lib.maintainers; [ alice bob ].

7.1.9. mainProgram

The name of the main binary for the package. This effects the binary nix run executes and falls back to the name of the package. Example: "rg"

7.1.10. priority

The priority of the package, used by nix-env to resolve file name conflicts between packages. See the Nix manual page for nix-env for details. Example: "10" (a low-priority package).

7.1.11. platforms

The list of Nix platform types on which the package is supported. Hydra builds packages according to the platform specified. If no platform is specified, the package does not have prebuilt binaries. An example is:

meta.platforms = lib.platforms.linux;

Attribute Set lib.platforms defines various common lists of platforms types.

7.1.12. tests

Warning: This attribute is special in that it is not actually under the meta attribute set but rather under the passthru attribute set. This is due to how meta attributes work, and the fact that they are supposed to contain only metadata, not derivations.

An attribute set with tests as values. A test is a derivation that builds when the test passes and fails to build otherwise.

You can run these tests with:

$ cd path/to/nixpkgs
$ nix-build -A your-package.tests Package tests

Tests that are part of the source package are often executed in the installCheckPhase.

Prefer passthru.tests for tests that are introduced in nixpkgs because:

  • passthru.tests tests the real package, independently from the environment in which it was built

  • we can run passthru.tests independently

  • installCheckPhase adds overhead to each build

For more on how to write and run package tests, see Section 18.7, “Package tests”. NixOS tests

The NixOS tests are available as nixosTests in parameters of derivations. For instance, the OpenSMTPD derivation includes lines similar to:

{ /* ... */, nixosTests }:
  # ...
  passthru.tests = {
    basic-functionality-and-dovecot-integration = nixosTests.opensmtpd;

NixOS tests run in a VM, so they are slower than regular package tests. For more information see NixOS module tests.

7.1.13. timeout

A timeout (in seconds) for building the derivation. If the derivation takes longer than this time to build, it can fail due to breaking the timeout. However, all computers do not have the same computing power, hence some builders may decide to apply a multiplicative factor to this value. When filling this value in, try to keep it approximately consistent with other values already present in nixpkgs.

7.1.14. hydraPlatforms

The list of Nix platform types for which the Hydra instance at will build the package. (Hydra is the Nix-based continuous build system.) It defaults to the value of meta.platforms. Thus, the only reason to set meta.hydraPlatforms is if you want to build the package on a subset of meta.platforms, or not at all, e.g.

meta.platforms = lib.platforms.linux;
meta.hydraPlatforms = [];

7.1.15. broken

If set to true, the package is marked as broken, meaning that it won’t show up in nix-env -qa, and cannot be built or installed. Such packages should be removed from Nixpkgs eventually unless they are fixed.

7.1.16. updateWalker

If set to true, the package is tested to be updated correctly by the script without additional settings. Such packages have meta.version set and their homepage (or the page specified by meta.downloadPage) contains a direct link to the package tarball.

7.2. Licenses

The meta.license attribute should preferably contain a value from lib.licenses defined in nixpkgs/lib/licenses.nix, or in-place license description of the same format if the license is unlikely to be useful in another expression.

Although it’s typically better to indicate the specific license, a few generic options are available:

7.2.1., "free"

Catch-all for free software licenses not listed above.

7.2.2. lib.licenses.unfreeRedistributable, "unfree-redistributable"

Unfree package that can be redistributed in binary form. That is, it’s legal to redistribute the output of the derivation. This means that the package can be included in the Nixpkgs channel.

Sometimes proprietary software can only be redistributed unmodified. Make sure the builder doesn’t actually modify the original binaries; otherwise we’re breaking the license. For instance, the NVIDIA X11 drivers can be redistributed unmodified, but our builder applies patchelf to make them work. Thus, its license is "unfree" and it cannot be included in the Nixpkgs channel.

7.2.3. lib.licenses.unfree, "unfree"

Unfree package that cannot be redistributed. You can build it yourself, but you cannot redistribute the output of the derivation. Thus it cannot be included in the Nixpkgs channel.

7.2.4. lib.licenses.unfreeRedistributableFirmware, "unfree-redistributable-firmware"

This package supplies unfree, redistributable firmware. This is a separate value from unfree-redistributable because not everybody cares whether firmware is free.

Chapter 8. Multiple-output packages

8.1. Introduction

The Nix language allows a derivation to produce multiple outputs, which is similar to what is utilized by other Linux distribution packaging systems. The outputs reside in separate Nix store paths, so they can be mostly handled independently of each other, including passing to build inputs, garbage collection or binary substitution. The exception is that building from source always produces all the outputs.

The main motivation is to save disk space by reducing runtime closure sizes; consequently also sizes of substituted binaries get reduced. Splitting can be used to have more granular runtime dependencies, for example the typical reduction is to split away development-only files, as those are typically not needed during runtime. As a result, closure sizes of many packages can get reduced to a half or even much less.

Note: The reduction effects could be instead achieved by building the parts in completely separate derivations. That would often additionally reduce build-time closures, but it tends to be much harder to write such derivations, as build systems typically assume all parts are being built at once. This compromise approach of single source package producing multiple binary packages is also utilized often by rpm and deb.

A number of attributes can be used to work with a derivation with multiple outputs. The attribute outputs is a list of strings, which are the names of the outputs. For each of these names, an identically named attribute is created, corresponding to that output. The attribute meta.outputsToInstall is used to determine the default set of outputs to install when using the derivation name unqualified.

8.2. Installing a split package

When installing a package with multiple outputs, the package’s meta.outputsToInstall attribute determines which outputs are actually installed. meta.outputsToInstall is a list whose default installs binaries and the associated man pages. The following sections describe ways to install different outputs.

8.2.1. Selecting outputs to install via NixOS

NixOS provides two ways to select the outputs to install for packages listed in environment.systemPackages:

  • The configuration option environment.extraOutputsToInstall is appended to each package’s meta.outputsToInstall attribute to determine the outputs to install. It can for example be used to install info documentation or debug symbols for all packages.

  • The outputs can be listed as packages in environment.systemPackages. For example, the "out" and "info" outputs for the coreutils package can be installed by including coreutils and in environment.systemPackages.

8.2.2. Selecting outputs to install via nix-env

nix-env lacks an easy way to select the outputs to install. When installing a package, nix-env always installs the outputs listed in meta.outputsToInstall, even when the user explicitly selects an output.

nix-env silenty disregards the outputs selected by the user, and instead installs the outputs from meta.outputsToInstall. For example,

$ nix-env -iA

installs the "out" output (coreutils.meta.outputsToInstall is [ "out" ]) instead of the requested "info".

The only recourse to select an output with nix-env is to override the package’s meta.outputsToInstall, using the functions described in Chapter 4, Overriding. For example, the following overlay adds the "info" output for the coreutils package:

self: super:
  coreutils = super.coreutils.overrideAttrs (oldAttrs: {
    meta = oldAttrs.meta // { outputsToInstall = oldAttrs.meta.outputsToInstall or [ "out" ] ++ [ "info" ]; };

8.3. Using a split package

In the Nix language the individual outputs can be reached explicitly as attributes, e.g., but the typical case is just using packages as build inputs.

When a multiple-output derivation gets into a build input of another derivation, the dev output is added if it exists, otherwise the first output is added. In addition to that, propagatedBuildOutputs of that package which by default contain $outputBin and $outputLib are also added. (See Section 8.4.2, “File type groups”.)

In some cases it may be desirable to combine different outputs under a single store path. A function symlinkJoin can be used to do this. (Note that it may negate some closure size benefits of using a multiple-output package.)

8.4. Writing a split derivation

Here you find how to write a derivation that produces multiple outputs.

In nixpkgs there is a framework supporting multiple-output derivations. It tries to cover most cases by default behavior. You can find the source separated in <nixpkgs/pkgs/build-support/setup-hooks/>; it’s relatively well-readable. The whole machinery is triggered by defining the outputs attribute to contain the list of desired output names (strings).

outputs = [ "bin" "dev" "out" "doc" ];

Often such a single line is enough. For each output an equally named environment variable is passed to the builder and contains the path in nix store for that output. Typically you also want to have the main out output, as it catches any files that didn’t get elsewhere.

Note: There is a special handling of the debug output, described at Section, “separateDebugInfo.

8.4.1. “Binaries first”

A commonly adopted convention in nixpkgs is that executables provided by the package are contained within its first output. This convention allows the dependent packages to reference the executables provided by packages in a uniform manner. For instance, provided with the knowledge that the perl package contains a perl executable it can be referenced as ${pkgs.perl}/bin/perl within a Nix derivation that needs to execute a Perl script.

The glibc package is a deliberate single exception to the “binaries first” convention. The glibc has libs as its first output allowing the libraries provided by glibc to be referenced directly (e.g. ${stdenv.glibc}/lib/ The executables provided by glibc can be accessed via its bin attribute (e.g. ${stdenv.glibc.bin}/bin/ldd).

The reason for why glibc deviates from the convention is because referencing a library provided by glibc is a very common operation among Nix packages. For instance, third-party executables packaged by Nix are typically patched and relinked with the relevant version of glibc libraries from Nix packages (please see the documentation on patchelf for more details).

8.4.2. File type groups

The support code currently recognizes some particular kinds of outputs and either instructs the build system of the package to put files into their desired outputs or it moves the files during the fixup phase. Each group of file types has an outputFoo variable specifying the output name where they should go. If that variable isn’t defined by the derivation writer, it is guessed – a default output name is defined, falling back to other possibilities if the output isn’t defined. $outputDev

is for development-only files. These include C(++) headers (include/), pkg-config (lib/pkgconfig/), cmake (lib/cmake/) and aclocal files (share/aclocal/). They go to dev or out by default. $outputBin

is meant for user-facing binaries, typically residing in bin/. They go to bin or out by default. $outputLib

is meant for libraries, typically residing in lib/ and libexec/. They go to lib or out by default. $outputDoc

is for user documentation, typically residing in share/doc/. It goes to doc or out by default. $outputDevdoc

is for developer documentation. Currently we count gtk-doc and devhelp books, typically residing in share/gtk-doc/ and share/devhelp/, in there. It goes to devdoc or is removed (!) by default. This is because e.g. gtk-doc tends to be rather large and completely unused by nixpkgs users. $outputMan

is for man pages (except for section 3), typically residing in share/man/man[0-9]/. They go to man or $outputBin by default. $outputDevman

is for section 3 man pages, typically residing in share/man/man[0-9]/. They go to devman or $outputMan by default. $outputInfo

is for info pages, typically residing in share/info/. They go to info or $outputBin by default.

8.4.3. Common caveats

  • Some configure scripts don’t like some of the parameters passed by default by the framework, e.g. --docdir=/foo/bar. You can disable this by setting setOutputFlags = false;.

  • The outputs of a single derivation can retain references to each other, but note that circular references are not allowed. (And each strongly-connected component would act as a single output anyway.)

  • Most of split packages contain their core functionality in libraries. These libraries tend to refer to various kind of data that typically gets into out, e.g. locale strings, so there is often no advantage in separating the libraries into lib, as keeping them in out is easier.

  • Some packages have hidden assumptions on install paths, which complicates splitting.

Chapter 9. Cross-compilation

9.1. Introduction

Cross-compilation means compiling a program on one machine for another type of machine. For example, a typical use of cross-compilation is to compile programs for embedded devices. These devices often don’t have the computing power and memory to compile their own programs. One might think that cross-compilation is a fairly niche concern. However, there are significant advantages to rigorously distinguishing between build-time and run-time environments! Significant, because the benefits apply even when one is developing and deploying on the same machine. Nixpkgs is increasingly adopting the opinion that packages should be written with cross-compilation in mind, and Nixpkgs should evaluate in a similar way (by minimizing cross-compilation-specific special cases) whether or not one is cross-compiling.

This chapter will be organized in three parts. First, it will describe the basics of how to package software in a way that supports cross-compilation. Second, it will describe how to use Nixpkgs when cross-compiling. Third, it will describe the internal infrastructure supporting cross-compilation.

9.2. Packaging in a cross-friendly manner

9.2.1. Platform parameters

Nixpkgs follows the conventions of GNU autoconf. We distinguish between 3 types of platforms when building a derivation: build, host, and target. In summary, build is the platform on which a package is being built, host is the platform on which it will run. The third attribute, target, is relevant only for certain specific compilers and build tools.

In Nixpkgs, these three platforms are defined as attribute sets under the names buildPlatform, hostPlatform, and targetPlatform. They are always defined as attributes in the standard environment. That means one can access them like:

{ stdenv, fooDep, barDep, ... }: ...stdenv.buildPlatform...

The build platform is the platform on which a package is built. Once someone has a built package, or pre-built binary package, the build platform should not matter and can be ignored.


The host platform is the platform on which a package will be run. This is the simplest platform to understand, but also the one with the worst name.


The target platform attribute is, unlike the other two attributes, not actually fundamental to the process of building software. Instead, it is only relevant for compatibility with building certain specific compilers and build tools. It can be safely ignored for all other packages.

The build process of certain compilers is written in such a way that the compiler resulting from a single build can itself only produce binaries for a single platform. The task of specifying this single target platform is thus pushed to build time of the compiler. The root cause of this is that the compiler (which will be run on the host) and the standard library/runtime (which will be run on the target) are built by a single build process.

There is no fundamental need to think about a single target ahead of time like this. If the tool supports modular or pluggable backends, both the need to specify the target at build time and the constraint of having only a single target disappear. An example of such a tool is LLVM.

Although the existence of a target platform is arguably a historical mistake, it is a common one: examples of tools that suffer from it are GCC, Binutils, GHC and Autoconf. Nixpkgs tries to avoid sharing in the mistake where possible. Still, because the concept of a target platform is so ingrained, it is best to support it as is.

The exact schema these fields follow is a bit ill-defined due to a long and convoluted evolution, but this is slowly being cleaned up. You can see examples of ones used in practice in; note how they are not all very consistent. For now, here are few fields can count on them containing:


This is a two-component shorthand for the platform. Examples of this would be x86_64-darwin and i686-linux; see for more. The first component corresponds to the CPU architecture of the platform and the second to the operating system of the platform ([cpu]-[os]). This format has built-in support in Nix, such as the builtins.currentSystem impure string.


This is a 3- or 4- component shorthand for the platform. Examples of this would be x86_64-unknown-linux-gnu and aarch64-apple-darwin14. This is a standard format called the LLVM target triple, as they are pioneered by LLVM. In the 4-part form, this corresponds to [cpu]-[vendor]-[os]-[abi]. This format is strictly more informative than the Nix host double, as the previous format could analogously be termed. This needs a better name than config!


This is a Nix representation of a parsed LLVM target triple with white-listed components. This can be specified directly, or actually parsed from the config. See for the exact representation.


This is a string identifying the standard C library used. Valid identifiers include glibc for GNU libc, libSystem for Darwin’s Libsystem, and uclibc for µClibc. It should probably be refactored to use the module system, like parse.


These predicates are defined in, and slapped onto every platform. They are superior to the ones in stdenv as they force the user to be explicit about which platform they are inspecting. Please use these instead of those.


This is, quite frankly, a dumping ground of ad-hoc settings (it’s an attribute set). See for examples—there’s hopefully one in there that will work verbatim for each platform that is working. Please help us triage these flags and give them better homes!

9.2.2. Theory of dependency categorization

Note: This is a rather philosophical description that isn’t very Nixpkgs-specific. For an overview of all the relevant attributes given to mkDerivation, see Section 6.3, “Specifying dependencies”. For a description of how everything is implemented, see Section 9.4.1, “Implementation of dependencies”.

In this section we explore the relationship between both runtime and build-time dependencies and the 3 Autoconf platforms.

A run time dependency between two packages requires that their host platforms match. This is directly implied by the meaning of host platform and runtime dependency: The package dependency exists while both packages are running on a single host platform.

A build time dependency, however, has a shift in platforms between the depending package and the depended-on package. build time dependency means that to build the depending package we need to be able to run the depended-on’s package. The depending package’s build platform is therefore equal to the depended-on package’s host platform.

If both the dependency and depending packages aren’t compilers or other machine-code-producing tools, we’re done. And indeed buildInputs and nativeBuildInputs have covered these simpler cases for many years. But if the dependency does produce machine code, we might need to worry about its target platform too. In principle, that target platform might be any of the depending package’s build, host, or target platforms, but we prohibit dependencies from a later platform to an earlier platform to limit confusion because we’ve never seen a legitimate use for them.

Finally, if the depending package is a compiler or other machine-code-producing tool, it might need dependencies that run at emit time. This is for compilers that (regrettably) insist on being built together with their source languages’ standard libraries. Assuming build != host != target, a run-time dependency of the standard library cannot be run at the compiler’s build time or run time, but only at the run time of code emitted by the compiler.

Putting this all together, that means we have dependencies in the form host → target, in at most the following six combinations: Possible dependency types

Dependency’s host platform Dependency’s target platform
build build
build host
build target
host host
host target
target target

Some examples will make this table clearer. Suppose there’s some package that is being built with a (build, host, target) platform triple of (foo, bar, baz). If it has a build-time library dependency, that would be a host → build dependency with a triple of (foo, foo, *) (the target platform is irrelevant). If it needs a compiler to be built, that would be a build → host dependency with a triple of (foo, foo, *) (the target platform is irrelevant). That compiler, would be built with another compiler, also build → host dependency, with a triple of (foo, foo, foo).

9.2.3. Cross packaging cookbook

Some frequently encountered problems when packaging for cross-compilation should be answered here. Ideally, the information above is exhaustive, so this section cannot provide any new information, but it is ludicrous and cruel to expect everyone to spend effort working through the interaction of many features just to figure out the same answer to the same common problem. Feel free to add to this list! My package fails to find a binutils command (cc/ar/ld etc.)

Many packages assume that an unprefixed binutils (cc/ar/ld etc.) is available, but Nix doesn’t provide one. It only provides a prefixed one, just as it only does for all the other binutils programs. It may be necessary to patch the package to fix the build system to use a prefix. For instance, instead of cc, use ${}cc.

makeFlags = [ "CC=${}cc" ]; How do I avoid compiling a GCC cross-compiler from source?

On less powerful machines, it can be inconvenient to cross-compile a package only to find out that GCC has to be compiled from source, which could take up to several hours. Nixpkgs maintains a limited cross-related jobset on Hydra, which tests cross-compilation to various platforms from build platforms x86_64-darwin, x86_64-linux, and aarch64-linux. See pkgs/top-level/release-cross.nix for the full list of target platforms and packages. For instance, the following invocation fetches the pre-built cross-compiled GCC for armv6l-unknown-linux-gnueabihf and builds GNU Hello from source.

$ nix-build '<nixpkgs>' -A pkgsCross.raspberryPi.hello What if my package’s build system needs to build a C program to be run under the build environment?

Add the following to your mkDerivation invocation.

depsBuildBuild = [ ]; My package’s testsuite needs to run host platform code.

Add the following to your mkDerivation invocation.

doCheck = stdenv.hostPlatform == stdenv.buildPlatform;

9.3. Cross-building packages

Nixpkgs can be instantiated with localSystem alone, in which case there is no cross-compiling and everything is built by and for that system, or also with crossSystem, in which case packages run on the latter, but all building happens on the former. Both parameters take the same schema as the 3 (build, host, and target) platforms defined in the previous section. As mentioned above, has some platforms which are used as arguments for these parameters in practice. You can use them programmatically, or on the command line:

$ nix-build '<nixpkgs>' --arg crossSystem '(import <nixpkgs/lib>).systems.examples.fooBarBaz' -A whatever


Eventually we would like to make these platform examples an unnecessary convenience so that

$ nix-build '<nixpkgs>' --arg crossSystem '{ config = "<arch>-<os>-<vendor>-<abi>"; }' -A whatever

works in the vast majority of cases. The problem today is dependencies on other sorts of configuration which aren’t given proper defaults. We rely on the examples to crudely to set those configuration parameters in some vaguely sane manner on the users behalf. Issue #34274 tracks this inconvenience along with its root cause in crufty configuration options.

While one is free to pass both parameters in full, there’s a lot of logic to fill in missing fields. As discussed in the previous section, only one of system, config, and parsed is needed to infer the other two. Additionally, libc will be inferred from parse. Finally, localSystem.system is also impurely inferred based on the platform evaluation occurs. This means it is often not necessary to pass localSystem at all, as in the command-line example in the previous paragraph.

Note: Many sources (manual, wiki, etc) probably mention passing system, platform, along with the optional crossSystem to Nixpkgs: import <nixpkgs> { system = ..; platform = ..; crossSystem = ..; }. Passing those two instead of localSystem is still supported for compatibility, but is discouraged. Indeed, much of the inference we do for these parameters is motivated by compatibility as much as convenience.

One would think that localSystem and crossSystem overlap horribly with the three *Platforms (buildPlatform, hostPlatform, and targetPlatform; see stage.nix or the manual). Actually, those identifiers are purposefully not used here to draw a subtle but important distinction: While the granularity of having 3 platforms is necessary to properly build packages, it is overkill for specifying the user’s intent when making a build plan or package set. A simple build vs deploy dichotomy is adequate: the sliding window principle described in the previous section shows how to interpolate between the these two end points to get the 3 platform triple for each bootstrapping stage. That means for any package a given package set, even those not bound on the top level but only reachable via dependencies or buildPackages, the three platforms will be defined as one of localSystem or crossSystem, with the former replacing the latter as one traverses build-time dependencies. A last simple difference is that crossSystem should be null when one doesn’t want to cross-compile, while the *Platforms are always non-null. localSystem is always non-null.

9.4. Cross-compilation infrastructure

9.4.1. Implementation of dependencies

The categories of dependencies developed in Section 9.2.2, “Theory of dependency categorization” are specified as lists of derivations given to mkDerivation, as documented in Section 6.3, “Specifying dependencies”. In short, each list of dependencies for host → target is called deps<host><target> (where host, and target values are either build, host, or target), with exceptions for backwards compatibility that depsBuildHost is instead called nativeBuildInputs and depsHostTarget is instead called buildInputs. Nixpkgs is now structured so that each deps<host><target> is automatically taken from pkgs<host><target>. (These pkgs<host><target>s are quite new, so there is no special case for nativeBuildInputs and buildInputs.) For example, pkgsBuildHost.gcc should be used at build-time, while pkgsHostTarget.gcc should be used at run-time.

Now, for most of Nixpkgs’s history, there were no pkgs<host><target> attributes, and most packages have not been refactored to use it explicitly. Prior to those, there were just buildPackages, pkgs, and targetPackages. Those are now redefined as aliases to pkgsBuildHost, pkgsHostTarget, and pkgsTargetTarget. It is acceptable, even recommended, to use them for libraries to show that the host platform is irrelevant.

But before that, there was just pkgs, even though both buildInputs and nativeBuildInputs existed. [Cross barely worked, and those were implemented with some hacks on mkDerivation to override dependencies.] What this means is the vast majority of packages do not use any explicit package set to populate their dependencies, just using whatever callPackage gives them even if they do correctly sort their dependencies into the multiple lists described above. And indeed, asking that users both sort their dependencies, and take them from the right attribute set, is both too onerous and redundant, so the recommended approach (for now) is to continue just categorizing by list and not using an explicit package set.

To make this work, we splice together the six pkgsFooBar package sets and have callPackage actually take its arguments from that. This is currently implemented in pkgs/top-level/splice.nix. mkDerivation then, for each dependency attribute, pulls the right derivation out from the splice. This splicing can be skipped when not cross-compiling as the package sets are the same, but still is a bit slow for cross-compiling. We’d like to do something better, but haven’t come up with anything yet.

9.4.2. Bootstrapping

Each of the package sets described above come from a single bootstrapping stage. While pkgs/top-level/default.nix, coordinates the composition of stages at a high level, pkgs/top-level/stage.nix ties the knot (creates the fixed point) of each stage. The package sets are defined per-stage however, so they can be thought of as edges between stages (the nodes) in a graph. Compositions like pkgsBuildTarget.targetPackages can be thought of as paths to this graph.

While there are many package sets, and thus many edges, the stages can also be arranged in a linear chain. In other words, many of the edges are redundant as far as connectivity is concerned. This hinges on the type of bootstrapping we do. Currently for cross it is:

  1. (native, native, native)

  2. (native, native, foreign)

  3. (native, foreign, foreign)

In each stage, pkgsBuildHost refers to the previous stage, pkgsBuildBuild refers to the one before that, and pkgsHostTarget refers to the current one, and pkgsTargetTarget refers to the next one. When there is no previous or next stage, they instead refer to the current stage. Note how all the invariants regarding the mapping between dependency and depending packages’ build host and target platforms are preserved. pkgsBuildTarget and pkgsHostHost are more complex in that the stage fitting the requirements isn’t always a fixed chain of prevs and nexts away (modulo the saturating self-references at the ends). We just special case each instead. All the primary edges are implemented is in pkgs/stdenv/booter.nix, and secondarily aliases in pkgs/top-level/stage.nix.

Note: The native stages are bootstrapped in legacy ways that predate the current cross implementation. This is why the bootstrapping stages leading up to the final stages are ignored in the previous paragraph.

If one looks at the 3 platform triples, one can see that they overlap such that one could put them together into a chain like:

(native, native, native, foreign, foreign)

If one imagines the saturating self references at the end being replaced with infinite stages, and then overlays those platform triples, one ends up with the infinite tuple:

(native..., native, native, native, foreign, foreign, foreign...)

One can then imagine any sequence of platforms such that there are bootstrap stages with their 3 platforms determined by sliding a window that is the 3 tuple through the sequence. This was the original model for bootstrapping. Without a target platform (assume a better world where all compilers are multi-target and all standard libraries are built in their own derivation), this is sufficient. Conversely if one wishes to cross compile faster, with a Canadian Cross bootstrapping stage where build != host != target, more bootstrapping stages are needed since no sliding window provides the pesky pkgsBuildTarget package set since it skips the Canadian cross stage’s host.


It is much better to refer to buildPackages than targetPackages, or more broadly package sets that do not mention “target”. There are three reasons for this.

First, it is because bootstrapping stages do not have a unique targetPackages. For example a (x86-linux, x86-linux, arm-linux) and (x86-linux, x86-linux, x86-windows) package set both have a (x86-linux, x86-linux, x86-linux) package set. Because there is no canonical targetPackages for such a native (build == host == target) package set, we set their targetPackages

Second, it is because this is a frequent source of hard-to-follow infinite recursions / cycles. When only package sets that don’t mention target are used, the package set forms a directed acyclic graph. This means that all cycles that exist are confined to one stage. This means they are a lot smaller, and easier to follow in the code or a backtrace. It also means they are present in native and cross builds alike, and so more likely to be caught by CI and other users.

Thirdly, it is because everything target-mentioning only exists to accommodate compilers with lousy build systems that insist on the compiler itself and standard library being built together. Of course that is bad because bigger derivations means longer rebuilds. It is also problematic because it tends to make the standard libraries less like other libraries than they could be, complicating code and build systems alike. Because of the other problems, and because of these innate disadvantages, compilers ought to be packaged another way where possible.

Note: If one explores Nixpkgs, they will see derivations with names like gccCross. Such *Cross derivations is a holdover from before we properly distinguished between the host and target platforms—the derivation with “Cross” in the name covered the build = host != target case, while the other covered the host = target, with build platform the same or not based on whether one was using its .nativeDrv or .crossDrv. This ugliness will disappear soon.

Chapter 10. Platform Notes

Table of Contents

10.1. Darwin (macOS)

10.1. Darwin (macOS)

Some common issues when packaging software for Darwin:

  • The Darwin stdenv uses clang instead of gcc. When referring to the compiler $CC or cc will work in both cases. Some builds hardcode gcc/g++ in their build scripts, that can usually be fixed with using something like makeFlags = [ "CC=cc" ]; or by patching the build scripts.

    stdenv.mkDerivation {
      name = "libfoo-1.2.3";
      # ...
      buildPhase = ''
        $CC -o hello hello.c
  • On Darwin, libraries are linked using absolute paths, libraries are resolved by their install_name at link time. Sometimes packages won’t set this correctly causing the library lookups to fail at runtime. This can be fixed by adding extra linker flags or by running install_name_tool -id during the fixupPhase.

    stdenv.mkDerivation {
      name = "libfoo-1.2.3";
      # ...
      makeFlags = lib.optional stdenv.isDarwin "LDFLAGS=-Wl,-install_name,$(out)/lib/libfoo.dylib";
  • Even if the libraries are linked using absolute paths and resolved via their install_name correctly, tests can sometimes fail to run binaries. This happens because the checkPhase runs before the libraries are installed.

    This can usually be solved by running the tests after the installPhase or alternatively by using DYLD_LIBRARY_PATH. More information about this variable can be found in the dyld(1) manpage.

    dyld: Library not loaded: /nix/store/7hnmbscpayxzxrixrgxvvlifzlxdsdir-jq-1.5-lib/lib/libjq.1.dylib
    Referenced from: /private/tmp/nix-build-jq-1.5.drv-0/jq-1.5/tests/../jq
    Reason: image not found
    ./tests/jqtest: line 5: 75779 Abort trap: 6
    stdenv.mkDerivation {
      name = "libfoo-1.2.3";
      # ...
      doInstallCheck = true;
      installCheckTarget = "check";
  • Some packages assume xcode is available and use xcrun to resolve build tools like clang, etc. This causes errors like xcode-select: error: no developer tools were found at '/Applications/' while the build doesn’t actually depend on xcode.

    stdenv.mkDerivation {
      name = "libfoo-1.2.3";
      # ...
      prePatch = ''
        substituteInPlace Makefile \
            --replace '/usr/bin/xcrun clang' clang

    The package xcbuild can be used to build projects that really depend on Xcode. However, this replacement is not 100% compatible with Xcode and can occasionally cause issues.

Chapter 11. Fetchers

When using Nix, you will frequently need to download source code and other files from the internet. For this purpose, Nix provides the fixed output derivation feature and Nixpkgs provides various functions that implement the actual fetching from various protocols and services.

11.1. Caveats

Because fixed output derivations are identified by their hash, a common mistake is to update a fetcher’s URL or a version parameter, without updating the hash. This will cause the old contents to be used. So remember to always invalidate the hash argument.

For those who develop and maintain fetchers, a similar problem arises with changes to the implementation of a fetcher. These may cause a fixed output derivation to fail, but won’t normally be caught by tests because the supposed output is already in the store or cache. For the purpose of testing, you can use a trick that is embodied by the invalidateFetcherByDrvHash function. It uses the derivation name to create a unique output path per fetcher implementation, defeating the caching precisely where it would be harmful.

11.2. fetchurl and fetchzip

Two basic fetchers are fetchurl and fetchzip. Both of these have two required arguments, a URL and a hash. The hash is typically sha256, although many more hash algorithms are supported. Nixpkgs contributors are currently recommended to use sha256. This hash will be used by Nix to identify your source. A typical usage of fetchurl is provided below.

{ stdenv, fetchurl }:

stdenv.mkDerivation {
  name = "hello";
  src = fetchurl {
    url = "";
    sha256 = "1111111111111111111111111111111111111111111111111111";

The main difference between fetchurl and fetchzip is in how they store the contents. fetchurl will store the unaltered contents of the URL within the Nix store. fetchzip on the other hand will decompress the archive for you, making files and directories directly accessible in the future. fetchzip can only be used with archives. Despite the name, fetchzip is not limited to .zip files and can also be used with any tarball.

fetchpatch works very similarly to fetchurl with the same arguments expected. It expects patch files as a source and performs normalization on them before computing the checksum. For example it will remove comments or other unstable parts that are sometimes added by version control systems and can change over time.

Most other fetchers return a directory rather than a single file.

11.3. fetchsvn

Used with Subversion. Expects url to a Subversion directory, rev, and sha256.

11.4. fetchgit

Used with Git. Expects url to a Git repo, rev, and sha256. rev in this case can be full the git commit id (SHA1 hash) or a tag name like refs/tags/v1.0.

Additionally the following optional arguments can be given: fetchSubmodules = true makes fetchgit also fetch the submodules of a repository. If deepClone is set to true, the entire repository is cloned as opposing to just creating a shallow clone. deepClone = true also implies leaveDotGit = true which means that the .git directory of the clone won’t be removed after checkout.

11.5. fetchfossil

Used with Fossil. Expects url to a Fossil archive, rev, and sha256.

11.6. fetchcvs

Used with CVS. Expects cvsRoot, tag, and sha256.

11.7. fetchhg

Used with Mercurial. Expects url, rev, and sha256.

A number of fetcher functions wrap part of fetchurl and fetchzip. They are mainly convenience functions intended for commonly used destinations of source code in Nixpkgs. These wrapper fetchers are listed below.

11.8. fetchFromGitHub

fetchFromGitHub expects four arguments. owner is a string corresponding to the GitHub user or organization that controls this repository. repo corresponds to the name of the software repository. These are located at the top of every GitHub HTML page as owner/repo. rev corresponds to the Git commit hash or tag (e.g v1.0) that will be downloaded from Git. Finally, sha256 corresponds to the hash of the extracted directory. Again, other hash algorithms are also available but sha256 is currently preferred.

fetchFromGitHub uses fetchzip to download the source archive generated by GitHub for the specified revision. If leaveDotGit, deepClone or fetchSubmodules are set to true, fetchFromGitHub will use fetchgit instead. Refer to its section for documentation of these options.

11.9. fetchFromGitLab

This is used with GitLab repositories. The arguments expected are very similar to fetchFromGitHub above.

11.10. fetchFromGitiles

This is used with Gitiles repositories. The arguments expected are similar to fetchgit.

11.11. fetchFromBitbucket

This is used with BitBucket repositories. The arguments expected are very similar to fetchFromGitHub above.

11.12. fetchFromSavannah

This is used with Savannah repositories. The arguments expected are very similar to fetchFromGitHub above.

11.13. fetchFromRepoOrCz

This is used with repositories. The arguments expected are very similar to fetchFromGitHub above.

11.14. fetchFromSourcehut

This is used with sourcehut repositories. The arguments expected are very similar to fetchFromGitHub above. Don’t forget the tilde (~) in front of the user name!

Chapter 12. Trivial builders

Nixpkgs provides a couple of functions that help with building derivations. The most important one, stdenv.mkDerivation, has already been documented above. The following functions wrap stdenv.mkDerivation, making it easier to use in certain cases.

12.1. runCommand

This takes three arguments, name, env, and buildCommand. name is just the name that Nix will append to the store path in the same way that stdenv.mkDerivation uses its name attribute. env is an attribute set specifying environment variables that will be set for this derivation. These attributes are then passed to the wrapped stdenv.mkDerivation. buildCommand specifies the commands that will be run to create this derivation. Note that you will need to create $out for Nix to register the command as successful.

An example of using runCommand is provided below.

(import <nixpkgs> {}).runCommand "my-example" {} ''
  echo My example command is running

  mkdir $out

  echo I can write data to the Nix store > $out/message

  echo I can also run basic commands like:

  echo ls

  echo whoami

  echo date

12.2. runCommandCC

This works just like runCommand. The only difference is that it also provides a C compiler in buildCommand’s environment. To minimize your dependencies, you should only use this if you are sure you will need a C compiler as part of running your command.

12.3. runCommandLocal

Variant of runCommand that forces the derivation to be built locally, it is not substituted. This is intended for very cheap commands (<1s execution time). It saves on the network roundrip and can speed up a build.

Note: This sets allowSubstitutes to false, so only use runCommandLocal if you are certain the user will always have a builder for the system of the derivation. This should be true for most trivial use cases (e.g. just copying some files to a different location or adding symlinks), because there the system is usually the same as builtins.currentSystem.

12.4. writeTextFile, writeText, writeTextDir, writeScript, writeScriptBin

These functions write text to the Nix store. This is useful for creating scripts from Nix expressions. writeTextFile takes an attribute set and expects two arguments, name and text. name corresponds to the name used in the Nix store path. text will be the contents of the file. You can also set executable to true to make this file have the executable bit set.

Many more commands wrap writeTextFile including writeText, writeTextDir, writeScript, and writeScriptBin. These are convenience functions over writeTextFile.

12.5. writeShellApplication

This can be used to easily produce a shell script that has some dependencies (runtimeInputs). It automatically sets the PATH of the script to contain all of the listed inputs, sets some sanity shellopts (errexit, nounset, pipefail), and checks the resulting script with shellcheck.

For example, look at the following code:

writeShellApplication {
  name = "show-nixos-org";

  runtimeInputs = [ curl w3m ];

  text = ''
    curl -s '' | w3m -dump -T text/html

Unlike with normal writeShellScriptBin, there is no need to manually write out ${curl}/bin/curl, setting the PATH was handled by writeShellApplication. Moreover, the script is being checked with shellcheck for more strict validation.

12.6. symlinkJoin

This can be used to put many derivations into the same directory structure. It works by creating a new derivation and adding symlinks to each of the paths listed. It expects two arguments, name, and paths. name is the name used in the Nix store path for the created derivation. paths is a list of paths that will be symlinked. These paths can be to Nix store derivations or any other subdirectory contained within.

12.7. writeReferencesToFile

Writes the closure of transitive dependencies to a file.

This produces the equivalent of nix-store -q --requisites.

For example,

writeReferencesToFile (writeScriptBin "hi" ''${hello}/bin/hello'')

produces an output path /nix/store/<hash>-runtime-deps containing


You can see that this includes hi, the original input path, hello, which is a direct reference, but also the other paths that are indirectly required to run hello.

12.8. writeDirectReferencesToFile

Writes the set of references to the output file, that is, their immediate dependencies.

This produces the equivalent of nix-store -q --references.

For example,

writeDirectReferencesToFile (writeScriptBin "hi" ''${hello}/bin/hello'')

produces an output path /nix/store/<hash>-runtime-references containing


but none of hello’s dependencies, because those are not referenced directly by hi’s output.

Chapter 13. Special builders

This chapter describes several special builders.

13.1. buildFHSUserEnv

buildFHSUserEnv provides a way to build and run FHS-compatible lightweight sandboxes. It creates an isolated root with bound /nix/store, so its footprint in terms of disk space needed is quite small. This allows one to run software which is hard or unfeasible to patch for NixOS – 3rd-party source trees with FHS assumptions, games distributed as tarballs, software with integrity checking and/or external self-updated binaries. It uses Linux namespaces feature to create temporary lightweight environments which are destroyed after all child processes exit, without root user rights requirement. Accepted arguments are:

  • name Environment name.

  • targetPkgs Packages to be installed for the main host’s architecture (i.e. x86_64 on x86_64 installations). Along with libraries binaries are also installed.

  • multiPkgs Packages to be installed for all architectures supported by a host (i.e. i686 and x86_64 on x86_64 installations). Only libraries are installed by default.

  • extraBuildCommands Additional commands to be executed for finalizing the directory structure.

  • extraBuildCommandsMulti Like extraBuildCommands, but executed only on multilib architectures.

  • extraOutputsToInstall Additional derivation outputs to be linked for both target and multi-architecture packages.

  • extraInstallCommands Additional commands to be executed for finalizing the derivation with runner script.

  • runScript A command that would be executed inside the sandbox and passed all the command line arguments. It defaults to bash.

  • profile Optional script for /etc/profile within the sandbox.

One can create a simple environment using a shell.nix like that:

{ pkgs ? import <nixpkgs> {} }:

(pkgs.buildFHSUserEnv {
  name = "simple-x11-env";
  targetPkgs = pkgs: (with pkgs;
    [ udev
    ]) ++ (with pkgs.xorg;
    [ libX11
  multiPkgs = pkgs: (with pkgs;
    [ udev
  runScript = "bash";

Running nix-shell would then drop you into a shell with these libraries and binaries available. You can use this to run closed-source applications which expect FHS structure without hassles: simply change runScript to the application path, e.g. ./bin/ – relative paths are supported.

13.2. pkgs.mkShell

pkgs.mkShell is a special kind of derivation that is only useful when using it combined with nix-shell. It will in fact fail to instantiate when invoked with nix-build.

13.2.1. Usage

{ pkgs ? import <nixpkgs> {} }:
pkgs.mkShell {
  # specify which packages to add to the shell environment
  packages = [ pkgs.gnumake ];
  # add all the dependencies, of the given packages, to the shell environment
  inputsFrom = with pkgs; [ hello gnutar ];

13.3. invalidateFetcherByDrvHash

Use the derivation hash to invalidate the output via name, for testing.

Type: (a@{ name, ... } -> Derivation) -> a -> Derivation

Normally, fixed output derivations can and should be cached by their output hash only, but for testing we want to re-fetch everytime the fetcher changes.

Changes to the fetcher become apparent in the drvPath, which is a hash of how to fetch, rather than a fixed store path. By inserting this hash into the name, we can make sure to re-run the fetcher every time the fetcher changes.

This relies on the assumption that Nix isn’t clever enough to reuse its database of local store contents to optimize fetching.

You might notice that the salted name derives from the normal invocation, not the final derivation. invalidateFetcherByDrvHash has to invoke the fetcher function twice: once to get a derivation hash, and again to produce the final fixed output derivation.


tests.fetchgit = invalidateFetcherByDrvHash fetchgit {
  name = "nix-source";
  url = "";
  rev = "9d9dbe6ed05854e03811c361a3380e09183f4f4a";
  sha256 = "sha256-7DszvbCNTjpzGRmpIVAWXk20P0/XTrWZ79KSOGLrUWY=";

Chapter 14. Images

This chapter describes tools for creating various types of images.

14.1. pkgs.appimageTools

pkgs.appimageTools is a set of functions for extracting and wrapping AppImage files. They are meant to be used if traditional packaging from source is infeasible, or it would take too long. To quickly run an AppImage file, pkgs.appimage-run can be used as well.

Warning: The appimageTools API is unstable and may be subject to backwards-incompatible changes in the future.

14.1.1. AppImage formats

There are different formats for AppImages, see the specification for details.

  • Type 1 images are ISO 9660 files that are also ELF executables.

  • Type 2 images are ELF executables with an appended filesystem.

They can be told apart with file -k:

$ file -k type1.AppImage
type1.AppImage: ELF 64-bit LSB executable, x86-64, version 1 (SYSV) ISO 9660 CD-ROM filesystem data 'AppImage' (Lepton 3.x), scale 0-0,
spot sensor temperature 0.000000, unit celsius, color scheme 0, calibration: offset 0.000000, slope 0.000000, dynamically linked, interpreter /lib64/, for GNU/Linux 2.6.18, BuildID[sha1]=d629f6099d2344ad82818172add1d38c5e11bc6d, stripped\012- data

$ file -k type2.AppImage
type2.AppImage: ELF 64-bit LSB executable, x86-64, version 1 (SYSV) (Lepton 3.x), scale 232-60668, spot sensor temperature -4.187500, color scheme 15, show scale bar, calibration: offset -0.000000, slope 0.000000 (Lepton 2.x), scale 4111-45000, spot sensor temperature 412442.250000, color scheme 3, minimum point enabled, calibration: offset -75402534979642766821519867692934234112.000000, slope 5815371847733706829839455140374904832.000000, dynamically linked, interpreter /lib64/, for GNU/Linux 2.6.18, BuildID[sha1]=79dcc4e55a61c293c5e19edbd8d65b202842579f, stripped\012- data

Note how the type 1 AppImage is described as an ISO 9660 CD-ROM filesystem, and the type 2 AppImage is not.

14.1.2. Wrapping

Depending on the type of AppImage you’re wrapping, you’ll have to use wrapType1 or wrapType2.

appimageTools.wrapType2 { # or wrapType1
  name = "patchwork";
  src = fetchurl {
    url = "";
    sha256 = "1blsprpkvm0ws9b96gb36f0rbf8f5jgmw4x6dsb1kswr4ysf591s";
  extraPkgs = pkgs: with pkgs; [ ];
  • name specifies the name of the resulting image.

  • src specifies the AppImage file to extract.

  • extraPkgs allows you to pass a function to include additional packages inside the FHS environment your AppImage is going to run in. There are a few ways to learn which dependencies an application needs:

    • Looking through the extracted AppImage files, reading its scripts and running patchelf and ldd on its executables. This can also be done in appimage-run, by setting APPIMAGE_DEBUG_EXEC=bash.

    • Running strace -vfefile on the wrapped executable, looking for libraries that can’t be found.

14.2. pkgs.dockerTools

pkgs.dockerTools is a set of functions for creating and manipulating Docker images according to the Docker Image Specification v1.2.0. Docker itself is not used to perform any of the operations done by these functions.

14.2.1. buildImage

This function is analogous to the docker build command, in that it can be used to build a Docker-compatible repository tarball containing a single image with one or multiple layers. As such, the result is suitable for being loaded in Docker with docker load.

The parameters of buildImage with relative example values are described below:

buildImage {
  name = "redis";
  tag = "latest";

  fromImage = someBaseImage;
  fromImageName = null;
  fromImageTag = "latest";

  contents = pkgs.redis;
  runAsRoot = ''
    mkdir -p /data

  config = {
    Cmd = [ "/bin/redis-server" ];
    WorkingDir = "/data";
    Volumes = { "/data" = { }; };

The above example will build a Docker image redis/latest from the given base image. Loading and running this image in Docker results in redis-server being started automatically.

  • name specifies the name of the resulting image. This is the only required argument for buildImage.

  • tag specifies the tag of the resulting image. By default it’s null, which indicates that the nix output hash will be used as tag.

  • fromImage is the repository tarball containing the base image. It must be a valid Docker image, such as exported by docker save. By default it’s null, which can be seen as equivalent to FROM scratch of a Dockerfile.

  • fromImageName can be used to further specify the base image within the repository, in case it contains multiple images. By default it’s null, in which case buildImage will peek the first image available in the repository.

  • fromImageTag can be used to further specify the tag of the base image within the repository, in case an image contains multiple tags. By default it’s null, in which case buildImage will peek the first tag available for the base image.

  • contents is a derivation that will be copied in the new layer of the resulting image. This can be similarly seen as ADD contents/ / in a Dockerfile. By default it’s null.

  • runAsRoot is a bash script that will run as root in an environment that overlays the existing layers of the base image with the new resulting layer, including the previously copied contents derivation. This can be similarly seen as RUN ... in a Dockerfile.

NOTE: Using this parameter requires the kvm device to be available.

  • config is used to specify the configuration of the containers that will be started off the built image in Docker. The available options are listed in the Docker Image Specification v1.2.0.

After the new layer has been created, its closure (to which contents, config and runAsRoot contribute) will be copied in the layer itself. Only new dependencies that are not already in the existing layers will be copied.

At the end of the process, only one new single layer will be produced and added to the resulting image.

The resulting repository will only list the single image image/tag. In the case of the buildImage example it would be redis/latest.

It is possible to inspect the arguments with which an image was built using its buildArgs attribute.

NOTE: If you see errors similar to getProtocolByName: does not exist (no such protocol name: tcp) you may need to add pkgs.iana-etc to contents.

NOTE: If you see errors similar to Error_Protocol ("certificate has unknown CA",True,UnknownCa) you may need to add pkgs.cacert to contents.

By default buildImage will use a static date of one second past the UNIX Epoch. This allows buildImage to produce binary reproducible images. When listing images with docker images, the newly created images will be listed like this:

$ docker images
hello        latest   08c791c7846e   48 years ago   25.2MB

You can break binary reproducibility but have a sorted, meaningful CREATED column by setting created to now.

pkgs.dockerTools.buildImage {
  name = "hello";
  tag = "latest";
  created = "now";
  contents = pkgs.hello;

  config.Cmd = [ "/bin/hello" ];

and now the Docker CLI will display a reasonable date and sort the images as expected:

$ docker images
REPOSITORY   TAG      IMAGE ID       CREATED              SIZE
hello        latest   de2bf4786de6   About a minute ago   25.2MB

however, the produced images will not be binary reproducible.

14.2.2. buildLayeredImage

Create a Docker image with many of the store paths being on their own layer to improve sharing between images. The image is realized into the Nix store as a gzipped tarball. Depending on the intended usage, many users might prefer to use streamLayeredImage instead, which this function uses internally.


The name of the resulting image.

tag optional

Tag of the generated image.

Default: the output path’s hash

fromImage optional

The repository tarball containing the base image. It must be a valid Docker image, such as one exported by docker save.

Default: null, which can be seen as equivalent to FROM scratch of a Dockerfile.

contents optional

Top level paths in the container. Either a single derivation, or a list of derivations.

Default: []

config optional

Run-time configuration of the container. A full list of the options are available at in the Docker Image Specification v1.2.0 .

Default: {}

created optional

Date and time the layers were created. Follows the same now exception supported by buildImage.

Default: 1970-01-01T00:00:01Z

maxLayers optional

Maximum number of layers to create.

Default: 100

Maximum: 125

extraCommands optional

Shell commands to run while building the final layer, without access to most of the layer contents. Changes to this layer are on top of all the other layers, so can create additional directories and files.

fakeRootCommands optional

Shell commands to run while creating the archive for the final layer in a fakeroot environment. Unlike extraCommands, you can run chown to change the owners of the files in the archive, changing fakeroot’s state instead of the real filesystem. The latter would require privileges that the build user does not have. Static binaries do not interact with the fakeroot environment. By default all files in the archive will be owned by root. Behavior of contents in the final image

Each path directly listed in contents will have a symlink in the root of the image.

For example:

pkgs.dockerTools.buildLayeredImage {
  name = "hello";
  contents = [ pkgs.hello ];

will create symlinks for all the paths in the hello package:

/bin/hello -> /nix/store/h1zb1padqbbb7jicsvkmrym3r6snphxg-hello-2.10/bin/hello
/share/info/ -> /nix/store/h1zb1padqbbb7jicsvkmrym3r6snphxg-hello-2.10/share/info/
/share/locale/bg/LC_MESSAGES/ -> /nix/store/h1zb1padqbbb7jicsvkmrym3r6snphxg-hello-2.10/share/locale/bg/LC_MESSAGES/ Automatic inclusion of config references

The closure of config is automatically included in the closure of the final image.

This allows you to make very simple Docker images with very little code. This container will start up and run hello:

pkgs.dockerTools.buildLayeredImage {
  name = "hello";
  config.Cmd = [ "${pkgs.hello}/bin/hello" ];
} Adjusting maxLayers

Increasing the maxLayers increases the number of layers which have a chance to be shared between different images.

Modern Docker installations support up to 128 layers, however older versions support as few as 42.

If the produced image will not be extended by other Docker builds, it is safe to set maxLayers to 128. However it will be impossible to extend the image further.

The first (maxLayers-2) most popular paths will have their own individual layers, then layer #maxLayers-1 will contain all the remaining unpopular paths, and finally layer #maxLayers will contain the Image configuration.

Docker’s Layers are not inherently ordered, they are content-addressable and are not explicitly layered until they are composed in to an Image.

14.2.3. streamLayeredImage

Builds a script which, when run, will stream an uncompressed tarball of a Docker image to stdout. The arguments to this function are as for buildLayeredImage. This method of constructing an image does not realize the image into the Nix store, so it saves on IO and disk/cache space, particularly with large images.

The image produced by running the output script can be piped directly into docker load, to load it into the local docker daemon:

$(nix-build) | docker load

Alternatively, the image be piped via gzip into skopeo, e.g. to copy it into a registry:

$(nix-build) | gzip --fast | skopeo copy docker-archive:/dev/stdin docker://some_docker_registry/myimage:tag

14.2.4. pullImage

This function is analogous to the docker pull command, in that it can be used to pull a Docker image from a Docker registry. By default Docker Hub is used to pull images.

Its parameters are described in the example below:

pullImage {
  imageName = "nixos/nix";
  imageDigest =
  finalImageName = "nix";
  finalImageTag = "1.11";
  sha256 = "0mqjy3zq2v6rrhizgb9nvhczl87lcfphq9601wcprdika2jz7qh8";
  os = "linux";
  arch = "x86_64";
  • imageName specifies the name of the image to be downloaded, which can also include the registry namespace (e.g. nixos). This argument is required.

  • imageDigest specifies the digest of the image to be downloaded. This argument is required.

  • finalImageName, if specified, this is the name of the image to be created. Note it is never used to fetch the image since we prefer to rely on the immutable digest ID. By default it’s equal to imageName.

  • finalImageTag, if specified, this is the tag of the image to be created. Note it is never used to fetch the image since we prefer to rely on the immutable digest ID. By default it’s latest.

  • sha256 is the checksum of the whole fetched image. This argument is required.

  • os, if specified, is the operating system of the fetched image. By default it’s linux.

  • arch, if specified, is the cpu architecture of the fetched image. By default it’s x86_64.

nix-prefetch-docker command can be used to get required image parameters:

$ nix run nixpkgs.nix-prefetch-docker -c nix-prefetch-docker --image-name mysql --image-tag 5

Since a given imageName may transparently refer to a manifest list of images which support multiple architectures and/or operating systems, you can supply the --os and --arch arguments to specify exactly which image you want. By default it will match the OS and architecture of the host the command is run on.

$ nix-prefetch-docker --image-name mysql --image-tag 5 --arch x86_64 --os linux

Desired image name and tag can be set using --final-image-name and --final-image-tag arguments:

$ nix-prefetch-docker --image-name mysql --image-tag 5 --final-image-name --final-image-tag prod

14.2.5. exportImage

This function is analogous to the docker export command, in that it can be used to flatten a Docker image that contains multiple layers. It is in fact the result of the merge of all the layers of the image. As such, the result is suitable for being imported in Docker with docker import.

NOTE: Using this function requires the kvm device to be available.

The parameters of exportImage are the following:

exportImage {
  fromImage = someLayeredImage;
  fromImageName = null;
  fromImageTag = null;

  name =;

The parameters relative to the base image have the same synopsis as described in buildImage, except that fromImage is the only required argument in this case.

The name argument is the name of the derivation output, which defaults to

14.2.6. shadowSetup

This constant string is a helper for setting up the base files for managing users and groups, only if such files don’t exist already. It is suitable for being used in a buildImage runAsRoot script for cases like in the example below:

buildImage {
  name = "shadow-basic";

  runAsRoot = ''
    groupadd -r redis
    useradd -r -g redis redis
    mkdir /data
    chown redis:redis /data

Creating base files like /etc/passwd or /etc/login.defs is necessary for shadow-utils to manipulate users and groups.

14.3. pkgs.ociTools

pkgs.ociTools is a set of functions for creating containers according to the OCI container specification v1.0.0. Beyond that it makes no assumptions about the container runner you choose to use to run the created container.

14.3.1. buildContainer

This function creates a simple OCI container that runs a single command inside of it. An OCI container consists of a config.json and a rootfs directory.The nix store of the container will contain all referenced dependencies of the given command.

The parameters of buildContainer with an example value are described below:

buildContainer {
  args = [
    (with pkgs;
      writeScript "" ''
        exec ${bash}/bin/bash

  mounts = {
    "/data" = {
      type = "none";
      source = "/var/lib/mydata";
      options = [ "bind" ];

  readonly = false;
  • args specifies a set of arguments to run inside the container. This is the only required argument for buildContainer. All referenced packages inside the derivation will be made available inside the container

  • mounts specifies additional mount points chosen by the user. By default only a minimal set of necessary filesystems are mounted into the container (e.g procfs, cgroupfs)

  • readonly makes the container's rootfs read-only if it is set to true. The default value is false false.

14.4. pkgs.snapTools

pkgs.snapTools is a set of functions for creating Snapcraft images. Snap and Snapcraft is not used to perform these operations.

14.4.1. The makeSnap Function

makeSnap takes a single named argument, meta. This argument mirrors the upstream snap.yaml format exactly.

The base should not be specified, as makeSnap will force set it.

Currently, makeSnap does not support creating GUI stubs.

14.4.2. Build a Hello World Snap

The following expression packages GNU Hello as a Snapcraft snap.

  inherit (import <nixpkgs> { }) snapTools hello;
in snapTools.makeSnap {
  meta = {
    name = "hello";
    summary = hello.meta.description;
    description = hello.meta.longDescription;
    architectures = [ "amd64" ];
    confinement = "strict";
    apps.hello.command = "${hello}/bin/hello";

nix-build this expression and install it with snap install ./result --dangerous. hello will now be the Snapcraft version of the package.

14.4.3. Build a Graphical Snap

Graphical programs require many more integrations with the host. This example uses Firefox as an example, because it is one of the most complicated programs we could package.

  inherit (import <nixpkgs> { }) snapTools firefox;
in snapTools.makeSnap {
  meta = {
    name = "nix-example-firefox";
    summary = firefox.meta.description;
    architectures = [ "amd64" ];
    apps.nix-example-firefox = {
      command = "${firefox}/bin/firefox";
      plugs = [
    confinement = "strict";

nix-build this expression and install it with snap install ./result --dangerous. nix-example-firefox will now be the Snapcraft version of the Firefox package.

The specific meaning behind plugs can be looked up in the Snapcraft interface documentation.

Chapter 15. Languages and frameworks

The standard build environment makes it easy to build typical Autotools-based packages with very little code. Any other kind of package can be accomodated by overriding the appropriate phases of stdenv. However, there are specialised functions in Nixpkgs to easily build packages for other programming languages, such as Perl or Haskell. These are described in this chapter.

15.1. Agda

15.1.1. How to use Agda

Agda is available as the agda package.

The agda package installs an Agda-wrapper, which calls agda with --library-file set to a generated library-file within the nix store, this means your library-file in $HOME/.agda/libraries will be ignored. By default the agda package installs Agda with no libraries, i.e. the generated library-file is empty. To use Agda with libraries, the agda.withPackages function can be used. This function either takes:

  • A list of packages,

  • or a function which returns a list of packages when given the agdaPackages attribute set,

  • or an attribute set containing a list of packages and a GHC derivation for compilation (see below).

  • or an attribute set containing a function which returns a list of packages when given the agdaPackages attribute set and a GHC derivation for compilation (see below).

For example, suppose we wanted a version of Agda which has access to the standard library. This can be obtained with the expressions:

agda.withPackages [ agdaPackages.standard-library ]


agda.withPackages (p: [ p.standard-library ])

or can be called as in the Compiling Agda section.

If you want to use a different version of a library (for instance a development version) override the src attribute of the package to point to your local repository

agda.withPackages (p: [
  (p.standard-library.overrideAttrs (oldAttrs: {
    version = "local version";
    src = /path/to/local/repo/agda-stdlib;

You can also reference a GitHub repository

agda.withPackages (p: [
  (p.standard-library.overrideAttrs (oldAttrs: {
    version = "1.5";
    src =  fetchFromGitHub {
      repo = "agda-stdlib";
      owner = "agda";
      rev = "v1.5";
      sha256 = "16fcb7ssj6kj687a042afaa2gq48rc8abihpm14k684ncihb2k4w";

If you want to use a library not added to Nixpkgs, you can add a dependency to a local library by calling agdaPackages.mkDerivation.

agda.withPackages (p: [
  (p.mkDerivation {
    pname = "your-agda-lib";
    version = "1.0.0";
    src = /path/to/your-agda-lib;

Again you can reference GitHub

agda.withPackages (p: [
  (p.mkDerivation {
    pname = "your-agda-lib";
    version = "1.0.0";
    src = fetchFromGitHub {
      repo = "repo";
      owner = "owner";
      version = "...";
      rev = "...";
      sha256 = "...";

See Building Agda Packages for more information on mkDerivation.

Agda will not by default use these libraries. To tell Agda to use a library we have some options:

  • Call agda with the library flag:

    $ agda -l standard-library -i . MyFile.agda
  • Write a my-library.agda-lib file for the project you are working on which may look like:

    name: my-library
    include: .
    depend: standard-library
  • Create the file ~/.agda/defaults and add any libraries you want to use by default.

More information can be found in the official Agda documentation on library management.

15.1.2. Compiling Agda

Agda modules can be compiled using the GHC backend with the --compile flag. A version of ghc with ieee754 is made available to the Agda program via the --with-compiler flag. This can be overridden by a different version of ghc as follows:

agda.withPackages {
  pkgs = [ ... ];
  ghc = haskell.compiler.ghcHEAD;

15.1.3. Writing Agda packages

To write a nix derivation for an Agda library, first check that the library has a *.agda-lib file.

A derivation can then be written using agdaPackages.mkDerivation. This has similar arguments to stdenv.mkDerivation with the following additions:

  • everythingFile can be used to specify the location of the Everything.agda file, defaulting to ./Everything.agda. If this file does not exist then either it should be patched in or the buildPhase should be overridden (see below).

  • libraryName should be the name that appears in the *.agda-lib file, defaulting to pname.

  • libraryFile should be the file name of the *.agda-lib file, defaulting to ${libraryName}.agda-lib.

Here is an example default.nix

{ nixpkgs ?  <nixpkgs> }:
with (import nixpkgs {});
agdaPackages.mkDerivation {
  version = "1.0";
  pname = "my-agda-lib";
  src = ./.;
  buildInputs = [
} Building Agda packages

The default build phase for agdaPackages.mkDerivation simply runs agda on the Everything.agda file. If something else is needed to build the package (e.g. make) then the buildPhase should be overridden. Additionally, a preBuild or configurePhase can be used if there are steps that need to be done prior to checking the Everything.agda file. agda and the Agda libraries contained in buildInputs are made available during the build phase. Installing Agda packages

The default install phase copies Agda source files, Agda interface files (*.agdai) and *.agda-lib files to the output directory. This can be overridden.

By default, Agda sources are files ending on .agda, or literate Agda files ending on .lagda, .lagda.tex,,, .lagda.rst. The list of recognised Agda source extensions can be extended by setting the extraExtensions config variable.

15.1.4. Maintaining the Agda package set on Nixpkgs

We are aiming at providing all common Agda libraries as packages on nixpkgs, and keeping them up to date. Contributions and maintenance help is always appreciated, but the maintenance effort is typically low since the Agda ecosystem is quite small.

The nixpkgs Agda package set tries to take up a role similar to that of Stackage in the Haskell world. It is a curated set of libraries that:

  1. Always work together.

  2. Are as up-to-date as possible.

While the Haskell ecosystem is huge, and Stackage is highly automatised, the Agda package set is small and can (still) be maintained by hand. Adding Agda packages to Nixpkgs

To add an Agda package to nixpkgs, the derivation should be written to pkgs/development/libraries/agda/${library-name}/ and an entry should be added to pkgs/top-level/agda-packages.nix. Here it is called in a scope with access to all other Agda libraries, so the top line of the default.nix can look like:

{ mkDerivation, standard-library, fetchFromGitHub }:

Note that the derivation function is called with mkDerivation set to agdaPackages.mkDerivation, therefore you could use a similar set as in your default.nix from Writing Agda Packages with agdaPackages.mkDerivation replaced with mkDerivation.

Here is an example skeleton derivation for iowa-stdlib:

mkDerivation {
  version = "1.5.0";
  pname = "iowa-stdlib";

  src = ...

  libraryFile = "";
  libraryName = "IAL-1.3";

  buildPhase = ''

This library has a file called .agda-lib, and so we give an empty string to libraryFile as nothing precedes .agda-lib in the filename. This file contains name: IAL-1.3, and so we let libraryName = "IAL-1.3". This library does not use an Everything.agda file and instead has a Makefile, so there is no need to set everythingFile and we set a custom buildPhase.

When writing an Agda package it is essential to make sure that no .agda-lib file gets added to the store as a single file (for example by using writeText). This causes Agda to think that the nix store is a Agda library and it will attempt to write to it whenever it typechecks something. See

In the pull request adding this library, you can test whether it builds correctly by writing in a comment:

@ofborg build agdaPackages.iowa-stdlib Maintaining Agda packages

As mentioned before, the aim is to have a compatible, and up-to-date package set. These two conditions sometimes exclude each other: For example, if we update agdaPackages.standard-library because there was an upstream release, this will typically break many reverse dependencies, i.e. downstream Agda libraries that depend on the standard library. In nixpkgs we are typically among the first to notice this, since we have build tests in place to check this.

In a pull request updating e.g. the standard library, you should write the following comment:

@ofborg build agdaPackages.standard-library.passthru.tests

This will build all reverse dependencies of the standard library, for example agdaPackages.agda-categories, or agdaPackages.generic.

In some cases it is useful to build all Agda packages. This can be done with the following Github comment:

@ofborg build agda.passthru.tests.allPackages

Sometimes, the builds of the reverse dependencies fail because they have not yet been updated and released. You should drop the maintainers a quick issue notifying them of the breakage, citing the build error (which you can get from the ofborg logs). If you are motivated, you might even send a pull request that fixes it. Usually, the maintainers will answer within a week or two with a new release. Bumping the version of that reverse dependency should be a further commit on your PR.

In the rare case that a new release is not to be expected within an acceptable time, simply mark the broken package as broken by setting meta.broken = true;. This will exclude it from the build test. It can be added later when it is fixed, and does not hinder the advancement of the whole package set in the meantime.

15.2. Android

The Android build environment provides three major features and a number of supporting features.

15.2.1. Deploying an Android SDK installation with plugins

The first use case is deploying the SDK with a desired set of plugins or subsets of an SDK.

with import <nixpkgs> {};

  androidComposition = androidenv.composeAndroidPackages {
    toolsVersion = "26.1.1";
    platformToolsVersion = "30.0.5";
    buildToolsVersions = [ "30.0.3" ];
    includeEmulator = false;
    emulatorVersion = "30.3.4";
    platformVersions = [ "28" "29" "30" ];
    includeSources = false;
    includeSystemImages = false;
    systemImageTypes = [ "google_apis_playstore" ];
    abiVersions = [ "armeabi-v7a" "arm64-v8a" ];
    cmakeVersions = [ "3.10.2" ];
    includeNDK = true;
    ndkVersions = ["22.0.7026061"];
    useGoogleAPIs = false;
    useGoogleTVAddOns = false;
    includeExtras = [

The above function invocation states that we want an Android SDK with the above specified plugin versions. By default, most plugins are disabled. Notable exceptions are the tools, platform-tools and build-tools sub packages.

The following parameters are supported:

  • toolsVersion, specifies the version of the tools package to use

  • platformsToolsVersion specifies the version of the platform-tools plugin

  • buildToolsVersions specifies the versions of the build-tools plugins to use.

  • includeEmulator specifies whether to deploy the emulator package (false by default). When enabled, the version of the emulator to deploy can be specified by setting the emulatorVersion parameter.

  • cmakeVersions specifies which CMake versions should be deployed.

  • includeNDK specifies that the Android NDK bundle should be included. Defaults to: false.

  • ndkVersions specifies the NDK versions that we want to use. These are linked under the ndk directory of the SDK root, and the first is linked under the ndk-bundle directory.

  • ndkVersion is equivalent to specifying one entry in ndkVersions, and ndkVersions overrides this parameter if provided.

  • includeExtras is an array of identifier strings referring to arbitrary add-on packages that should be installed.

  • platformVersions specifies which platform SDK versions should be included.

For each platform version that has been specified, we can apply the following options:

  • includeSystemImages specifies whether a system image for each platform SDK should be included.

  • includeSources specifies whether the sources for each SDK version should be included.

  • useGoogleAPIs specifies that for each selected platform version the Google API should be included.

  • useGoogleTVAddOns specifies that for each selected platform version the Google TV add-on should be included.

For each requested system image we can specify the following options:

  • systemImageTypes specifies what kind of system images should be included. Defaults to: default.

  • abiVersions specifies what kind of ABI version of each system image should be included. Defaults to: armeabi-v7a.

Most of the function arguments have reasonable default settings.

You can specify license names:

  • extraLicenses is a list of license names. You can get these names from repo.json or licenses. The SDK license (android-sdk-license) is accepted for you if you set accept_license to true. If you are doing something like working with preview SDKs, you will want to add android-sdk-preview-license or whichever license applies here.

Additionally, you can override the repositories that composeAndroidPackages will pull from:

  • repoJson specifies a path to a generated repo.json file. You can generate this by running, which in turn will call into mkrepo.rb.

  • repoXmls is an attribute set containing paths to repo XML files. If specified, it takes priority over repoJson, and will trigger a local build writing out a repo.json to the Nix store based on the given repository XMLs.

repoXmls = {
  packages = [ ./xml/repository2-1.xml ];
  images = [
  addons = [ ./xml/addon2-1.xml ];

When building the above expression with:

$ nix-build

The Android SDK gets deployed with all desired plugin versions.

We can also deploy subsets of the Android SDK. For example, to only the platform-tools package, you can evaluate the following expression:

with import <nixpkgs> {};

  androidComposition = androidenv.composeAndroidPackages {
    # ...

15.2.2. Using predefined Android package compositions

In addition to composing an Android package set manually, it is also possible to use a predefined composition that contains all basic packages for a specific Android version, such as version 9.0 (API-level 28).

The following Nix expression can be used to deploy the entire SDK with all basic plugins:

with import <nixpkgs> {};


It is also possible to use one plugin only:

with import <nixpkgs> {};


15.2.3. Building an Android application

In addition to the SDK, it is also possible to build an Ant-based Android project and automatically deploy all the Android plugins that a project requires.

with import <nixpkgs> {};

androidenv.buildApp {
  name = "MyAndroidApp";
  src = ./myappsources;
  release = true;

  # If release is set to true, you need to specify the following parameters
  keyStore = ./keystore;
  keyAlias = "myfirstapp";
  keyStorePassword = "mykeystore";
  keyAliasPassword = "myfirstapp";

  # Any Android SDK parameters that install all the relevant plugins that a
  # build requires
  platformVersions = [ "24" ];

  # When we include the NDK, then ndk-build is invoked before Ant gets invoked
  includeNDK = true;

Aside from the app-specific build parameters (name, src, release and keystore parameters), the buildApp {} function supports all the function parameters that the SDK composition function (the function shown in the previous section) supports.

This build function is particularly useful when it is desired to use Hydra: the Nix-based continuous integration solution to build Android apps. An Android APK gets exposed as a build product and can be installed on any Android device with a web browser by navigating to the build result page.

15.2.4. Spawning emulator instances

For testing purposes, it can also be quite convenient to automatically generate scripts that spawn emulator instances with all desired configuration settings.

An emulator spawn script can be configured by invoking the emulateApp {} function:

with import <nixpkgs> {};

androidenv.emulateApp {
  name = "emulate-MyAndroidApp";
  platformVersion = "28";
  abiVersion = "x86"; # armeabi-v7a, mips, x86_64
  systemImageType = "google_apis_playstore";

Additional flags may be applied to the Android SDK’s emulator through the runtime environment variable $NIX_ANDROID_EMULATOR_FLAGS.

It is also possible to specify an APK to deploy inside the emulator and the package and activity names to launch it:

with import <nixpkgs> {};

androidenv.emulateApp {
  name = "emulate-MyAndroidApp";
  platformVersion = "24";
  abiVersion = "armeabi-v7a"; # mips, x86, x86_64
  systemImageType = "default";
  useGoogleAPIs = false;
  app = ./MyApp.apk;
  package = "MyApp";
  activity = "MainActivity";

In addition to prebuilt APKs, you can also bind the APK parameter to a buildApp {} function invocation shown in the previous example.

15.2.5. Notes on environment variables in Android projects

  • ANDROID_SDK_ROOT should point to the Android SDK. In your Nix expressions, this should be ${androidComposition.androidsdk}/libexec/android-sdk. Note that ANDROID_HOME is deprecated, but if you rely on tools that need it, you can export it too.

  • ANDROID_NDK_ROOT should point to the Android NDK, if you’re doing NDK development. In your Nix expressions, this should be ${ANDROID_SDK_ROOT}/ndk-bundle.

If you are running the Android Gradle plugin, you need to export GRADLE_OPTS to override aapt2 to point to the aapt2 binary in the Nix store as well, or use a FHS environment so the packaged aapt2 can run. If you don’t want to use a FHS environment, something like this should work:

  buildToolsVersion = "30.0.3";

  # Use buildToolsVersion when you define androidComposition
  androidComposition = <...>;
pkgs.mkShell rec {
  ANDROID_SDK_ROOT = "${androidComposition.androidsdk}/libexec/android-sdk";

  # Use the same buildToolsVersion here
  GRADLE_OPTS = "${ANDROID_SDK_ROOT}/build-tools/${buildToolsVersion}/aapt2";

If you are using cmake, you need to add it to PATH in a shell hook or FHS env profile. The path is suffixed with a build number, but properly prefixed with the version. So, something like this should suffice:

  cmakeVersion = "3.10.2";

  # Use cmakeVersion when you define androidComposition
  androidComposition = <...>;
pkgs.mkShell rec {
  ANDROID_SDK_ROOT = "${androidComposition.androidsdk}/libexec/android-sdk";

  # Use the same cmakeVersion here
  shellHook = ''
    export PATH="$(echo "$ANDROID_SDK_ROOT/cmake/${cmakeVersion}".*/bin):$PATH"

Note that running Android Studio with ANDROID_SDK_ROOT set will automatically write a file with sdk.dir set to $ANDROID_SDK_ROOT if one does not already exist. If you are using the NDK as well, you may have to add ndk.dir to this file.

An example shell.nix that does all this for you is provided in examples/shell.nix. This shell.nix includes a shell hook that overwrites with the correct sdk.dir and ndk.dir values. This will ensure that the SDK and NDK directories will both be correct when you run Android Studio inside nix-shell.

15.2.6. Notes on improving build.gradle compatibility

Ensure that your buildToolsVersion and ndkVersion match what is declared in androidenv. If you are using cmake, make sure its declared version is correct too.

Otherwise, you may get cryptic errors from aapt2 and the Android Gradle plugin warning that it cannot install the build tools because the SDK directory is not writeable.

android {
    buildToolsVersion "30.0.3"
    ndkVersion = "22.0.7026061"
    externalNativeBuild {
        cmake {
            version "3.10.2"

15.2.7. Querying the available versions of each plugin

repo.json provides all the options in one file now.

A shell script in the pkgs/development/mobile/androidenv/ subdirectory can be used to retrieve all possible options:

./ packages

The above command-line instruction queries all package versions in repo.json.

15.2.8. Updating the generated expressions

repo.json is generated from XML files that the Android Studio package manager uses. To update the expressions run the script that is stored in the pkgs/development/mobile/androidenv/ subdirectory:


15.3. BEAM Languages (Erlang, Elixir & LFE)

15.3.1. Introduction

In this document and related Nix expressions, we use the term, BEAM, to describe the environment. BEAM is the name of the Erlang Virtual Machine and, as far as we’re concerned, from a packaging perspective, all languages that run on the BEAM are interchangeable. That which varies, like the build system, is transparent to users of any given BEAM package, so we make no distinction.

15.3.2. Available versions and deprecations schedule Elixir

nixpkgs follows the official elixir deprecation schedule and keeps the last 5 released versions of Elixir available.

15.3.3. Structure

All BEAM-related expressions are available via the top-level beam attribute, which includes:

  • interpreters: a set of compilers running on the BEAM, including multiple Erlang/OTP versions (beam.interpreters.erlangR22, etc), Elixir (beam.interpreters.elixir) and LFE (Lisp Flavoured Erlang) (beam.interpreters.lfe).

  • packages: a set of package builders (Mix and rebar3), each compiled with a specific Erlang/OTP version, e.g. beam.packages.erlang22.

The default Erlang compiler, defined by beam.interpreters.erlang, is aliased as erlang. The default BEAM package set is defined by beam.packages.erlang and aliased at the top level as beamPackages.

To create a package builder built with a custom Erlang version, use the lambda, beam.packagesWith, which accepts an Erlang/OTP derivation and produces a package builder similar to beam.packages.erlang.

Many Erlang/OTP distributions available in beam.interpreters have versions with ODBC and/or Java enabled or without wx (no observer support). For example, there’s beam.interpreters.erlangR22_odbc_javac, which corresponds to beam.interpreters.erlangR22 and beam.interpreters.erlangR22_nox, which corresponds to beam.interpreters.erlangR22.

15.3.4. Build Tools Rebar3

We provide a version of Rebar3, under rebar3. We also provide a helper to fetch Rebar3 dependencies from a lockfile under fetchRebar3Deps.

We also provide a version on Rebar3 with plugins included, under rebar3WithPlugins. This package is a function which takes two arguments: plugins, a list of nix derivations to include as plugins (loaded only when specified in rebar.config), and globalPlugins, which should always be loaded by rebar3. Example: rebar3WithPlugins { globalPlugins = [beamPackages.pc]; }.

When adding a new plugin it is important that the packageName attribute is the same as the atom used by rebar3 to refer to the plugin. Mix & works exactly as expected. There is a bootstrap process that needs to be run, which is supported by the buildErlangMk derivation.

For Elixir applications use mixRelease to make a release. See examples for more details.

There is also a buildMix helper, whose behavior is closer to that of buildErlangMk and buildRebar3. The primary difference is that mixRelease makes a release, while buildMix only builds the package, making it useful for libraries and other dependencies.

15.3.5. How to Install BEAM Packages

BEAM builders are not registered at the top level, simply because they are not relevant to the vast majority of Nix users. To install any of those builders into your profile, refer to them by their attribute path beamPackages.rebar3:

$ nix-env -f "<nixpkgs>" -iA beamPackages.rebar3

15.3.6. Packaging BEAM Applications Erlang Applications Rebar3 Packages

The Nix function, buildRebar3, defined in beam.packages.erlang.buildRebar3 and aliased at the top level, can be used to build a derivation that understands how to build a Rebar3 project.

If a package needs to compile native code via Rebar3’s port compilation mechanism, add compilePort = true; to the derivation. Packages functions similarly to Rebar3, except we use buildErlangMk instead of buildRebar3. Mix Packages

mixRelease is used to make a release in the mix sense. Dependencies will need to be fetched with fetchMixDeps and passed to it. mixRelease - Elixir Phoenix example

there are 3 steps, frontend dependencies (javascript), backend dependencies (elixir) and the final derivation that puts both of those together mixRelease - Frontend dependencies (javascript)

for phoenix projects, inside of nixpkgs you can either use yarn2nix (mkYarnModule) or node2nix. An example with yarn2nix can be found here. An example with node2nix will follow. To package something outside of nixpkgs, you have alternatives like npmlock2nix or nix-npm-buildpackage mixRelease - backend dependencies (mix)

There are 2 ways to package backend dependencies. With mix2nix and with a fixed-output-derivation (FOD).


mix2nix is a cli tool available in nixpkgs. it will generate a nix expression from a mix.lock file. It is quite standard in the 2nix tool series.

Note that currently mix2nix can’t handle git dependencies inside the mix.lock file. If you have git dependencies, you can either add them manually (see example) or use the FOD method.

The advantage of using mix2nix is that nix will know your whole dependency graph. On a dependency update, this won’t trigger a full rebuild and download of all the dependencies, where FOD will do so.

practical steps:

  • run mix2nix > mix_deps.nix in the upstream repo.

  • pass mixNixDeps = with pkgs; import ./mix_deps.nix { inherit lib beamPackages; }; as an argument to mixRelease.

If there are git depencencies.

  • You’ll need to fix the version artificially in mix.exs and regenerate the mix.lock with fixed version (on upstream). This will enable you to run mix2nix > mix_deps.nix.

  • From the mix_deps.nix file, remove the dependencies that had git versions and pass them as an override to the import function.

  mixNixDeps = import ./mix.nix {
    inherit beamPackages lib;
    overrides = (final: prev: {
      # mix2nix does not support git dependencies yet,
      # so we need to add them manually
      prometheus_ex = beamPackages.buildMix rec {
        name = "prometheus_ex";
        version = "3.0.5";

        # Change the argument src with the git src that you actually need
        src = fetchFromGitLab {
          domain = "";
          group = "pleroma";
          owner = "elixir-libraries";
          repo = "prometheus.ex";
          rev = "a4e9beb3c1c479d14b352fd9d6dd7b1f6d7deee5";
          sha256 = "1v0q4bi7sb253i8q016l7gwlv5562wk5zy3l2sa446csvsacnpjk";
        # you can re-use the same beamDeps argument as generated
        beamDeps = with final; [ prometheus ];

You will need to run the build process once to fix the sha256 to correspond to your new git src.


A fixed output derivation will download mix dependencies from the internet. To ensure reproducibility, a hash will be supplied. Note that mix is relatively reproducible. An FOD generating a different hash on each run hasn’t been observed (as opposed to npm where the chances are relatively high). See elixir_ls for a usage example of FOD.

Practical steps

  • start with the following argument to mixRelease

  mixFodDeps = fetchMixDeps {
    pname = "mix-deps-${pname}";
    inherit src version;
    sha256 = lib.fakeSha256;

The first build will complain about the sha256 value, you can replace with the suggested value after that.

Note that if after you’ve replaced the value, nix suggests another sha256, then mix is not fetching the dependencies reproducibly. An FOD will not work in that case and you will have to use mix2nix. mixRelease - example

Here is how your default.nix file would look for a phoenix project.

with import <nixpkgs> { };

  # beam.interpreters.erlangR23 is available if you need a particular version
  packages = beam.packagesWith beam.interpreters.erlang;

  pname = "your_project";
  version = "0.0.1";

  src = builtins.fetchgit {
    url = "ssh://";
    rev = "replace_with_your_commit";

  # if using mix2nix you can use the mixNixDeps attribute
  mixFodDeps = packages.fetchMixDeps {
    pname = "mix-deps-${pname}";
    inherit src version;
    # nix will complain and tell you the right value to replace this with
    sha256 = lib.fakeSha256;
    # if you have build time environment variables add them here

  nodeDependencies = (pkgs.callPackage ./assets/default.nix { }).shell.nodeDependencies;

in packages.mixRelease {
  inherit src pname version mixFodDeps;
  # if you have build time environment variables add them here

  postBuild = ''
    ln -sf ${nodeDependencies}/lib/node_modules assets/node_modules
    npm run deploy --prefix ./assets

    # for external task you need a workaround for the no deps check flag
    mix do deps.loadpaths --no-deps-check, phx.digest
    mix phx.digest --no-deps-check

Setup will require the following steps:

  • Move your secrets to runtime environment variables. For more information refer to the runtime.exs docs. On a fresh Phoenix build that would mean that both DATABASE_URL and SECRET_KEY need to be moved to runtime.exs.

  • cd assets and nix-shell -p node2nix --run node2nix --development will generate a Nix expression containing your frontend dependencies

  • commit and push those changes

  • you can now nix-build .

  • To run the release, set the RELEASE_TMP environment variable to a directory that your program has write access to. It will be used to store the BEAM settings. Example of creating a service for an Elixir - Phoenix project

In order to create a service with your release, you could add a service.nix in your project with the following

{config, pkgs, lib, ...}:

  release = pkgs.callPackage ./default.nix;
  release_name = "app";
  working_directory = "/home/app";
{${release_name} = {
    wantedBy = [ "" ];
    after = [ "" "postgresql.service" ];
    # note that if you are connecting to a postgres instance on a different host
    # postgresql.service should not be included in the requires.
    requires = [ "" "postgresql.service" ];
    description = "my app";
    environment = {
      # RELEASE_TMP is used to write the state of the
      # VM configuration when the system is running
      # it needs to be a writable directory
      RELEASE_TMP = working_directory;
      # can be generated in an elixir console with
      # Base.encode32(:crypto.strong_rand_bytes(32))
      RELEASE_COOKIE = "my_cookie";
      MY_VAR = "my_var";
    serviceConfig = {
      Type = "exec";
      DynamicUser = true;
      WorkingDirectory = working_directory;
      # Implied by DynamicUser, but just to emphasize due to RELEASE_TMP
      PrivateTmp = true;
      ExecStart = ''
        ${release}/bin/${release_name} start
      ExecStop = ''
        ${release}/bin/${release_name} stop
      ExecReload = ''
        ${release}/bin/${release_name} restart
      Restart = "on-failure";
      RestartSec = 5;
      StartLimitBurst = 3;
      StartLimitInterval = 10;
    # disksup requires bash
    path = [ pkgs.bash ];

  # in case you have migration scripts or you want to use a remote shell
  environment.systemPackages = [ release ];

15.3.7. How to Develop Creating a Shell

Usually, we need to create a shell.nix file and do our development inside of the environment specified therein. Just install your version of Erlang and any other interpreters, and then use your normal build tools. As an example with Elixir:

{ pkgs ? import <nixpkgs> {} }:

with pkgs;
  elixir = beam.packages.erlangR24.elixir_1_12;
mkShell {
  buildInputs = [ elixir ];
} Elixir - Phoenix project

Here is an example shell.nix.

with import <nixpkgs> { };

  # define packages to install
  basePackages = [
    # replace with beam.packages.erlang.elixir_1_11 if you need
    # only used for frontend dependencies
    # you are free to use yarn2nix as well
    # formatting js file

  inputs = basePackages ++ lib.optionals stdenv.isLinux [ inotify-tools ]
    ++ lib.optionals stdenv.isDarwin
    (with darwin.apple_sdk.frameworks; [ CoreFoundation CoreServices ]);

  # define shell startup command
  hooks = ''
    # this allows mix to work on the local directory
    mkdir -p .nix-mix .nix-hex
    export MIX_HOME=$PWD/.nix-mix
    export HEX_HOME=$PWD/.nix-mix
    export PATH=$MIX_HOME/bin:$HEX_HOME/bin:$PATH
    # TODO: not sure how to make hex available without installing it afterwards.
    mix local.hex --if-missing
    export LANG=en_US.UTF-8
    # keep your shell history in iex
    export ERL_AFLAGS="-kernel shell_history enabled"

    # postges related
    # keep all your db data in a folder inside the project
    export PGDATA="$PWD/db"

    # phoenix related env vars
    export POOL_SIZE=15
    export DB_URL="postgresql://postgres:postgres@localhost:5432/db"
    export PORT=4000
    export MIX_ENV=dev
    # add your project env vars here, word readable in the nix store.
    export ENV_VAR="your_env_var"

in mkShell {
  buildInputs = inputs;
  shellHook = hooks;

Initializing the project will require the following steps:

  • create the db directory initdb ./db (inside your mix project folder)

  • create the postgres user createuser postgres -ds

  • create the db createdb db

  • start the postgres instance pg_ctl -l "$PGDATA/server.log" start

  • add the /db folder to your .gitignore

  • you can start your phoenix server and get a shell with iex -S mix phx.server

15.4. Bower

Bower is a package manager for web site front-end components. Bower packages (comprising of build artefacts and sometimes sources) are stored in git repositories, typically on Github. The package registry is run by the Bower team with package metadata coming from the bower.json file within each package.

The end result of running Bower is a bower_components directory which can be included in the web app’s build process.

Bower can be run interactively, by installing nodePackages.bower. More interestingly, the Bower components can be declared in a Nix derivation, with the help of nodePackages.bower2nix.

15.4.1. bower2nix usage

Suppose you have a bower.json with the following contents: Example bower.json

  "name": "my-web-app",
  "dependencies": {
    "angular": "~1.5.0",
    "bootstrap": "~3.3.6"

Running bower2nix will produce something like the following output:

{ fetchbower, buildEnv }:
buildEnv { name = "bower-env"; ignoreCollisions = true; paths = [
  (fetchbower "angular" "1.5.3" "~1.5.0" "1749xb0firxdra4rzadm4q9x90v6pzkbd7xmcyjk6qfza09ykk9y")
  (fetchbower "bootstrap" "3.3.6" "~3.3.6" "1vvqlpbfcy0k5pncfjaiskj3y6scwifxygfqnw393sjfxiviwmbv")
  (fetchbower "jquery" "2.2.2" "1.9.1 - 2" "10sp5h98sqwk90y4k6hbdviwqzvzwqf47r3r51pakch5ii2y7js1")

Using the bower2nix command line arguments, the output can be redirected to a file. A name like bower-packages.nix would be fine.

The resulting derivation is a union of all the downloaded Bower packages (and their dependencies). To use it, they still need to be linked together by Bower, which is where buildBowerComponents is useful.

15.4.2. buildBowerComponents function

The function is implemented in pkgs/development/bower-modules/generic/default.nix. Example buildBowerComponents

      bowerComponents = buildBowerComponents {
        name = "my-web-app";
        generated = ./bower-packages.nix; 1
        src = myWebApp; 2

In buildBowerComponents example the following arguments are of special significance to the function:


generated specifies the file which was created by bower2nix.


src is your project's sources. It needs to contain a bower.json file.

buildBowerComponents will run Bower to link together the output of bower2nix, resulting in a bower_components directory which can be used.

Here is an example of a web frontend build process using gulp. You might use grunt, or anything else. Example build script (gulpfile.js)

var gulp = require('gulp');

gulp.task('default', [], function () {

gulp.task('build', [], function () {
  console.log("Just a dummy gulp build");
}); Example Full example — default.nix

      { myWebApp ? { outPath = ./.; name = "myWebApp"; }
      , pkgs ? import <nixpkgs> {}

      pkgs.stdenv.mkDerivation {
        name = "my-web-app-frontend";
        src = myWebApp;

        buildInputs = [ pkgs.nodePackages.gulp ];

        bowerComponents = pkgs.buildBowerComponents { 1
          name = "my-web-app";
          generated = ./bower-packages.nix;
          src = myWebApp;

        buildPhase = ''
          cp --reflink=auto --no-preserve=mode -R $bowerComponents/bower_components . 2
          export HOME=$PWD 3
          ${pkgs.nodePackages.gulp}/bin/gulp build 4

        installPhase = "mv gulpdist $out";

A few notes about Full example — default.nix:


The result of buildBowerComponents is an input to the frontend build.


Whether to symlink or copy the bower_components directory depends on the build tool in use. In this case a copy is used to avoid gulp silliness with permissions.


gulp requires HOME to refer to a writeable directory.


The actual build command. Other tools could be used.

15.4.3. Troubleshooting ENOCACHE errors from buildBowerComponents

This means that Bower was looking for a package version which doesn’t exist in the generated bower-packages.nix.

If bower.json has been updated, then run bower2nix again.

It could also be a bug in bower2nix or fetchbower. If possible, try reformulating the version specification in bower.json.

15.5. Coq and coq packages

15.5.1. Coq derivation: coq

The Coq derivation is overridable through the coq.override overrides, where overrides is an attribute set which contains the arguments to override. We recommend overriding either of the following

  • version (optional, defaults to the latest version of Coq selected for nixpkgs, see pkgs/top-level/coq-packages to witness this choice), which follows the conventions explained in the coqPackages section below,

  • customOCamlPackage (optional, defaults to null, which lets Coq choose a version automatically), which can be set to any of the ocaml packages attribute of ocaml-ng (such as ocaml-ng.ocamlPackages_4_10 which is the default for Coq 8.11 for example).

  • coq-version (optional, defaults to the short version e.g. 8.10), is a version number of the form x.y that indicates which Coq’s version build behavior to mimic when using a source which is not a release. E.g. coq.override { version = "d370a9d1328a4e1cdb9d02ee032f605a9d94ec7a"; coq-version = "8.10"; }.

15.5.2. Coq packages attribute sets: coqPackages

The recommended way of defining a derivation for a Coq library, is to use the coqPackages.mkCoqDerivation function, which is essentially a specialization of mkDerivation taking into account most of the specifics of Coq libraries. The following attributes are supported:

  • pname (required) is the name of the package,

  • version (optional, defaults to null), is the version to fetch and build, this attribute is interpreted in several ways depending on its type and pattern:

    • if it is a known released version string, i.e. from the release attribute below, the according release is picked, and the version attribute of the resulting derivation is set to this release string,

    • if it is a majorMinor "x.y" prefix of a known released version (as defined above), then the latest "x.y.z" known released version is selected (for the ordering given by versionAtLeast),

    • if it is a path or a string representing an absolute path (i.e. starting with "/"), the provided path is selected as a source, and the version attribute of the resulting derivation is set to "dev",

    • if it is a string of the form owner:branch then it tries to download the branch of owner owner for a project of the same name using the same vcs, and the version attribute of the resulting derivation is set to "dev", additionally if the owner is not provided (i.e. if the owner: prefix is missing), it defaults to the original owner of the package (see below),

    • if it is a string of the form "#N", and the domain is github, then it tries to download the current head of the pull request #N from github,

  • defaultVersion (optional). Coq libraries may be compatible with some specific versions of Coq only. The defaultVersion attribute is used when no version is provided (or if version = null) to select the version of the library to use by default, depending on the context. This selection will mainly depend on a coq version number but also possibly on other packages versions (e.g. mathcomp). If its value ends up to be null, the package is marked for removal in end-user coqPackages attribute set.

  • release (optional, defaults to {}), lists all the known releases of the library and for each of them provides an attribute set with at least a sha256 attribute (you may put the empty string "" in order to automatically insert a fake sha256, this will trigger an error which will allow you to find the correct sha256), each attribute set of the list of releases also takes optional overloading arguments for the fetcher as below (i.e.domain, owner, repo, rev assuming the default fetcher is used) and optional overrides for the result of the fetcher (i.e. version and src).

  • fetcher (optional, defaults to a generic fetching mechanism supporting github or gitlab based infrastructures), is a function that takes at least an owner, a repo, a rev, and a sha256 and returns an attribute set with a version and src.

  • repo (optional, defaults to the value of pname),

  • owner (optional, defaults to "coq-community").

  • domain (optional, defaults to ""), domains including the strings "github" or "gitlab" in their names are automatically supported, otherwise, one must change the fetcher argument to support them (cf pkgs/development/coq-modules/heq/default.nix for an example),

  • releaseRev (optional, defaults to (v: v)), provides a default mapping from release names to revision hashes/branch names/tags,

  • displayVersion (optional), provides a way to alter the computation of name from pname, by explaining how to display version numbers,

  • namePrefix (optional, defaults to [ "coq" ]), provides a way to alter the computation of name from pname, by explaining which dependencies must occur in name,

  • extraBuildInputs (optional), by default buildInputs just contains coq, this allows to add more build inputs,

  • mlPlugin (optional, defaults to false). Some extensions (plugins) might require OCaml and sometimes other OCaml packages. Standard dependencies can be added by setting the current option to true. For a finer grain control, the coq.ocamlPackages attribute can be used in extraBuildInputs to depend on the same package set Coq was built against.

  • useDune2ifVersion (optional, default to (x: false) uses Dune2 to build the package if the provided predicate evaluates to true on the version, e.g. useDune2if = versions.isGe "1.1" will use dune if the version of the package is greater or equal to "1.1",

  • useDune2 (optional, defaults to false) uses Dune2 to build the package if set to true, the presence of this attribute overrides the behavior of the previous one.

  • opam-name (optional, defaults to concatenating with a dash separator the components of namePrefix and pname), name of the Dune package to build.

  • enableParallelBuilding (optional, defaults to true), since it is activated by default, we provide a way to disable it.

  • extraInstallFlags (optional), allows to extend installFlags which initializes the variable COQMF_COQLIB so as to install in the proper subdirectory. Indeed Coq libraries should be installed in $(out)/lib/coq/${coq.coq-version}/user-contrib/. Such directories are automatically added to the $COQPATH environment variable by the hook defined in the Coq derivation.

  • setCOQBIN (optional, defaults to true), by default, the environment variable $COQBIN is set to the current Coq’s binary, but one can disable this behavior by setting it to false,

  • useMelquiondRemake (optional, default to null) is an attribute set, which, if given, overloads the preConfigurePhases, configureFlags, buildPhase, and installPhase attributes of the derivation for a specific use in libraries using remake as set up by Guillaume Melquiond for flocq, gappalib, interval, and coquelicot (see the corresponding derivation for concrete examples of use of this option). For backward compatibility, the attribute useMelquiondRemake.logpath must be set to the logical root of the library (otherwise, one can pass useMelquiondRemake = {} to activate this without backward compatibility).

  • dropAttrs, keepAttrs, dropDerivationAttrs are all optional and allow to tune which attribute is added or removed from the final call to mkDerivation.

It also takes other standard mkDerivation attributes, they are added as such, except for meta which extends an automatically computed meta (where the platform is the same as coq and the homepage is automatically computed).

Here is a simple package example. It is a pure Coq library, thus it depends on Coq. It builds on the Mathematical Components library, thus it also takes some mathcomp derivations as extraBuildInputs.

{ lib, mkCoqDerivation, version ? null
, coq, mathcomp, mathcomp-finmap, mathcomp-bigenough }:
with lib; mkCoqDerivation {
  /* namePrefix leads to e.g. `name = coq8.11-mathcomp1.11-multinomials-1.5.2` */
  namePrefix = [ "coq" "mathcomp" ];
  pname = "multinomials";
  owner = "math-comp";
  inherit version;
  defaultVersion =  with versions; switch [ coq.version mathcomp.version ] [
      { cases = [ (range "8.7" "8.12")  "1.11.0" ];             out = "1.5.2"; }
      { cases = [ (range "8.7" "8.11")  (range "1.8" "1.10") ]; out = "1.5.0"; }
      { cases = [ (range "8.7" "8.10")  (range "1.8" "1.10") ]; out = "1.4"; }
      { cases = [ "8.6"                 (range "1.6" "1.7") ];  out = "1.1"; }
    ] null;
  release = {
    "1.5.2".sha256 = "15aspf3jfykp1xgsxf8knqkxv8aav2p39c2fyirw7pwsfbsv2c4s";
    "1.5.1".sha256 = "13nlfm2wqripaq671gakz5mn4r0xwm0646araxv0nh455p9ndjs3";
    "1.5.0".sha256 = "064rvc0x5g7y1a0nip6ic91vzmq52alf6in2bc2dmss6dmzv90hw";
    "1.5.0".rev    = "1.5";
    "1.4".sha256   = "0vnkirs8iqsv8s59yx1fvg1nkwnzydl42z3scya1xp1b48qkgn0p";
    "1.3".sha256   = "0l3vi5n094nx3qmy66hsv867fnqm196r8v605kpk24gl0aa57wh4";
    "1.2".sha256   = "1mh1w339dslgv4f810xr1b8v2w7rpx6fgk9pz96q0fyq49fw2xcq";
    "1.1".sha256   = "1q8alsm89wkc0lhcvxlyn0pd8rbl2nnxg81zyrabpz610qqjqc3s";
    "1.0".sha256   = "1qmbxp1h81cy3imh627pznmng0kvv37k4hrwi2faa101s6bcx55m";

  propagatedBuildInputs =
    [ mathcomp.ssreflect mathcomp.algebra mathcomp-finmap mathcomp-bigenough ];

  meta = {
    description = "A Coq/SSReflect Library for Monoidal Rings and Multinomials";
    license = licenses.cecill-c;

15.6. Crystal

15.6.1. Building a Crystal package

This section uses Mint as an example for how to build a Crystal package.

If the Crystal project has any dependencies, the first step is to get a shards.nix file encoding those. Get a copy of the project and go to its root directory such that its shard.lock file is in the current directory. Executable projects should usually commit the shard.lock file, but sometimes that’s not the case, which means you need to generate it yourself. With an existing shard.lock file, crystal2nix can be run.

$ git clone
$ cd mint
$ git checkout 0.5.0
$ if [ ! -f shard.lock ]; then nix-shell -p shards --run "shards lock"; fi
$ nix-shell -p crystal2nix --run crystal2nix

This should have generated a shards.nix file.

Next create a Nix file for your derivation and use pkgs.crystal.buildCrystalPackage as follows:

with import <nixpkgs> {};
crystal.buildCrystalPackage rec {
  pname = "mint";
  version = "0.5.0";

  src = fetchFromGitHub {
    owner = "mint-lang";
    repo = "mint";
    rev = version;
    sha256 = "0vxbx38c390rd2ysvbwgh89v2232sh5rbsp3nk9wzb70jybpslvl";

  # Insert the path to your shards.nix file here
  shardsFile = ./shards.nix;


This won’t build anything yet, because we haven’t told it what files build. We can specify a mapping from binary names to source files with the crystalBinaries attribute. The project’s compilation instructions should show this. For Mint, the binary is called mint, which is compiled from the source file src/, so we’ll specify this as follows: = "src/";

  # ...

Additionally you can override the default crystal build options (which are currently --release --progress --no-debug --verbose) with = [ "--release" "--verbose" ];

Depending on the project, you might need additional steps to get it to compile successfully. In Mint’s case, we need to link against openssl, so in the end the Nix file looks as follows:

with import <nixpkgs> {};
crystal.buildCrystalPackage rec {
  version = "0.5.0";
  pname = "mint";
  src = fetchFromGitHub {
    owner = "mint-lang";
    repo = "mint";
    rev = version;
    sha256 = "0vxbx38c390rd2ysvbwgh89v2232sh5rbsp3nk9wzb70jybpslvl";

  shardsFile = ./shards.nix; = "src/";

  buildInputs = [ openssl ];

15.7. Dhall

The Nixpkgs support for Dhall assumes some familiarity with Dhall’s language support for importing Dhall expressions, which is documented here:

15.7.1. Remote imports

Nixpkgs bypasses Dhall’s support for remote imports using Dhall’s semantic integrity checks. Specifically, any Dhall import can be protected by an integrity check like:

… and if the import is cached then the interpreter will load the import from cache instead of fetching the URL.

Nixpkgs uses this trick to add all of a Dhall expression’s dependencies into the cache so that the Dhall interpreter never needs to resolve any remote URLs. In fact, Nixpkgs uses a Dhall interpreter with remote imports disabled when packaging Dhall expressions to enforce that the interpreter never resolves a remote import. This means that Nixpkgs only supports building Dhall expressions if all of their remote imports are protected by semantic integrity checks.

Instead of remote imports, Nixpkgs uses Nix to fetch remote Dhall code. For example, the Prelude Dhall package uses pkgs.fetchFromGitHub to fetch the dhall-lang repository containing the Prelude. Relying exclusively on Nix to fetch Dhall code ensures that Dhall packages built using Nix remain pure and also behave well when built within a sandbox.

15.7.2. Packaging a Dhall expression from scratch

We can illustrate how Nixpkgs integrates Dhall by beginning from the following trivial Dhall expression with one dependency (the Prelude):

-- ./true.dhall

let Prelude =

in  Prelude.Bool.not False

As written, this expression cannot be built using Nixpkgs because the expression does not protect the Prelude import with a semantic integrity check, so the first step is to freeze the expression using dhall freeze, like this:

$ dhall freeze --inplace ./true.dhall

… which gives us:

-- ./true.dhall

let Prelude =

in  Prelude.Bool.not False

To package that expression, we create a ./true.nix file containing the following specification for the Dhall package:

# ./true.nix

{ buildDhallPackage, Prelude }:

buildDhallPackage {
  name = "true";
  code = ./true.dhall;
  dependencies = [ Prelude ];
  source = true;

… and we complete the build by incorporating that Dhall package into the pkgs.dhallPackages hierarchy using an overlay, like this:

# ./example.nix

  nixpkgs = builtins.fetchTarball {
    url    = "";
    sha256 = "1pbl4c2dsaz2lximgd31m96jwbps6apn3anx8cvvhk1gl9rkg107";

  dhallOverlay = self: super: {
    true = self.callPackage ./true.nix { };

  overlay = self: super: {
    dhallPackages = super.dhallPackages.override (old: {
      overrides =
        self.lib.composeExtensions (old.overrides or (_: _: {})) dhallOverlay;

  pkgs = import nixpkgs { config = {}; overlays = [ overlay ]; };


… which we can then build using this command:

$ nix build --file ./example.nix dhallPackages.true

15.7.3. Contents of a Dhall package

The above package produces the following directory tree:

$ tree -a ./result
├── .cache
│   └── dhall
│       └── 122027abdeddfe8503496adeb623466caa47da5f63abd2bc6fa19f6cfcb73ecfed70
├── binary.dhall
└── source.dhall

… where:

  • source.dhall contains the result of interpreting our Dhall package:

    $ cat ./result/source.dhall
  • The .cache subdirectory contains one binary cache product encoding the same result as source.dhall:

    $ dhall decode < ./result/.cache/dhall/122027abdeddfe8503496adeb623466caa47da5f63abd2bc6fa19f6cfcb73ecfed70
  • binary.dhall contains a Dhall expression which handles fetching and decoding the same cache product:

    $ cat ./result/binary.dhall
    missing sha256:27abdeddfe8503496adeb623466caa47da5f63abd2bc6fa19f6cfcb73ecfed70
    $ cp -r ./result/.cache .cache
    $ chmod -R u+w .cache
    $ XDG_CACHE_HOME=.cache dhall --file ./result/binary.dhall

The source.dhall file is only present for packages that specify source = true;. By default, Dhall packages omit the source.dhall in order to conserve disk space when they are used exclusively as dependencies. For example, if we build the Prelude package it will only contain the binary encoding of the expression:

$ nix build --file ./example.nix dhallPackages.Prelude

$ tree -a result
├── .cache
│   └── dhall
│       └── 122026b0ef498663d269e4dc6a82b0ee289ec565d683ef4c00d0ebdd25333a5a3c98
└── binary.dhall

2 directories, 2 files

Typically, you only specify source = true; for the top-level Dhall expression of interest (such as our example true.nix Dhall package). However, if you wish to specify source = true for all Dhall packages, then you can amend the Dhall overlay like this:

  dhallOverrides = self: super: {
    # Enable source for all Dhall packages
    buildDhallPackage =
      args: super.buildDhallPackage (args // { source = true; });

    true = self.callPackage ./true.nix { };

… and now the Prelude will contain the fully decoded result of interpreting the Prelude:

$ nix build --file ./example.nix dhallPackages.Prelude

$ tree -a result
├── .cache
│   └── dhall
│       └── 122026b0ef498663d269e4dc6a82b0ee289ec565d683ef4c00d0ebdd25333a5a3c98
├── binary.dhall
└── source.dhall

$ cat ./result/source.dhall
{ Bool =
  { and =
      \(_ : List Bool) ->
        List/fold Bool _ Bool (\(_ : Bool) -> \(_ : Bool) -> _@1 && _) True
  , build = \(_ : Type -> _ -> _@1 -> _@2) -> _ Bool True False
  , even =
      \(_ : List Bool) ->
        List/fold Bool _ Bool (\(_ : Bool) -> \(_ : Bool) -> _@1 == _) True
  , fold =
      \(_ : Bool) ->

15.7.4. Packaging functions

We already saw an example of using buildDhallPackage to create a Dhall package from a single file, but most Dhall packages consist of more than one file and there are two derived utilities that you may find more useful when packaging multiple files:

  • buildDhallDirectoryPackage - build a Dhall package from a local directory

  • buildDhallGitHubPackage - build a Dhall package from a GitHub repository

The buildDhallPackage is the lowest-level function and accepts the following arguments:

  • name: The name of the derivation

  • dependencies: Dhall dependencies to build and cache ahead of time

  • code: The top-level expression to build for this package

    Note that the code field accepts an arbitrary Dhall expression. You’re not limited to just a file.

  • source: Set to true to include the decoded result as source.dhall in the build product, at the expense of requiring more disk space

  • documentationRoot: Set to the root directory of the package if you want dhall-docs to generate documentation underneath the docs subdirectory of the build product

The buildDhallDirectoryPackage is a higher-level function implemented in terms of buildDhallPackage that accepts the following arguments:

  • name: Same as buildDhallPackage

  • dependencies: Same as buildDhallPackage

  • source: Same as buildDhallPackage

  • src: The directory containing Dhall code that you want to turn into a Dhall package

  • file: The top-level file (package.dhall by default) that is the entrypoint to the rest of the package

  • document: Set to true to generate documentation for the package

The buildDhallGitHubPackage is another higher-level function implemented in terms of buildDhallPackage that accepts the following arguments:

  • name: Same as buildDhallPackage

  • dependencies: Same as buildDhallPackage

  • source: Same as buildDhallPackage

  • owner: The owner of the repository

  • repo: The repository name

  • rev: The desired revision (or branch, or tag)

  • directory: The subdirectory of the Git repository to package (if a directory other than the root of the repository)

  • file: The top-level file (${directory}/package.dhall by default) that is the entrypoint to the rest of the package

  • document: Set to true to generate documentation for the package

Additionally, buildDhallGitHubPackage accepts the same arguments as fetchFromGitHub, such as sha256 or fetchSubmodules.

15.7.5. dhall-to-nixpkgs

You can use the dhall-to-nixpkgs command-line utility to automate packaging Dhall code. For example:

$ nix-env --install --attr haskellPackages.dhall-nixpkgs

$ nix-env --install --attr nix-prefetch-git  # Used by dhall-to-nixpkgs

$ dhall-to-nixpkgs github
{ buildDhallGitHubPackage, Prelude }:
  buildDhallGitHubPackage {
    name = "dhall-semver";
    githubBase = "";
    owner = "Gabriel439";
    repo = "dhall-semver";
    rev = "2d44ae605302ce5dc6c657a1216887fbb96392a4";
    fetchSubmodules = false;
    sha256 = "0y8shvp8srzbjjpmnsvz9c12ciihnx1szs0yzyi9ashmrjvd0jcz";
    directory = "";
    file = "package.dhall";
    source = false;
    document = false;
    dependencies = [ (Prelude.overridePackage { file = "package.dhall"; }) ];

The utility takes care of automatically detecting remote imports and converting them to package dependencies. You can also use the utility on local Dhall directories, too:

$ dhall-to-nixpkgs directory ~/proj/dhall-semver
{ buildDhallDirectoryPackage, Prelude }:
  buildDhallDirectoryPackage {
    name = "proj";
    src = ~/proj/dhall-semver;
    file = "package.dhall";
    source = false;
    document = false;
    dependencies = [ (Prelude.overridePackage { file = "package.dhall"; }) ];
    } Remote imports as fixed-output derivations

dhall-to-nixpkgs has the ability to fetch and build remote imports as fixed-output derivations by using their Dhall integrity check. This is sometimes easier than manually packaging all remote imports.

This can be used like the following:

$ dhall-to-nixpkgs directory --fixed-output-derivations ~/proj/dhall-semver
{ buildDhallDirectoryPackage, buildDhallUrl }:
  buildDhallDirectoryPackage {
    name = "proj";
    src = ~/proj/dhall-semver;
    file = "package.dhall";
    source = false;
    document = false;
    dependencies = [
      (buildDhallUrl {
        url = "";
        hash = "sha256-ENs8kZwl6QRoM9+Jeo/+JwHcOQ+giT2VjDQwUkvlpD4=";
        dhallHash = "sha256:10db3c919c25e9046833df897a8ffe2701dc390fa0893d958c3430524be5a43e";

Here, dhall-semver’s Prelude dependency is fetched and built with the buildDhallUrl helper function, instead of being passed in as a function argument.

15.7.6. Overriding dependency versions

Suppose that we change our true.dhall example expression to depend on an older version of the Prelude (19.0.0):

-- ./true.dhall

let Prelude =

in  Prelude.Bool.not False

If we try to rebuild that expression the build will fail:

$ nix build --file ./example.nix dhallPackages.true
builder for '/nix/store/0f1hla7ff1wiaqyk1r2ky4wnhnw114fi-true.drv' failed with exit code 1; last 10 log lines:

  Dhall was compiled without the 'with-http' flag.

  The requested URL was:

  5│         sha256:eb693342eb769f782174157eba9b5924cf8ac6793897fc36a31ccbd6f56dafe2

[1 built (1 failed), 0.0 MiB DL]
error: build of '/nix/store/0f1hla7ff1wiaqyk1r2ky4wnhnw114fi-true.drv' failed

… because the default Prelude selected by Nixpkgs revision 94b2848559b12a8ed1fe433084686b2a81123c99is is version 20.1.0, which doesn’t have the same integrity check as version 19.0.0. This means that version 19.0.0 is not cached and the interpreter is not allowed to fall back to importing the URL.

However, we can override the default Prelude version by using dhall-to-nixpkgs to create a Dhall package for our desired Prelude:

$ dhall-to-nixpkgs github \
    --name Prelude \
    --directory Prelude \
    --rev v19.0.0 \
    > Prelude.nix

… and then referencing that package in our Dhall overlay, by either overriding the Prelude globally for all packages, like this:

  dhallOverrides = self: super: {
    true = self.callPackage ./true.nix { };

    Prelude = self.callPackage ./Prelude.nix { };

… or selectively overriding the Prelude dependency for just the true package, like this:

  dhallOverrides = self: super: {
    true = self.callPackage ./true.nix {
      Prelude = self.callPackage ./Prelude.nix { };

15.7.7. Overrides

You can override any of the arguments to buildDhallGitHubPackage or buildDhallDirectoryPackage using the overridePackage attribute of a package. For example, suppose we wanted to selectively enable source = true just for the Prelude. We can do that like this:

  dhallOverrides = self: super: {
    Prelude = super.Prelude.overridePackage { source = true; };


15.8. Dotnet

15.8.1. Local Development Workflow

For local development, it’s recommended to use nix-shell to create a dotnet environment:

# shell.nix
with import <nixpkgs> {};

mkShell {
  name = "dotnet-env";
  packages = [
} Using many sdks in a workflow

It’s very likely that more than one sdk will be needed on a given project. Dotnet provides several different frameworks (E.g dotnetcore, aspnetcore, etc.) as well as many versions for a given framework. Normally, dotnet is able to fetch a framework and install it relative to the executable. However, this would mean writing to the nix store in nixpkgs, which is read-only. To support the many-sdk use case, one can compose an environment using dotnetCorePackages.combinePackages:

with import <nixpkgs> {};

mkShell {
  name = "dotnet-env";
  packages = [
    (with dotnetCorePackages; combinePackages [

This will produce a dotnet installation that has the dotnet 3.1, 3.0, and 2.1 sdk. The first sdk listed will have it’s cli utility present in the resulting environment. Example info output:

$ dotnet --info
.NET Core SDK (reflecting any global.json):
 Version:   3.1.101
 Commit:    b377529961


.NET Core SDKs installed:
  2.1.803 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/sdk]
  3.0.102 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/sdk]
  3.1.101 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/sdk]

.NET Core runtimes installed:
  Microsoft.AspNetCore.All 2.1.15 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.AspNetCore.All]
  Microsoft.AspNetCore.App 2.1.15 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.AspNetCore.App]
  Microsoft.AspNetCore.App 3.0.2 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.AspNetCore.App]
  Microsoft.AspNetCore.App 3.1.1 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.AspNetCore.App]
  Microsoft.NETCore.App 2.1.15 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.NETCore.App]
  Microsoft.NETCore.App 3.0.2 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.NETCore.App]
  Microsoft.NETCore.App 3.1.1 [/nix/store/iiv98i2jdi226dgh4jzkkj2ww7f8jgpd-dotnet-core-combined/shared/Microsoft.NETCore.App]

15.8.2. dotnet-sdk vs dotnetCorePackages.sdk

The dotnetCorePackages.sdk_X_Y is preferred over the old dotnet-sdk as both major and minor version are very important for a dotnet environment. If a given minor version isn’t present (or was changed), then this will likely break your ability to build a project.

15.8.3. dotnetCorePackages.sdk vs dotnetCorePackages.runtime vs dotnetCorePackages.aspnetcore

The dotnetCorePackages.sdk contains both a runtime and the full sdk of a given version. The runtime and aspnetcore packages are meant to serve as minimal runtimes to deploy alongside already built applications.

15.8.4. Packaging a Dotnet Application

To package Dotnet applications, you can use buildDotnetModule. This has similar arguments to stdenv.mkDerivation, with the following additions:

  • projectFile has to be used for specifying the dotnet project file relative to the source root. These usually have .sln or .csproj file extensions. This can be an array of multiple projects as well.

  • nugetDeps has to be used to specify the NuGet dependency file. Unfortunately, these cannot be deterministically fetched without a lockfile. This file should be generated using nuget-to-nix tool, which is available in nixpkgs.

  • executables is used to specify which executables get wrapped to $out/bin, relative to $out/lib/$pname. If this is unset, all executables generated will get installed. If you do not want to install any, set this to [].

  • runtimeDeps is used to wrap libraries into LD_LIBRARY_PATH. This is how dotnet usually handles runtime dependencies.

  • buildType is used to change the type of build. Possible values are Release, Debug, etc. By default, this is set to Release.

  • dotnet-sdk is useful in cases where you need to change what dotnet SDK is being used.

  • dotnet-runtime is useful in cases where you need to change what dotnet runtime is being used. This can be either a regular dotnet runtime, or an aspnetcore.

  • dotnet-test-sdk is useful in cases where unit tests expect a different dotnet SDK. By default, this is set to the dotnet-sdk attribute.

  • testProjectFile is useful in cases where the regular project file does not contain the unit tests. By default, this is set to the projectFile attribute.

  • disabledTests is used to disable running specific unit tests. This gets passed as: dotnet test --filter "FullyQualifiedName!={}", to ensure compatibility with all unit test frameworks.

  • dotnetRestoreFlags can be used to pass flags to dotnet restore.

  • dotnetBuildFlags can be used to pass flags to dotnet build.

  • dotnetTestFlags can be used to pass flags to dotnet test.

  • dotnetInstallFlags can be used to pass flags to dotnet install.

  • dotnetFlags can be used to pass flags to all of the above phases.

Here is an example default.nix, using some of the previously discussed arguments:

{ lib, buildDotnetModule, dotnetCorePackages, ffmpeg }:

buildDotnetModule rec {
  pname = "someDotnetApplication";
  version = "0.1";

  src = ./.;

  projectFile = "src/project.sln";
  nugetDeps = ./deps.nix; # File generated with `nuget-to-nix path/to/src > deps.nix`.

  dotnet-sdk = dotnetCorePackages.sdk_3_1;
  dotnet-runtime = dotnetCorePackages.net_5_0;
  dotnetFlags = [ "--runtime linux-x64" ];

  executables = [ "foo" ]; # This wraps "$out/lib/$pname/foo" to `$out/bin/foo`.
  executables = []; # Don't install any executables.

  runtimeDeps = [ ffmpeg ]; # This will wrap ffmpeg's library path into `LD_LIBRARY_PATH`.

15.9. Emscripten

Emscripten: An LLVM-to-JavaScript Compiler

This section of the manual covers how to use emscripten in nixpkgs.

Minimal requirements:

  • nix

  • nixpkgs

Modes of use of emscripten:

  • Imperative usage (on the command line):

    If you want to work with emcc, emconfigure and emmake as you are used to from Ubuntu and similar distributions you can use these commands:

    • nix-env -i emscripten

    • nix-shell -p emscripten

  • Declarative usage:

    This mode is far more power full since this makes use of nix for dependency management of emscripten libraries and targets by using the mkDerivation which is implemented by pkgs.emscriptenStdenv and pkgs.buildEmscriptenPackage. The source for the packages is in pkgs/top-level/emscripten-packages.nix and the abstraction behind it in pkgs/development/em-modules/generic/default.nix.

    • build and install all packages:

      • nix-env -iA emscriptenPackages

    • dev-shell for zlib implementation hacking:

      • nix-shell -A emscriptenPackages.zlib

15.9.1. Imperative usage

A few things to note:

  • export EMCC_DEBUG=2 is nice for debugging

  • ~/.emscripten, the build artifact cache sometimes creates issues and needs to be removed from time to time

15.9.2. Declarative usage

Let’s see two different examples from pkgs/top-level/emscripten-packages.nix:

  • pkgs.zlib.override

  • pkgs.buildEmscriptenPackage

Both are interesting concepts.

A special requirement of the pkgs.buildEmscriptenPackage is the doCheck = true is a default meaning that each emscriptenPackage requires a checkPhase implemented.

  • Use export EMCC_DEBUG=2 from within a emscriptenPackage’s phase to get more detailed debug output what is going wrong.

  • ~/.emscripten cache is requiring us to set HOME=$TMPDIR in individual phases. This makes compilation slower but also makes it more deterministic. Usage 1: pkgs.zlib.override

This example uses zlib from nixpkgs but instead of compiling C to ELF it compiles C to JS since we were using pkgs.zlib.override and changed stdenv to pkgs.emscriptenStdenv. A few adaptions and hacks were set in place to make it working. One advantage is that when pkgs.zlib is updated, it will automatically update this package as well. However, this can also be the downside…

See the zlib example:

zlib = (pkgs.zlib.override {
  stdenv = pkgs.emscriptenStdenv;
(old: rec {
  buildInputs = old.buildInputs ++ [ pkg-config ];
  # we need to reset this setting!
  configurePhase = ''
    # FIXME: Some tests require writing at $HOME
    runHook preConfigure

    #export EMCC_DEBUG=2
    emconfigure ./configure --prefix=$out --shared

    runHook postConfigure
  dontStrip = true;
  outputs = [ "out" ];
  buildPhase = ''
    emmake make
  installPhase = ''
    emmake make install
  checkPhase = ''
    echo "================= testing zlib using node ================="

    echo "Compiling a custom test"
    set -x
    emcc -O2 -s EMULATE_FUNCTION_POINTER_CASTS=1 test/example.c -DZ_SOLO \${old.version} -I . -o example.js

    echo "Using node to execute the test"
    ${pkgs.nodejs}/bin/node ./example.js

    set +x
    if [ $? -ne 0 ]; then
      echo "test failed for some reason"
      exit 1;
      echo "it seems to work! very good."
    echo "================= /testing zlib using node ================="

  postPatch = pkgs.lib.optionalString pkgs.stdenv.isDarwin ''
    substituteInPlace configure \
      --replace '/usr/bin/libtool' 'ar' \
      --replace 'AR="libtool"' 'AR="ar"' \
      --replace 'ARFLAGS="-o"' 'ARFLAGS="-r"'
}); Usage 2: pkgs.buildEmscriptenPackage

This xmlmirror example features a emscriptenPackage which is defined completely from this context and no pkgs.zlib.override is used.

xmlmirror = pkgs.buildEmscriptenPackage rec {
  name = "xmlmirror";

  buildInputs = [ pkg-config autoconf automake libtool gnumake libxml2 nodejs openjdk json_c ];
  nativeBuildInputs = [ pkg-config zlib ];

  src = pkgs.fetchgit {
    url = "";
    rev = "4fd7e86f7c9526b8f4c1733e5c8b45175860a8fd";
    sha256 = "1jasdqnbdnb83wbcnyrp32f36w3xwhwp0wq8lwwmhqagxrij1r4b";

  configurePhase = ''
    rm -f fastXmlLint.js*
    # a fix for ERROR:root:For asm.js, TOTAL_MEMORY must be a multiple of 16MB, was 234217728
    sed -e "s/TOTAL_MEMORY=234217728/TOTAL_MEMORY=268435456/g" -i Makefile.emEnv
    sed -e "s/-o fastXmlLint.js/-s EXTRA_EXPORTED_RUNTIME_METHODS='[\"ccall\", \"cwrap\"]' -o fastXmlLint.js/g" -i Makefile.emEnv

  buildPhase = ''
    make -f Makefile.emEnv

  outputs = [ "out" "doc" ];

  installPhase = ''
    mkdir -p $out/share
    mkdir -p $doc/share/${name}

    cp Demo* $out/share
    cp -R codemirror-5.12 $out/share
    cp fastXmlLint.js* $out/share
    cp *.xsd $out/share
    cp *.js $out/share
    cp *.xhtml $out/share
    cp *.html $out/share
    cp *.json $out/share
    cp *.rng $out/share
    cp $doc/share/${name}
  checkPhase = ''

}; Declarative debugging

Use nix-shell -I nixpkgs=/some/dir/nixpkgs -A emscriptenPackages.libz and from there you can go trough the individual steps. This makes it easy to build a good unit test or list the files of the project.

  1. nix-shell -I nixpkgs=/some/dir/nixpkgs -A emscriptenPackages.libz

  2. cd /tmp/

  3. unpackPhase

  4. cd libz-1.2.3

  5. configurePhase

  6. buildPhase

  7. … happy hacking…

15.9.3. Summary

Using this toolchain makes it easy to leverage nix from NixOS, MacOSX or even Windows (WSL+ubuntu+nix). This toolchain is reproducible, behaves like the rest of the packages from nixpkgs and contains a set of well working examples to learn and adapt from.

If in trouble, ask the maintainers.

15.10. GNOME

15.10.1. Packaging GNOME applications

Programs in the GNOME universe are written in various languages but they all use GObject-based libraries like GLib, GTK or GStreamer. These libraries are often modular, relying on looking into certain directories to find their modules. However, due to Nix’s specific file system organization, this will fail without our intervention. Fortunately, the libraries usually allow overriding the directories through environment variables, either natively or thanks to a patch in nixpkgs. Wrapping the executables to ensure correct paths are available to the application constitutes a significant part of packaging a modern desktop application. In this section, we will describe various modules needed by such applications, environment variables needed to make the modules load, and finally a script that will do the work for us. Settings

GSettings API is often used for storing settings. GSettings schemas are required, to know the type and other metadata of the stored values. GLib looks for glib-2.0/schemas/gschemas.compiled files inside the directories of XDG_DATA_DIRS.

On Linux, GSettings API is implemented using dconf backend. You will need to add dconf GIO module to GIO_EXTRA_MODULES variable, otherwise the memory backend will be used and the saved settings will not be persistent.

Last you will need the dconf database D-Bus service itself. You can enable it using programs.dconf.enable.

Some applications will also require gsettings-desktop-schemas for things like reading proxy configuration or user interface customization. This dependency is often not mentioned by upstream, you should grep for org.gnome.desktop and org.gnome.system to see if the schemas are needed. GIO modules

GLib’s GIO library supports several extension points. Notably, they allow:

  • implementing settings backends (already mentioned)

  • adding TLS support

  • proxy settings

  • virtual file systems

The modules are typically installed to lib/gio/modules/ directory of a package and you need to add them to GIO_EXTRA_MODULES if you need any of those features.

In particular, we recommend:

  • adding dconf.lib for any software on Linux that reads GSettings (even transitivily through e.g. GTK’s file manager)

  • adding glib-networking for any software that accesses network using GIO or libsoup – glib-networking contains a module that implements TLS support and loads system-wide proxy settings

To allow software to use various virtual file systems, gvfs package can be also added. But that is usually an optional feature so we typically use gvfs from the system (e.g. installed globally using NixOS module). GdkPixbuf loaders

GTK applications typically use GdkPixbuf to load images. But gdk-pixbuf package only supports basic bitmap formats like JPEG, PNG or TIFF, requiring to use third-party loader modules for other formats. This is especially painful since GTK itself includes SVG icons, which cannot be rendered without a loader provided by librsvg.

Unlike other libraries mentioned in this section, GdkPixbuf only supports a single value in its controlling environment variable GDK_PIXBUF_MODULE_FILE. It is supposed to point to a cache file containing information about the available loaders. Each loader package will contain a lib/gdk-pixbuf-2.0/2.10.0/loaders.cache file describing the default loaders in gdk-pixbuf package plus the loader contained in the package itself. If you want to use multiple third-party loaders, you will need to create your own cache file manually. Fortunately, this is pretty rare as not many loaders exist.

gdk-pixbuf contains a setup hook that sets GDK_PIXBUF_MODULE_FILE from dependencies but as mentioned in further section, it is pretty limited. Loaders should propagate this setup hook. Icons

When an application uses icons, an icon theme should be available in XDG_DATA_DIRS during runtime. The package for the default, icon-less hicolor-icon-theme (should be propagated by every icon theme) contains a setup hook that will pick up icon themes from buildInputs and pass it to our wrapper. Unfortunately, relying on that would mean every user has to download the theme included in the package expression no matter their preference. For that reason, we leave the installation of icon theme on the user. If you use one of the desktop environments, you probably already have an icon theme installed.

To avoid costly file system access when locating icons, GTK, as well as Qt, can rely on icon-theme.cache files from the themes’ top-level directories. These files are generated using gtk-update-icon-cache, which is expected to be run whenever an icon is added or removed to an icon theme (typically an application icon into hicolor theme) and some programs do indeed run this after icon installation. However, since packages are installed into their own prefix by Nix, this would lead to conflicts. For that reason, gtk3 provides a setup hook that will clean the file from installation. Since most applications only ship their own icon that will be loaded on start-up, it should not affect them too much. On the other hand, icon themes are much larger and more widely used so we need to cache them. Because we recommend installing icon themes globally, we will generate the cache files from all packages in a profile using a NixOS module. You can enable the cache generation using gtk.iconCache.enable option if your desktop environment does not already do that. Packaging icon themes

Icon themes may inherit from other icon themes. The inheritance is specified using the Inherits key in the index.theme file distributed with the icon theme. According to the icon theme specification, icons not provided by the theme are looked for in its parent icon themes. Therefore the parent themes should be installed as dependencies for a more complete experience regarding the icon sets used.

The package hicolor-icon-theme provides a setup hook which makes symbolic links for the parent themes into the directory share/icons of the current theme directory in the nix store, making sure they can be found at runtime. For that to work the packages providing parent icon themes should be listed as propagated build dependencies, together with hicolor-icon-theme.

Also make sure that icon-theme.cache is installed for each theme provided by the package, and set dontDropIconThemeCache to true so that the cache file is not removed by the gtk3 setup hook. GTK Themes

Previously, a GTK theme needed to be in XDG_DATA_DIRS. This is no longer necessary for most programs since GTK incorporated Adwaita theme. Some programs (for example, those designed for elementary HIG) might require a special theme like pantheon.elementary-gtk-theme. GObject introspection typelibs

GObject introspection allows applications to use C libraries in other languages easily. It does this through typelib files searched in GI_TYPELIB_PATH. Various plug-ins

If your application uses GStreamer or Grilo, you should set GST_PLUGIN_SYSTEM_PATH_1_0 and GRL_PLUGIN_PATH, respectively.

15.10.2. Onto wrapGAppsHook

Given the requirements above, the package expression would become messy quickly:

preFixup = ''
  for f in $(find $out/bin/ $out/libexec/ -type f -executable); do
    wrapProgram "$f" \
      --prefix GIO_EXTRA_MODULES : "${getLib dconf}/lib/gio/modules" \
      --prefix XDG_DATA_DIRS : "$out/share" \
      --prefix XDG_DATA_DIRS : "$out/share/gsettings-schemas/${name}" \
      --prefix XDG_DATA_DIRS : "${gsettings-desktop-schemas}/share/gsettings-schemas/${}" \
      --prefix XDG_DATA_DIRS : "${hicolor-icon-theme}/share" \
      --prefix GI_TYPELIB_PATH : "${lib.makeSearchPath "lib/girepository-1.0" [ pango json-glib ]}"

Fortunately, there is wrapGAppsHook. It works in conjunction with other setup hooks that populate environment variables, and it will then wrap all executables in bin and libexec directories using said variables.

For convenience, it also adds dconf.lib for a GIO module implementing a GSettings backend using dconf, gtk3 for GSettings schemas, and librsvg for GdkPixbuf loader to the closure. There is also wrapGAppsHook4, which replaces GTK 3 with GTK 4. And in case you are packaging a program without a graphical interface, you might want to use wrapGAppsNoGuiHook, which runs the same script as wrapGAppsHook but does not bring gtk3 and librsvg into the closure.

  • wrapGAppsHook itself will add the package’s share directory to XDG_DATA_DIRS.

  • glib setup hook will populate GSETTINGS_SCHEMAS_PATH and then wrapGAppsHook will prepend it to XDG_DATA_DIRS.

  • gdk-pixbuf setup hook will populate GDK_PIXBUF_MODULE_FILE with the path to biggest loaders.cache file from the dependencies containing GdkPixbuf loaders. This works fine when there are only two packages containing loaders (gdk-pixbuf and e.g. librsvg) – it will choose the second one, reasonably expecting that it will be bigger since it describes extra loader in addition to the default ones. But when there are more than two loader packages, this logic will break. One possible solution would be constructing a custom cache file for each package containing a program like services/x11/gdk-pixbuf.nix NixOS module does. wrapGAppsHook copies the GDK_PIXBUF_MODULE_FILE environment variable into the produced wrapper.

  • One of gtk3’s setup hooks will remove icon-theme.cache files from package’s icon theme directories to avoid conflicts. Icon theme packages should prevent this with dontDropIconThemeCache = true;.

  • dconf.lib is a dependency of wrapGAppsHook, which then also adds it to the GIO_EXTRA_MODULES variable.

  • hicolor-icon-theme’s setup hook will add icon themes to XDG_ICON_DIRS which is prepended to XDG_DATA_DIRS by wrapGAppsHook.

  • gobject-introspection setup hook populates GI_TYPELIB_PATH variable with lib/girepository-1.0 directories of dependencies, which is then added to wrapper by wrapGAppsHook. It also adds share directories of dependencies to XDG_DATA_DIRS, which is intended to promote GIR files but it also pollutes the closures of packages using wrapGAppsHook.

    Warning: The setup hook currently does not work in expressions with strictDeps enabled, like Python packages. In those cases, you will need to disable it with strictDeps = false;.
  • Setup hooks of gst_all_1.gstreamer and grilo will populate the GST_PLUGIN_SYSTEM_PATH_1_0 and GRL_PLUGIN_PATH variables, respectively, which will then be added to the wrapper by wrapGAppsHook.

You can also pass additional arguments to makeWrapper using gappsWrapperArgs in preFixup hook:

preFixup = ''
    # Thumbnailers
    --prefix XDG_DATA_DIRS : "${gdk-pixbuf}/share"
    --prefix XDG_DATA_DIRS : "${librsvg}/share"
    --prefix XDG_DATA_DIRS : "${shared-mime-info}/share"

15.10.3. Updating GNOME packages

Most GNOME package offer updateScript, it is therefore possible to update to latest source tarball by running nix-shell maintainers/scripts/update.nix --argstr package gnome.nautilus or even en masse with nix-shell maintainers/scripts/update.nix --argstr path gnome. Read the package’s NEWS file to see what changed.

15.10.4. Frequently encountered issues GLib-GIO-ERROR **: 06:04:50.903: No GSettings schemas are installed on the system

There are no schemas available in XDG_DATA_DIRS. Temporarily add a random package containing schemas like gsettings-desktop-schemas to buildInputs. glib and wrapGAppsHook setup hooks will take care of making the schemas available to application and you will see the actual missing schemas with the next error. Or you can try looking through the source code for the actual schemas used. GLib-GIO-ERROR **: 06:04:50.903: Settings schema ‘’ is not installed

Package is missing some GSettings schemas. You can find out the package containing the schema with nix-locate and let the hooks handle the wrapping as above. When using wrapGAppsHook with special derivers you can end up with double wrapped binaries.

This is because derivers like python.pkgs.buildPythonApplication or qt5.mkDerivation have setup-hooks automatically added that produce wrappers with makeWrapper. The simplest way to workaround that is to disable the wrapGAppsHook automatic wrapping with dontWrapGApps = true; and pass the arguments it intended to pass to makeWrapper to another.

In the case of a Python application it could look like:

python3.pkgs.buildPythonApplication {
  pname = "gnome-music";
  version = "3.32.2";

  nativeBuildInputs = [

  dontWrapGApps = true;

  # Arguments to be passed to `makeWrapper`, only used by buildPython*
  preFixup = ''

And for a QT app like:

mkDerivation {
  pname = "calibre";
  version = "3.47.0";

  nativeBuildInputs = [

  dontWrapGApps = true;

  # Arguments to be passed to `makeWrapper`, only used by qt5’s mkDerivation
  preFixup = ''
} I am packaging a project that cannot be wrapped, like a library or GNOME Shell extension.

You can rely on applications depending on the library setting the necessary environment variables but that is often easy to miss. Instead we recommend to patch the paths in the source code whenever possible. Here are some examples: I need to wrap a binary outside bin and libexec directories.

You can manually trigger the wrapping with wrapGApp in preFixup phase. It takes a path to a program as a first argument; the remaining arguments are passed directly to wrapProgram function.

15.11. Go

15.11.1. Go modules

The function buildGoModule builds Go programs managed with Go modules. It builds a Go Modules through a two phase build:

  • An intermediate fetcher derivation. This derivation will be used to fetch all of the dependencies of the Go module.

  • A final derivation will use the output of the intermediate derivation to build the binaries and produce the final output. Example for buildGoModule

In the following is an example expression using buildGoModule, the following arguments are of special significance to the function:

  • vendorSha256: is the hash of the output of the intermediate fetcher derivation. vendorSha256 can also take null as an input. When null is used as a value, rather than fetching the dependencies and vendoring them, we use the vendoring included within the source repo. If you’d like to not have to update this field on dependency changes, run go mod vendor in your source repo and set vendorSha256 = null;

  • runVend: runs the vend command to generate the vendor directory. This is useful if your code depends on c code and go mod tidy does not include the needed sources to build.

  • proxyVendor: Fetches (go mod download) and proxies the vendor directory. This is useful if any dependency has case-insensitive conflicts which will produce platform dependant vendorSha256 checksums.

pet = buildGoModule rec {
  pname = "pet";
  version = "0.3.4";

  src = fetchFromGitHub {
    owner = "knqyf263";
    repo = "pet";
    rev = "v${version}";
    sha256 = "0m2fzpqxk7hrbxsgqplkg7h2p7gv6s1miymv3gvw0cz039skag0s";

  vendorSha256 = "1879j77k96684wi554rkjxydrj8g3hpp0kvxz03sd8dmwr3lh83j";

  runVend = true;

  meta = with lib; {
    description = "Simple command-line snippet manager, written in Go";
    homepage = "";
    license =;
    maintainers = with maintainers; [ kalbasit ];
    platforms = platforms.linux ++ platforms.darwin;

15.11.2. buildGoPackage (legacy)

The function buildGoPackage builds legacy Go programs, not supporting Go modules. Example for buildGoPackage

In the following is an example expression using buildGoPackage, the following arguments are of special significance to the function:

  • goPackagePath specifies the package’s canonical Go import path.

  • goDeps is where the Go dependencies of a Go program are listed as a list of package source identified by Go import path. It could be imported as a separate deps.nix file for readability. The dependency data structure is described below.

deis = buildGoPackage rec {
  pname = "deis";
  version = "1.13.0";

  goPackagePath = "";

  src = fetchFromGitHub {
    owner = "deis";
    repo = "deis";
    rev = "v${version}";
    sha256 = "1qv9lxqx7m18029lj8cw3k7jngvxs4iciwrypdy0gd2nnghc68sw";

  goDeps = ./deps.nix;

The goDeps attribute can be imported from a separate nix file that defines which Go libraries are needed and should be included in GOPATH for buildPhase:

# deps.nix
[ # goDeps is a list of Go dependencies.
    # goPackagePath specifies Go package import path.
    goPackagePath = "";
    fetch = {
      # `fetch type` that needs to be used to get package source.
      # If `git` is used there should be `url`, `rev` and `sha256` defined next to it.
      type = "git";
      url = "";
      rev = "a83829b6f1293c91addabc89d0571c246397bbf4";
      sha256 = "1m4dsmk90sbi17571h6pld44zxz7jc4lrnl4f27dpd1l8g5xvjhh";
    goPackagePath = "";
    fetch = {
      type = "git";
      url = "";
      rev = "784ddc588536785e7299f7272f39101f7faccc3f";
      sha256 = "0wwz48jl9fvl1iknvn9dqr4gfy1qs03gxaikrxxp9gry6773v3sj";

To extract dependency information from a Go package in automated way use go2nix. It can produce complete derivation and goDeps file for Go programs.

You may use Go packages installed into the active Nix profiles by adding the following to your ~/.bashrc:

for p in $NIX_PROFILES; do

15.11.3. Attributes used by the builders

Both buildGoModule and buildGoPackage can be tweaked to behave slightly differently, if the following attributes are used: ldflags

Arguments to pass to the Go linker tool via the -ldflags argument of go build. The most common use case for this argument is to make the resulting executable aware of its own version. For example:

  ldflags = [
    "-s" "-w"
    "-X main.Version=${version}"
    "-X main.Commit=${version}"
  ]; tags

Arguments to pass to the Go via the -tags argument of go build. For example:

  tags = [
  tags = [ "production" ] ++ lib.optionals withSqlite [ "sqlite" ]; deleteVendor

Removes the pre-existing vendor directory. This should only be used if the dependencies included in the vendor folder are broken or incomplete. subPackages

Limits the builder from building child packages that have not been listed. If subPackages is not specified, all child packages will be built.

15.12. Haskell

The documentation for the Haskell infrastructure is published at The source code for that site lives in the doc/ sub-directory of the cabal2nix Git repository and changes can be submitted there.

15.13. Hy

15.13.1. Installation Installation without packages

You can install hy via nix-env or by adding it to configuration.nix by reffering to it as a hy attribute. This kind of installation adds hy to your environment and it succesfully works with python3.


Packages that are installed with your python derivation, are not accesible by hy this way. Installation with packages

Creating hy derivation with custom python packages is really simple and similar to the way that python does it. Attribute hy provides function withPackages that creates custom hy derivation with specified packages.

For example if you want to create shell with matplotlib and numpy, you can do it like so:

$ nix-shell -p "hy.withPackages (ps: with ps; [ numpy matplotlib ])"

Or if you want to extend your configuration.nix:

{ # ...

  environment.systemPackages = with pkgs; [
    (hy.withPackages (py-packages: with py-packages; [ numpy matplotlib ]))

15.14. Idris

15.14.1. Installing Idris

The easiest way to get a working idris version is to install the idris attribute:

$ # On NixOS
$ nix-env -i nixos.idris
$ # On non-NixOS
$ nix-env -i nixpkgs.idris

This however only provides the prelude and base libraries. To install idris with additional libraries, you can use the idrisPackages.with-packages function, e.g. in an overlay in ~/.config/nixpkgs/overlays/my-idris.nix:

self: super: {
  myIdris = with self.idrisPackages; with-packages [ contrib pruviloj ];

And then:

$ # On NixOS
$ nix-env -iA nixos.myIdris
$ # On non-NixOS
$ nix-env -iA nixpkgs.myIdris

To see all available Idris packages:

$ # On NixOS
$ nix-env -qaPA nixos.idrisPackages
$ # On non-NixOS
$ nix-env -qaPA nixpkgs.idrisPackages

Similarly, entering a nix-shell:

$ nix-shell -p 'idrisPackages.with-packages (with idrisPackages; [ contrib pruviloj ])'

15.14.2. Starting Idris with library support

To have access to these libraries in idris, call it with an argument -p <library name> for each library:

$ nix-shell -p 'idrisPackages.with-packages (with idrisPackages; [ contrib pruviloj ])'
[nix-shell:~]$ idris -p contrib -p pruviloj

A listing of all available packages the Idris binary has access to is available via --listlibs:

$ idris --listlibs

15.14.3. Building an Idris project with Nix

As an example of how a Nix expression for an Idris package can be created, here is the one for idrisPackages.yaml:

{ lib
, build-idris-package
, fetchFromGitHub
, contrib
, lightyear
build-idris-package  {
  name = "yaml";
  version = "2018-01-25";

  # This is the .ipkg file that should be built, defaults to the package name
  # In this case it should build `Yaml.ipkg` instead of `yaml.ipkg`
  # This is only necessary because the yaml packages ipkg file is
  # different from its package name here.
  ipkgName = "Yaml";
  # Idris dependencies to provide for the build
  idrisDeps = [ contrib lightyear ];

  src = fetchFromGitHub {
    owner = "Heather";
    repo = "Idris.Yaml";
    rev = "5afa51ffc839844862b8316faba3bafa15656db4";
    sha256 = "1g4pi0swmg214kndj85hj50ccmckni7piprsxfdzdfhg87s0avw7";

  meta = with lib; {
    description = "Idris YAML lib";
    homepage = "";
    license =;
    maintainers = [ maintainers.brainrape ];

Assuming this file is saved as yaml.nix, it’s buildable using

$ nix-build -E '(import <nixpkgs> {}).idrisPackages.callPackage ./yaml.nix {}'

Or it’s possible to use

with import <nixpkgs> {};

  yaml = idrisPackages.callPackage ./yaml.nix {};

in another file (say default.nix) to be able to build it with

$ nix-build -A yaml

15.14.4. Passing options to idris commands

The build-idris-package function provides also optional input values to set additional options for the used idris commands.

Specifically, you can set idrisBuildOptions, idrisTestOptions, idrisInstallOptions and idrisDocOptions to provide additional options to the idris command respectively when building, testing, installing and generating docs for your package.

For example you could set

build-idris-package {
  idrisBuildOptions = [ "--log" "1" "--verbose" ]


to require verbose output during idris build phase.

15.15. iOS

This component is basically a wrapper/workaround that makes it possible to expose an Xcode installation as a Nix package by means of symlinking to the relevant executables on the host system.

Since Xcode can’t be packaged with Nix, nor we can publish it as a Nix package (because of its license) this is basically the only integration strategy making it possible to do iOS application builds that integrate with other components of the Nix ecosystem

The primary objective of this project is to use the Nix expression language to specify how iOS apps can be built from source code, and to automatically spawn iOS simulator instances for testing.

This component also makes it possible to use Hydra, the Nix-based continuous integration server to regularly build iOS apps and to do wireless ad-hoc installations of enterprise IPAs on iOS devices through Hydra.

The Xcode build environment implements a number of features.

15.15.1. Deploying a proxy component wrapper exposing Xcode

The first use case is deploying a Nix package that provides symlinks to the Xcode installation on the host system. This package can be used as a build input to any build function implemented in the Nix expression language that requires Xcode.

  pkgs = import <nixpkgs> {};

  xcodeenv = import ./xcodeenv {
    inherit (pkgs) stdenv;
xcodeenv.composeXcodeWrapper {
  version = "9.2";
  xcodeBaseDir = "/Applications/";

By deploying the above expression with nix-build and inspecting its content you will notice that several Xcode-related executables are exposed as a Nix package:

$ ls result/bin
lrwxr-xr-x  1 sander  staff  94  1 jan  1970 Simulator -> /Applications/
lrwxr-xr-x  1 sander  staff  17  1 jan  1970 codesign -> /usr/bin/codesign
lrwxr-xr-x  1 sander  staff  17  1 jan  1970 security -> /usr/bin/security
lrwxr-xr-x  1 sander  staff  21  1 jan  1970 xcode-select -> /usr/bin/xcode-select
lrwxr-xr-x  1 sander  staff  61  1 jan  1970 xcodebuild -> /Applications/
lrwxr-xr-x  1 sander  staff  14  1 jan  1970 xcrun -> /usr/bin/xcrun

15.15.2. Building an iOS application

We can build an iOS app executable for the simulator, or an IPA/xcarchive file for release purposes, e.g. ad-hoc, enterprise or store installations, by executing the xcodeenv.buildApp {} function:

  pkgs = import <nixpkgs> {};

  xcodeenv = import ./xcodeenv {
    inherit (pkgs) stdenv;
xcodeenv.buildApp {
  name = "MyApp";
  src = ./myappsources;
  sdkVersion = "11.2";

  target = null; # Corresponds to the name of the app by default
  configuration = null; # Release for release builds, Debug for debug builds
  scheme = null; # -scheme will correspond to the app name by default
  sdk = null; # null will set it to 'iphonesimulator` for simulator builds or `iphoneos` to real builds
  xcodeFlags = "";

  release = true;
  certificateFile = ./mycertificate.p12;
  certificatePassword = "secret";
  provisioningProfile = ./myprovisioning.profile;
  signMethod = "ad-hoc"; # 'enterprise' or 'store'
  generateIPA = true;
  generateXCArchive = false;

  enableWirelessDistribution = true;
  installURL = "/installipa.php";
  bundleId = "mycompany.myapp";
  appVersion = "1.0";

  # Supports all xcodewrapper parameters as well
  xcodeBaseDir = "/Applications/";

The above function takes a variety of parameters:

  • The name and src parameters are mandatory and specify the name of the app and the location where the source code resides

  • sdkVersion specifies which version of the iOS SDK to use.

It also possile to adjust the xcodebuild parameters. This is only needed in rare circumstances. In most cases the default values should suffice:

  • Specifies which xcodebuild target to build. By default it takes the target that has the same name as the app.

  • The configuration parameter can be overridden if desired. By default, it will do a debug build for the simulator and a release build for real devices.

  • The scheme parameter specifies which -scheme parameter to propagate to xcodebuild. By default, it corresponds to the app name.

  • The sdk parameter specifies which SDK to use. By default, it picks iphonesimulator for simulator builds and iphoneos for release builds.

  • The xcodeFlags parameter specifies arbitrary command line parameters that should be propagated to xcodebuild.

By default, builds are carried out for the iOS simulator. To do release builds (builds for real iOS devices), you must set the release parameter to true. In addition, you need to set the following parameters:

  • certificateFile refers to a P12 certificate file.

  • certificatePassword specifies the password of the P12 certificate.

  • provisioningProfile refers to the provision profile needed to sign the app

  • signMethod should refer to ad-hoc for signing the app with an ad-hoc certificate, enterprise for enterprise certificates and app-store for App store certificates.

  • generateIPA specifies that we want to produce an IPA file (this is probably what you want)

  • generateXCArchive specifies thet we want to produce an xcarchive file.

When building IPA files on Hydra and when it is desired to allow iOS devices to install IPAs by browsing to the Hydra build products page, you can enable the enableWirelessDistribution parameter.

When enabled, you need to configure the following options:

  • The installURL parameter refers to the URL of a PHP script that composes the itms-services:// URL allowing iOS devices to install the IPA file.

  • bundleId refers to the bundle ID value of the app

  • appVersion refers to the app’s version number

To use wireless adhoc distributions, you must also install the corresponding PHP script on a web server (see section: Installing the PHP script for wireless ad hoc installations from Hydra for more information).

In addition to the build parameters, you can also specify any parameters that the xcodeenv.composeXcodeWrapper {} function takes. For example, the xcodeBaseDir parameter can be overridden to refer to a different Xcode version.

15.15.3. Spawning simulator instances

In addition to building iOS apps, we can also automatically spawn simulator instances:

  pkgs = import <nixpkgs> {};

  xcodeenv = import ./xcodeenv {
    inherit (pkgs) stdenv;
xcode.simulateApp {
  name = "simulate";

  # Supports all xcodewrapper parameters as well
  xcodeBaseDir = "/Applications/";

The above expression produces a script that starts the simulator from the provided Xcode installation. The script can be started as follows:


By default, the script will show an overview of UDID for all available simulator instances and asks you to pick one. You can also provide a UDID as a command-line parameter to launch an instance automatically:

./result/bin/run-test-simulator 5C93129D-CF39-4B1A-955F-15180C3BD4B8

You can also extend the simulator script to automatically deploy and launch an app in the requested simulator instance:

  pkgs = import <nixpkgs> {};

  xcodeenv = import ./xcodeenv {
    inherit (pkgs) stdenv;
xcode.simulateApp {
  name = "simulate";
  bundleId = "mycompany.myapp";
  app = xcode.buildApp {
    # ...

  # Supports all xcodewrapper parameters as well
  xcodeBaseDir = "/Applications/";

By providing the result of an xcode.buildApp {} function and configuring the app bundle id, the app gets deployed automatically and started.

15.15.4. Troubleshooting

In some rare cases, it may happen that after a failure, changes are not picked up. Most likely, this is caused by a derived data cache that Xcode maintains. To wipe it you can run:

$ rm -rf ~/Library/Developer/Xcode/DerivedData

15.16. Java

Ant-based Java packages are typically built from source as follows:

stdenv.mkDerivation {
  name = "...";
  src = fetchurl { ... };

  nativeBuildInputs = [ jdk ant ];

  buildPhase = "ant";

Note that jdk is an alias for the OpenJDK (self-built where available, or pre-built via Zulu). Platforms with OpenJDK not (yet) in Nixpkgs (Aarch32, Aarch64) point to the (unfree) oraclejdk.

JAR files that are intended to be used by other packages should be installed in $out/share/java. JDKs have a stdenv setup hook that add any JARs in the share/java directories of the build inputs to the CLASSPATH environment variable. For instance, if the package libfoo installs a JAR named foo.jar in its share/java directory, and another package declares the attribute

buildInputs = [ libfoo ];
nativeBuildInputs = [ jdk ];

then CLASSPATH will be set to /nix/store/...-libfoo/share/java/foo.jar.

Private JARs should be installed in a location like $out/share/package-name.

If your Java package provides a program, you need to generate a wrapper script to run it using a JRE. You can use makeWrapper for this:

nativeBuildInputs = [ makeWrapper ];

installPhase = ''
  mkdir -p $out/bin
  makeWrapper ${jre}/bin/java $out/bin/foo \
    --add-flags "-cp $out/share/java/foo.jar"

Since the introduction of the Java Platform Module System in Java 9, Java distributions typically no longer ship with a general-purpose JRE: instead, they allow generating a JRE with only the modules required for your application(s). Because we can’t predict what modules will be needed on a general-purpose system, the default jre package is the full JDK. When building a minimal system/image, you can override the modules parameter on jre_minimal to build a JRE with only the modules relevant for you:

  my_jre = pkgs.jre_minimal.override {
    modules = [
      # The modules used by 'something' and 'other' combined:
  something = (pkgs.something.override { jre = my_jre; });
  other = (pkgs.other.override { jre = my_jre; });

You can also specify what JDK your JRE should be based on, for example selecting a headless build to avoid including a link to GTK+:

my_jre = pkgs.jre_minimal.override {
  jdk = jdk11_headless;

Note all JDKs passthru home, so if your application requires environment variables like JAVA_HOME being set, that can be done in a generic fashion with the --set argument of makeWrapper:

--set JAVA_HOME ${jdk.home}

It is possible to use a different Java compiler than javac from the OpenJDK. For instance, to use the GNU Java Compiler:

nativeBuildInputs = [ gcj ant ];

Here, Ant will automatically use gij (the GNU Java Runtime) instead of the OpenJRE.

15.17. Javascript

15.17.1. Introduction

This contains instructions on how to package javascript applications.

The various tools available will be listed in the tools-overview. Some general principles for packaging will follow. Finally some tool specific instructions will be given.

15.17.2. Getting unstuck / finding code examples

If you find you are lacking inspiration for packing javascript applications, the links below might prove useful. Searching online for prior art can be helpful if you are running into solved problems. Github Gitlab

15.17.3. Tools overview

15.17.4. General principles

The following principles are given in order of importance with potential exceptions. Try to use the same node version used upstream

It is often not documented which node version is used upstream, but if it is, try to use the same version when packaging.

This can be a problem if upstream is using the latest and greatest and you are trying to use an earlier version of node. Some cryptic errors regarding V8 may appear.

An exception to this: Try to respect the package manager originally used by upstream (and use the upstream lock file)

A lock file (package-lock.json, yarn.lock…) is supposed to make reproducible installations of node_modules for each tool.

Guidelines of package managers, recommend to commit those lock files to the repos. If a particular lock file is present, it is a strong indication of which package manager is used upstream.

It’s better to try to use a nix tool that understand the lock file. Using a different tool might give you hard to understand error because different packages have been installed. An example of problems that could arise can be found here. Upstream uses npm, but this is an attempt to package it with yarn2nix (that uses yarn.lock)

Using a different tool forces to commit a lock file to the repository. Those files are fairly large, so when packaging for nixpkgs, this approach does not scale well.

Exceptions to this rule are:

  • when you encounter one of the bugs from a nix tool. In each of the tool specific instructions, known problems will be detailed. If you have a problem with a particular tool, then it’s best to try another tool, even if this means you will have to recreate a lock file and commit it to nixpkgs. In general yarn2nix has less known problems and so a simple search in nixpkgs will reveal many yarn.lock files committed

  • Some lock files contain particular version of a package that has been pulled off npm for some reason. In that case, you can recreate upstream lock (by removing the original and npm install, yarn, …) and commit this to nixpkgs.

  • The only tool that supports workspaces (a feature of npm that helps manage sub-directories with different package.json from a single top level package.json) is yarn2nix. If upstream has workspaces you should try yarn2nix. Try to use upstream package.json

Exceptions to this rule are

  • Sometimes the upstream repo assumes some dependencies be installed globally. In that case you can add them manually to the upstream package.json (yarn add xxx or npm install xxx, …). Dependencies that are installed locally can be executed with npx for cli tools. (e.g. npx postcss ..., this is how you can call those dependencies in the phases).

  • Sometimes there is a version conflict between some dependency requirements. In that case you can fix a version (by removing the ^).

  • Sometimes the script defined in the package.json does not work as is. Some scripts for example use cli tools that might not be available, or cd in directory with a different package.json (for workspaces notably). In that case, it’s perfectly fine to look at what the particular script is doing and break this down in the phases. In the build script you can see build:* calling in turns several other build scripts like build:ui or build:server. If one of those fails, you can try to separate those into:

yarn build:ui
yarn build:server
# OR
npm run build:ui
npm run build:server

when you need to override a package.json. It’s nice to use the one from the upstream src and do some explicit override. Here is an example.

patchedPackageJSON = final.runCommand "package.json" { } ''
  ${jq}/bin/jq '.version = "0.4.0" |
    .devDependencies."@jsdoc/cli" = "^0.2.5"
    ${sonar-src}/package.json > $out

you will still need to commit the modified version of the lock files, but at least the overrides are explicit for everyone to see. Using node_modules directly

each tool has an abstraction to just build the node_modules (dependencies) directory. you can always use the stdenv.mkDerivation with the node_modules to build the package (symlink the node_modules directory and then use the package build command). the node_modules abstraction can be also used to build some web framework frontends. For an example of this see how plausible is built. mkYarnModules to make the derivation containing node_modules. Then when building the frontend you can just symlink the node_modules directory

15.17.5. javascript packages inside nixpkgs

The pkgs/development/node-packages folder contains a generated collection of NPM packages that can be installed with the Nix package manager.

As a rule of thumb, the package set should only provide end user software packages, such as command-line utilities. Libraries should only be added to the package set if there is a non-NPM package that requires it.

When it is desired to use NPM libraries in a development project, use the node2nix generator directly on the package.json configuration file of the project.

The package set provides support for the official stable Node.js versions. The latest stable LTS release in nodePackages, as well as the latest stable Current release in nodePackages_latest.

If your package uses native addons, you need to examine what kind of native build system it uses. Here are some examples:

  • node-gyp

  • node-gyp-builder

  • node-pre-gyp

After you have identified the correct system, you need to override your package expression while adding in build system as a build input. For example, dat requires node-gyp-build, so we override its expression in default.nix:

    dat = super.dat.override {
      buildInputs = [ self.node-gyp-build pkgs.libtool pkgs.autoconf pkgs.automake ];
      meta.broken = since "12";

To add a package from NPM to nixpkgs:

  1. Modify pkgs/development/node-packages/node-packages.json to add, update or remove package entries to have it included in nodePackages and nodePackages_latest.

  2. Run the script: cd pkgs/development/node-packages && ./

  3. Build your new package to test your changes: cd /path/to/nixpkgs && nix-build -A nodePackages.<new-or-updated-package>. To build against the latest stable Current Node.js version (e.g. 14.x): nix-build -A nodePackages_latest.<new-or-updated-package>

  4. Add and commit all modified and generated files.

For more information about the generation process, consult the file of the node2nix tool.

15.17.6. Tool specific instructions node2nix Preparation

you will need to generate a nix expression for the dependencies

  • don’t forget the -l package-lock.json if there is a lock file

  • Most probably you will need the --development to include the devDependencies

so the command will most likely be node2nix --development -l package-lock.json

link to the doc in the repo Pitfalls
  • if upstream package.json does not have a version attribute, node2nix will crash. You will need to add it like shown in the package.json section

  • node2nix has some bugs. related to working with lock files from npm distributed with nodejs-16_x

  • node2nix does not like missing packages from npm. If you see something like Cannot resolve version: vue-loader-v16@undefined then you might want to try another tool. The package might have been pulled off of npm. yarn2nix Preparation

you will need at least a yarn.lock and yarn.nix file

  • generate a yarn.lock in upstream if it is not already there

  • yarn2nix > yarn.nix will generate the dependencies in a nix format mkYarnPackage

this will by default try to generate a binary. For package only generating static assets (Svelte, Vue, React…), you will need to explicitly override the build step with your instructions. It’s important to use the --offline flag. For example if you script is "build": "something" in package.json use

buildPhase = ''
  yarn build --offline

The dist phase is also trying to build a binary, the only way to override it is with

distPhase = "true";

the configure phase can sometimes fail because it tries to be too clever. One common override is

configurePhase = "ln -s $node_modules node_modules"; mkYarnModules

this will generate a derivation including the node_modules. If you have to build a derivation for an integrated web framework (rails, phoenix..), this is probably the easiest way. Plausible offers a good example of how to do this. Overriding dependency behavior

In the mkYarnPackage record the property pkgConfig can be used to override packages when you encounter problems building.

For instance, say your package is throwing errors when trying to invoke node-sass: ENOENT: no such file or directory, scandir '/build/source/node_modules/node-sass/vendor'

To fix this we will specify different versions of build inputs to use, as well as some post install steps to get the software built the way we want:

mkYarnPackage rec {
  pkgConfig = {
    node-sass = {
      buildInputs = with final;[ python libsass pkg-config ];
      postInstall = ''
        LIBSASS_EXT=auto yarn --offline run build
        rm build/config.gypi
} Pitfalls
  • if version is missing from upstream package.json, yarn will silently install nothing. In that case, you will need to override package.json as shown in the package.json section

  • having trouble with node-gyp? Try adding these lines to the yarnPreBuild steps:

    yarnPreBuild = ''
      mkdir -p $HOME/.node-gyp/${nodejs.version}
      echo 9 > $HOME/.node-gyp/${nodejs.version}/installVersion
      ln -sfv ${nodejs}/include $HOME/.node-gyp/${nodejs.version}
      export npm_config_nodedir=${nodejs}

15.17.7. Outside of nixpkgs

There are some other options available that can’t be used inside nixpkgs. Those other options are written in nix. Importing them in nixpkgs will require moving the source code into nixpkgs. Using Import From Derivation is not allowed in hydra at present. If you are packaging something outside nixpkgs, those can be considered npmlock2nix

npmlock2nix aims at building node_modules without code generation. It hasn’t reached v1 yet, the api might be subject to change. Pitfalls nix-npm-buildpackage

nix-npm-buildpackage aims at building node_modules without code generation. It hasn’t reached v1 yet, the api might change. It supports both package-lock.json and yarn.lock. Pitfalls

15.18. User’s Guide to Lua Infrastructure

15.18.1. Using Lua Overview of Lua

Several versions of the Lua interpreter are available: luajit, lua 5.1, 5.2, 5.3. The attribute lua refers to the default interpreter, it is also possible to refer to specific versions, e.g. lua5_2 refers to Lua 5.2.

Lua libraries are in separate sets, with one set per interpreter version.

The interpreters have several common attributes. One of these attributes is pkgs, which is a package set of Lua libraries for this specific interpreter. E.g., the busted package corresponding to the default interpreter is lua.pkgs.busted, and the lua 5.2 version is lua5_2.pkgs.busted. The main package set contains aliases to these package sets, e.g. luaPackages refers to lua5_1.pkgs and lua52Packages to lua5_2.pkgs. Installing Lua and packages Lua environment defined in separate .nix file

Create a file, e.g. build.nix, with the following expression

with import <nixpkgs> {};

lua5_2.withPackages (ps: with ps; [ busted luafilesystem ])

and install it in your profile with

nix-env -if build.nix

Now you can use the Lua interpreter, as well as the extra packages (busted, luafilesystem) that you added to the environment. Lua environment defined in ~/.config/nixpkgs/config.nix

If you prefer to, you could also add the environment as a package override to the Nixpkgs set, e.g. using config.nix,

{ # ...

  packageOverrides = pkgs: with pkgs; {
    myLuaEnv = lua5_2.withPackages (ps: with ps; [ busted luafilesystem ]);

and install it in your profile with

nix-env -iA nixpkgs.myLuaEnv

The environment is installed by referring to the attribute, and considering the nixpkgs channel was used. Lua environment defined in /etc/nixos/configuration.nix

For the sake of completeness, here’s another example how to install the environment system-wide.

{ # ...

  environment.systemPackages = with pkgs; [
    (lua.withPackages(ps: with ps; [ busted luafilesystem ]))
} How to override a Lua package using overlays?

Use the following overlay template:

final: prev:

  lua = prev.lua.override {
    packageOverrides = luaself: luaprev: {

      luarocks-nix = luaprev.luarocks-nix.overrideAttrs(oa: {
        pname = "luarocks-nix";
        src = /home/my_luarocks/repository;

  luaPackages = lua.pkgs;
} Temporary Lua environment with nix-shell

There are two methods for loading a shell with Lua packages. The first and recommended method is to create an environment with lua.buildEnv or lua.withPackages and load that. E.g.

$ nix-shell -p 'lua.withPackages(ps: with ps; [ busted luafilesystem ])'

opens a shell from which you can launch the interpreter

[nix-shell:~] lua

The other method, which is not recommended, does not create an environment and requires you to list the packages directly,

$ nix-shell -p lua.pkgs.busted lua.pkgs.luafilesystem

Again, it is possible to launch the interpreter from the shell. The Lua interpreter has the attribute pkgs which contains all Lua libraries for that specific interpreter.

15.18.2. Developing with Lua

Now that you know how to get a working Lua environment with Nix, it is time to go forward and start actually developing with Lua. There are two ways to package lua software, either it is on luarocks and most of it can be taken care of by the luarocks2nix converter or the packaging has to be done manually. Let’s present the luarocks way first and the manual one in a second time. Packaging a library on luarocks is the main repository of lua packages. The site proposes two types of packages, the rockspec and the src.rock (equivalent of a rockspec but with the source). These packages can have different build types such as cmake, builtin etc .

Luarocks-based packages are generated in pkgs/development/lua-modules/generated-packages.nix from the whitelist maintainers/scripts/luarocks-packages.csv and updated by running maintainers/scripts/update-luarocks-packages.

luarocks2nix is a tool capable of generating nix derivations from both rockspec and src.rock (and favors the src.rock). The automation only goes so far though and some packages need to be customized. These customizations go in pkgs/development/lua-modules/overrides.nix. For instance if the rockspec defines external_dependencies, these need to be manually added to the overrides.nix.

You can try converting luarocks packages to nix packages with the command nix-shell -p luarocks-nix and then luarocks nix PKG_NAME. Packaging a library manually

You can develop your package as you usually would, just don’t forget to wrap it within a toLuaModule call, for instance

mynewlib = toLuaModule ( stdenv.mkDerivation { ... });

There is also the buildLuaPackage function that can be used when lua modules are not packaged for luarocks. You can see a few examples at pkgs/top-level/lua-packages.nix.

15.18.3. Lua Reference Lua interpreters

Versions 5.1, 5.2, 5.3 and 5.4 of the lua interpreter are available as respectively lua5_1, lua5_2, lua5_3 and lua5_4. Luajit is available too. The Nix expressions for the interpreters can be found in pkgs/development/interpreters/lua-5. Attributes on lua interpreters packages

Each interpreter has the following attributes:

  • interpreter. Alias for ${pkgs.lua}/bin/lua.

  • buildEnv. Function to build lua interpreter environments with extra packages bundled together. See section lua.buildEnv function for usage and documentation.

  • withPackages. Simpler interface to buildEnv.

  • pkgs. Set of Lua packages for that specific interpreter. The package set can be modified by overriding the interpreter and passing packageOverrides. buildLuarocksPackage function

The buildLuarocksPackage function is implemented in pkgs/development/interpreters/lua-5/build-lua-package.nix The following is an example:

luaposix = buildLuarocksPackage {
  pname = "luaposix";
  version = "34.0.4-1";

  src = fetchurl {
    url    = "";
    sha256 = "0yrm5cn2iyd0zjd4liyj27srphvy0gjrjx572swar6zqr4dwjqp2";
  disabled = (luaOlder "5.1") || (luaAtLeast "5.4");
  propagatedBuildInputs = [ bit32 lua std_normalize ];

  meta = with lib; {
    homepage = "";
    description = "Lua bindings for POSIX";
    maintainers = with maintainers; [ vyp lblasc ];
    license.fullName = "MIT/X11";

The buildLuarocksPackage delegates most tasks to luarocks:

  • it adds luarocks as an unpacker for src.rock files (zip files really).

  • configurePhasewrites a temporary luarocks configuration file which location is exported via the environment variableLUAROCKS_CONFIG`.

  • the buildPhase does nothing.

  • installPhase calls luarocks make --deps-mode=none --tree $out to build and install the package

  • In the postFixup phase, the wrapLuaPrograms bash function is called to wrap all programs in the $out/bin/* directory to include $PATH environment variable and add dependent libraries to script’s LUA_PATH and LUA_CPATH.

By default meta.platforms is set to the same value as the interpreter unless overridden otherwise. buildLuaApplication function

The buildLuaApplication function is practically the same as buildLuaPackage. The difference is that buildLuaPackage by default prefixes the names of the packages with the version of the interpreter. Because with an application we’re not interested in multiple version the prefix is dropped. lua.withPackages function

The lua.withPackages takes a function as an argument that is passed the set of lua packages and returns the list of packages to be included in the environment. Using the withPackages function, the previous example for the luafilesystem environment can be written like this:

with import <nixpkgs> {};

lua.withPackages (ps: [ps.luafilesystem])

withPackages passes the correct package set for the specific interpreter version as an argument to the function. In the above example, ps equals luaPackages. But you can also easily switch to using lua5_2:

with import <nixpkgs> {};

lua5_2.withPackages (ps: [ps.lua])

Now, ps is set to lua52Packages, matching the version of the interpreter. Possible Todos

  • export/use version specific variables such as LUA_PATH_5_2/LUAROCKS_CONFIG_5_2

  • let luarocks check for dependencies via exporting the different rocktrees in temporary config Lua Contributing guidelines

Following rules should be respected:

  • Make sure libraries build for all Lua interpreters.

  • Commit names of Lua libraries should reflect that they are Lua libraries, so write for example luaPackages.luafilesystem: 1.11 -> 1.12.

15.19. Maven

Maven is a well-known build tool for the Java ecosystem however it has some challenges when integrating into the Nix build system.

The following provides a list of common patterns with how to package a Maven project (or any JVM language that can export to Maven) as a Nix package.

For the purposes of this example let’s consider a very basic Maven project with the following pom.xml with a single dependency on emoji-java.

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="" xmlns:xsi=""
  <name>NixOS Maven Demo</name>


Our main class file will be very simple:

import com.vdurmont.emoji.EmojiParser;

public class Main {
  public static void main(String[] args) {
    String str = "NixOS :grinning: is super cool :smiley:!";
    String result = EmojiParser.parseToUnicode(str);

You find this demo project at

15.19.1. Solving for dependencies buildMaven with NixOS/mvn2nix-maven-plugin

⚠️ Although buildMaven is the blessed way within nixpkgs, as of 2020, it hasn’t seen much activity in quite a while.

buildMaven is an alternative method that tries to follow similar patterns of other programming languages by generating a lock file. It relies on the maven plugin mvn2nix-maven-plugin.

First you generate a project-info.json file using the maven plugin.

This should be executed in the project’s source repository or be told which pom.xml to execute with.

# run this step within the project's source repository
❯ mvn org.nixos.mvn2nix:mvn2nix-maven-plugin:mvn2nix

❯ cat project-info.json | jq | head
  "project": {
    "artifactId": "maven-demo",
    "groupId": "org.nixos",
    "version": "1.0",
    "classifier": "",
    "extension": "jar",
    "dependencies": [
        "artifactId": "maven-resources-plugin",

This file is then given to the buildMaven function, and it returns 2 attributes.

repo: A Maven repository that is a symlink farm of all the dependencies found in the project-info.json

build: A simple derivation that runs through mvn compile & mvn package to build the JAR. You may use this as inspiration for more complicated derivations.

Here is an example of building the Maven repository

{ pkgs ? import <nixpkgs> { } }:
with pkgs;
(buildMaven ./project-info.json).repo

The benefit over the double invocation as we will see below, is that the /nix/store entry is a linkFarm of every package, so that changes to your dependency set doesn’t involve downloading everything from scratch.

❯ tree $(nix-build --no-out-link build-maven-repository.nix) | head
├── antlr
│   └── antlr
│       └── 2.7.2
│           ├── antlr-2.7.2.jar -> /nix/store/d027c8f2cnmj5yrynpbq2s6wmc9cb559-antlr-2.7.2.jar
│           └── antlr-2.7.2.pom -> /nix/store/mv42fc5gizl8h5g5vpywz1nfiynmzgp2-antlr-2.7.2.pom
├── avalon-framework
│   └── avalon-framework
│       └── 4.1.3
│           ├── avalon-framework-4.1.3.jar -> /nix/store/iv5fp3955w3nq28ff9xfz86wvxbiw6n9-avalon-framework-4.1.3.jar Double Invocation

⚠️ This pattern is the simplest but may cause unnecessary rebuilds due to the output hash changing.

The double invocation is a simple way to get around the problem that nix-build may be sandboxed and have no Internet connectivity.

It treats the entire Maven repository as a single source to be downloaded, relying on Maven’s dependency resolution to satisfy the output hash. This is similar to fetchers like fetchgit, except it has to run a Maven build to determine what to download.

The first step will be to build the Maven project as a fixed-output derivation in order to collect the Maven repository – below is an example.

Traditionally the Maven repository is at ~/.m2/repository. We will override this to be the $out directory.

{ lib, stdenv, maven }:
stdenv.mkDerivation {
  name = "maven-repository";
  buildInputs = [ maven ];
  src = ./.; # or fetchFromGitHub, cleanSourceWith, etc
  buildPhase = ''
    mvn package -Dmaven.repo.local=$out

  # keep only *.{pom,jar,sha1,nbm} and delete all ephemeral files with lastModified timestamps inside
  installPhase = ''
    find $out -type f \
      -name \*.lastUpdated -or \
      -name -or \
      -name _remote.repositories \

  # don't do any fixup
  dontFixup = true;
  outputHashAlgo = "sha256";
  outputHashMode = "recursive";
  # replace this with the correct SHA256
  outputHash = lib.fakeSha256;

The build will fail, and tell you the expected outputHash to place. When you’ve set the hash, the build will return with a /nix/store entry whose contents are the full Maven repository.

Some additional files are deleted that would cause the output hash to change potentially on subsequent runs.

❯ tree $(nix-build --no-out-link double-invocation-repository.nix) | head
├── backport-util-concurrent
│   └── backport-util-concurrent
│       └── 3.1
│           ├── backport-util-concurrent-3.1.pom
│           └── backport-util-concurrent-3.1.pom.sha1
├── classworlds
│   └── classworlds
│       ├── 1.1
│       │   ├── classworlds-1.1.jar

If your package uses SNAPSHOT dependencies or version ranges; there is a strong likelihood that over-time your output hash will change since the resolved dependencies may change. Hence this method is less recommended then using buildMaven.

15.19.2. Building a JAR

Regardless of which strategy is chosen above, the step to build the derivation is the same.

{ stdenv, maven, callPackage }:
# pick a repository derivation, here we will use buildMaven
let repository = callPackage ./build-maven-repository.nix { };
in stdenv.mkDerivation rec {
  pname = "maven-demo";
  version = "1.0";

  src = builtins.fetchTarball "";
  buildInputs = [ maven ];

  buildPhase = ''
    echo "Using repository ${repository}"
    mvn --offline -Dmaven.repo.local=${repository} package;

  installPhase = ''
    install -Dm644 target/${pname}-${version}.jar $out/share/java

We place the library in $out/share/java since JDK package has a stdenv setup hook that adds any JARs in the share/java directories of the build inputs to the CLASSPATH environment.

❯ tree $(nix-build --no-out-link build-jar.nix)
└── share
    └── java
        └── maven-demo-1.0.jar

2 directories, 1 file

15.19.3. Runnable JAR

The previous example builds a jar file but that’s not a file one can run.

You need to use it with java -jar $out/share/java/output.jar and make sure to provide the required dependencies on the classpath.

The following explains how to use makeWrapper in order to make the derivation produce an executable that will run the JAR file you created.

We will use the same repository we built above (either double invocation or buildMaven) to setup a CLASSPATH for our JAR.

The following two methods are more suited to Nix then building an UberJar which may be the more traditional approach. CLASSPATH

This is ideal if you are providing a derivation for nixpkgs and don’t want to patch the project’s pom.xml.

We will read the Maven repository and flatten it to a single list. This list will then be concatenated with the CLASSPATH separator to create the full classpath.

We make sure to provide this classpath to the makeWrapper.

{ stdenv, maven, callPackage, makeWrapper, jre }:
  repository = callPackage ./build-maven-repository.nix { };
in stdenv.mkDerivation rec {
  pname = "maven-demo";
  version = "1.0";

  src = builtins.fetchTarball
  buildInputs = [ maven makeWrapper ];

  buildPhase = ''
    echo "Using repository ${repository}"
    mvn --offline -Dmaven.repo.local=${repository} package;

  installPhase = ''
    mkdir -p $out/bin

    classpath=$(find ${repository} -name "*.jar" -printf ':%h/%f');
    install -Dm644 target/${pname}-${version}.jar $out/share/java
    # create a wrapper that will automatically set the classpath
    # this should be the paths from the dependency derivation
    makeWrapper ${jre}/bin/java $out/bin/${pname} \
          --add-flags "-classpath $out/share/java/${pname}-${version}.jar:''${classpath#:}" \
          --add-flags "Main"
} MANIFEST file via Maven Plugin

This is ideal if you are the project owner and want to change your pom.xml to set the CLASSPATH within it.

Augment the pom.xml to create a JAR with the following manifest:


The above plugin instructs the JAR to look for the necessary dependencies in the lib/ relative folder. The layout of the folder is also in the maven repository style.

❯ unzip -q -c $(nix-build --no-out-link runnable-jar.nix)/share/java/maven-demo-1.0.jar META-INF/MANIFEST.MF

Manifest-Version: 1.0
Archiver-Version: Plexus Archiver
Built-By: nixbld
Class-Path: . ../../repository/com/vdurmont/emoji-java/5.1.1/emoji-jav
 a-5.1.1.jar ../../repository/org/json/json/20170516/json-20170516.jar
Created-By: Apache Maven 3.6.3
Build-Jdk: 1.8.0_265
Main-Class: Main

We will modify the derivation above to add a symlink to our repository so that it’s accessible to our JAR during the installPhase.

{ stdenv, maven, callPackage, makeWrapper, jre }:
# pick a repository derivation, here we will use buildMaven
let repository = callPackage ./build-maven-repository.nix { };
in stdenv.mkDerivation rec {
  pname = "maven-demo";
  version = "1.0";

  src = builtins.fetchTarball
  buildInputs = [ maven makeWrapper ];

  buildPhase = ''
    echo "Using repository ${repository}"
    mvn --offline -Dmaven.repo.local=${repository} package;

  installPhase = ''
    mkdir -p $out/bin

    # create a symbolic link for the repository directory
    ln -s ${repository} $out/repository

    install -Dm644 target/${pname}-${version}.jar $out/share/java
    # create a wrapper that will automatically set the classpath
    # this should be the paths from the dependency derivation
    makeWrapper ${jre}/bin/java $out/bin/${pname} \
          --add-flags "-jar $out/share/java/${pname}-${version}.jar"

Our script produces a dependency on jre rather than jdk to restrict the runtime closure necessary to run the application.

This will give you an executable shell-script that launches your JAR with all the dependencies available.

❯ tree $(nix-build --no-out-link runnable-jar.nix)
├── bin
│   └── maven-demo
├── repository -> /nix/store/g87va52nkc8jzbmi1aqdcf2f109r4dvn-maven-repository
└── share
    └── java
        └── maven-demo-1.0.jar

❯ $(nix-build --no-out-link --option tarball-ttl 1 runnable-jar.nix)/bin/maven-demo
NixOS 😀 is super cool 😃!

15.20. Nim

15.20.1. Overview

The Nim compiler, a builder function, and some packaged libraries are available in Nixpkgs. Until now each compiler release has been effectively backwards compatible so only the latest version is available.

15.20.2. Nim program packages in Nixpkgs

Nim programs can be built using nimPackages.buildNimPackage. In the case of packages not containing exported library code the attribute nimBinOnly should be set to true.

The following example shows a Nim program that depends only on Nim libraries:

{ lib, nimPackages, fetchurl }:

nimPackages.buildNimPackage rec {
  pname = "hottext";
  version = "1.4";

  nimBinOnly = true;

  src = fetchurl {
    url = "${version}.tar.gz";
    sha256 = "sha256-hIUofi81zowSMbt1lUsxCnVzfJGN3FEiTtN8CEFpwzY=";

  buildInputs = with nimPackages; [

15.20.3. Nim library packages in Nixpkgs

Nim libraries can also be built using nimPackages.buildNimPackage, but often the product of a fetcher is sufficient to satisfy a dependency. The fetchgit, fetchFromGitHub, and fetchNimble functions yield an output that can be discovered during the configurePhase of buildNimPackage.

Nim library packages are listed in pkgs/top-level/nim-packages.nix and implemented at pkgs/development/nim-packages.

The following example shows a Nim library that propagates a dependency on a non-Nim package:

{ lib, buildNimPackage, fetchNimble, SDL2 }:

buildNimPackage rec {
  pname = "sdl2";
  version = "2.0.4";
  src = fetchNimble {
    inherit pname version;
    hash = "sha256-Vtcj8goI4zZPQs2TbFoBFlcR5UqDtOldaXSH/+/xULk=";
  propagatedBuildInputs = [ SDL2 ];

15.20.4. buildNimPackage parameters

All parameters from stdenv.mkDerivation function are still supported. The following are specific to buildNimPackage:

  • nimBinOnly ? false: If true then build only the programs listed in the Nimble file in the packages sources.

  • nimbleFile: Specify the Nimble file location of the package being built rather than discover the file at build-time.

  • nimRelease ? true: Build the package in release mode.

  • nimDefines ? []: A list of Nim defines. Key-value tuples are not supported.

  • nimFlags ? []: A list of command line arguments to pass to the Nim compiler. Use this to specify defines with arguments in the form of -d:${name}=${value}.

  • nimDoc ? false`: Build and install HTML documentation.

  • buildInputs ? []: The packages listed here will be searched for *.nimble files which are used to populate the Nim library path. Otherwise the standard behavior is in effect.

15.21. OCaml

15.21.1. User guide

OCaml libraries are available in attribute sets of the form ocaml-ng.ocamlPackages_X_XX where X is to be replaced with the desired compiler version. For example, ocamlgraph compiled with OCaml 4.12 can be found in ocaml-ng.ocamlPackages_4_12.ocamlgraph. The compiler itself is also located in this set, under the name ocaml.

If you don’t care about the exact compiler version, ocamlPackages is a top-level alias pointing to a recent version of OCaml.

OCaml applications are usually available top-level, and not inside ocamlPackages. Notable exceptions are build tools that must be built with the same compiler version as the compiler you intend to use like dune or ocaml-lsp.

To open a shell able to build a typical OCaml project, put the dependencies in buildInputs and add ocamlPackages.ocaml and ocamlPackages.findlib to nativeBuildInputs at least. For example:

 pkgs = import <nixpkgs> {};
 # choose the ocaml version you want to use
 ocamlPackages = pkgs.ocaml-ng.ocamlPackages_4_12;
pkgs.mkShell {
  # build tools
  nativeBuildInputs = with ocamlPackages; [ ocaml findlib dune_2 ocaml-lsp ];
  # dependencies
  buildInputs = with ocamlPackages; [ ocamlgraph ];

15.21.2. Packaging guide

OCaml libraries should be installed in $(out)/lib/ocaml/${ocaml.version}/site-lib/. Such directories are automatically added to the $OCAMLPATH environment variable when building another package that depends on them or when opening a nix-shell.

Given that most of the OCaml ecosystem is now built with dune, nixpkgs includes a convenience build support function called buildDunePackage that will build an OCaml package using dune, OCaml and findlib and any additional dependencies provided as buildInputs or propagatedBuildInputs.

Here is a simple package example.

  • It defines an (optional) attribute minimalOCamlVersion that will be used to throw a descriptive evaluation error if building with an older OCaml is attempted.

  • It uses the fetchFromGitHub fetcher to get its source.

  • useDune2 = true ensures that the latest version of Dune is used for the build (this may become the default value in a future release).

  • It sets the optional doCheck attribute such that tests will be run with dune runtest -p angstrom after the build (dune build -p angstrom) is complete, but only if the Ocaml version is at at least "4.05".

  • It uses the package ocaml-syntax-shims as a build input, alcotest and ppx_let as check inputs (because they are needed to run the tests), and bigstringaf and result as propagated build inputs (thus they will also be available to libraries depending on this library).

  • The library will be installed using the angstrom.install file that dune generates.

{ lib,
  ppx_let }:

buildDunePackage rec {
  pname = "angstrom";
  version = "0.15.0";
  useDune2 = true;

  minimalOCamlVersion = "4.04";

  src = fetchFromGitHub {
    owner  = "inhabitedtype";
    repo   = pname;
    rev    = version;
    sha256 = "1hmrkdcdlkwy7rxhngf3cv3sa61cznnd9p5lmqhx20664gx2ibrh";

  checkInputs = [ alcotest ppx_let ];
  buildInputs = [ ocaml-syntax-shims ];
  propagatedBuildInputs = [ bigstringaf result ];
  doCheck = lib.versionAtLeast ocaml.version "4.05";

  meta = {
    homepage = "";
    description = "OCaml parser combinators built for speed and memory efficiency";
    license = lib.licenses.bsd3;
    maintainers = with lib.maintainers; [ sternenseemann ];

Here is a second example, this time using a source archive generated with dune-release. It is a good idea to use this archive when it is available as it will usually contain substituted variables such as a %%VERSION%% field. This library does not depend on any other OCaml library and no tests are run after building it.

{ lib, fetchurl, buildDunePackage }:

buildDunePackage rec {
  pname = "wtf8";
  version = "1.0.2";

  useDune2 = true;

  minimalOCamlVersion = "4.02";

  src = fetchurl {
    url = "${pname}/releases/download/v${version}/${pname}-v${version}.tbz";
    sha256 = "09ygcxxd5warkdzz17rgpidrd0pg14cy2svvnvy1hna080lzg7vp";

  meta = with lib; {
    homepage = "";
    description = "WTF-8 is a superset of UTF-8 that allows unpaired surrogates.";
    license =;
    maintainers = [ maintainers.eqyiel ];

15.22. Octave

15.22.1. Introduction

Octave is a modular scientific programming language and environment. A majority of the packages supported by Octave from their website are packaged in nixpkgs.

15.22.2. Structure

All Octave add-on packages are available in two ways:

  1. Under the top-level Octave attribute, octave.pkgs.

  2. As a top-level attribute, octavePackages.

15.22.3. Packaging Octave Packages

Nixpkgs provides a function buildOctavePackage, a generic package builder function for any Octave package that complies with the Octave’s current packaging format.

All Octave packages are defined in pkgs/top-level/octave-packages.nix rather than pkgs/all-packages.nix. Each package is defined in their own file in the pkgs/development/octave-modules directory. Octave packages are made available through all-packages.nix through both the attribute octavePackages and octave.pkgs. You can test building an Octave package as follows:

$ nix-build -A octavePackages.symbolic

When building Octave packages with nix-build, the buildOctavePackage function adds octave-octaveVersion to; the start of the package’s name attribute.

This can be required when installing the package using nix-env:

$ nix-env -i octave-6.2.0-symbolic

Although, you can also install it using the attribute name:

$ nix-env -i -A octavePackages.symbolic

You can build Octave with packages by using the withPackages passed-through function.

$ nix-shell -p 'octave.withPackages (ps: with ps; [ symbolic ])'

This will also work in a shell.nix file.

{ pkgs ? import <nixpkgs> { }}:

pkgs.mkShell {
  nativeBuildInputs = with pkgs; [
    (octave.withPackages (opkgs: with opkgs; [ symbolic ]))
} buildOctavePackage Steps

The buildOctavePackage does several things to make sure things work properly.

  1. Sets the environment variable OCTAVE_HISTFILE to /dev/null during package compilation so that the commands run through the Octave interpreter directly are not logged.

  2. Skips the configuration step, because the packages are stored as gzipped tarballs, which Octave itself handles directly.

  3. Change the hierarchy of the tarball so that only a single directory is at the top-most level of the tarball.

  4. Use Octave itself to run the pkg build command, which unzips the tarball, extracts the necessary files written in Octave, and compiles any code written in C++ or Fortran, and places the fully compiled artifact in $out.

buildOctavePackage is built on top of stdenv in a standard way, allowing most things to be customized. Handling Dependencies

In Octave packages, there are four sets of dependencies that can be specified:


Just like other packages, nativeBuildInputs is intended for architecture-dependent build-time-only dependencies.


Like other packages, buildInputs is intended for architecture-independent build-time-only dependencies.


Similar to other packages, propagatedBuildInputs is intended for packages that are required for both building and running of the package. See Symbolic for how this works and why it is needed.


This is a special dependency that ensures the specified Octave packages are dependent on others, and are made available simultaneously when loading them in Octave. Installing Octave Packages

By default, the buildOctavePackage function does not install the requested package into Octave for use. The function will only build the requested package. This is due to Octave maintaining an text-based database about which packages are installed where. To this end, when all the requested packages have been built, the Octave package and all its add-on packages are put together into an environment, similar to Python.

  1. First, all the Octave binaries are wrapped with the environment variable OCTAVE_SITE_INITFILE set to a file in $out, which is required for Octave to be able to find the non-standard package database location.

  2. Because of the way buildEnv works, all tarballs that are present (which should be all Octave packages to install) should be removed.

  3. The path down to the default install location of Octave packages is recreated so that Nix-operated Octave can install the packages.

  4. Install the packages into the $out environment while writing package entries to the database file. This database file is unique for each different (according to Nix) environment invocation.

  5. Rewrite the Octave-wide startup file to read from the list of packages installed in that particular environment.

  6. Wrap any programs that are required by the Octave packages so that they work with all the paths defined within the environment.

15.23. Perl

15.23.1. Running perl programs on the shell

When executing a Perl script, it is possible you get an error such as ./ bad interpreter: /usr/bin/perl: no such file or directory. This happens when the script expects Perl to be installed at /usr/bin/perl, which is not the case when using Perl from nixpkgs. You can fix the script by changing the first line to:

#!/usr/bin/env perl

to take the Perl installation from the PATH environment variable, or invoke Perl directly with:

$ perl ./

When the script is using a Perl library that is not installed globally, you might get an error such as Can't locate in @INC (you may need to install the DB_File module). In that case, you can use nix-shell to start an ad-hoc shell with that library installed, for instance:

$ nix-shell -p perl perlPackages.DBFile --run ./

If you are always using the script in places where nix-shell is available, you can embed the nix-shell invocation in the shebang like this:

#!/usr/bin/env nix-shell
#! nix-shell -i perl -p perl perlPackages.DBFile

15.23.2. Packaging Perl programs

Nixpkgs provides a function buildPerlPackage, a generic package builder function for any Perl package that has a standard Makefile.PL. It’s implemented in pkgs/development/perl-modules/generic.

Perl packages from CPAN are defined in pkgs/top-level/perl-packages.nix rather than pkgs/all-packages.nix. Most Perl packages are so straight-forward to build that they are defined here directly, rather than having a separate function for each package called from perl-packages.nix. However, more complicated packages should be put in a separate file, typically in pkgs/development/perl-modules. Here is an example of the former:

ClassC3 = buildPerlPackage rec {
  name = "Class-C3-0.21";
  src = fetchurl {
    url = "mirror://cpan/authors/id/F/FL/FLORA/${name}.tar.gz";
    sha256 = "1bl8z095y4js66pwxnm7s853pi9czala4sqc743fdlnk27kq94gz";

Note the use of mirror://cpan/, and the ${name} in the URL definition to ensure that the name attribute is consistent with the source that we’re actually downloading. Perl packages are made available in all-packages.nix through the variable perlPackages. For instance, if you have a package that needs ClassC3, you would typically write

foo = import ../path/to/foo.nix {
  inherit stdenv fetchurl ...;
  inherit (perlPackages) ClassC3;

in all-packages.nix. You can test building a Perl package as follows:

$ nix-build -A perlPackages.ClassC3

buildPerlPackage adds perl- to the start of the name attribute, so the package above is actually called perl-Class-C3-0.21. So to install it, you can say:

$ nix-env -i perl-Class-C3

(Of course you can also install using the attribute name: nix-env -i -A perlPackages.ClassC3.)

So what does buildPerlPackage do? It does the following:

  1. In the configure phase, it calls perl Makefile.PL to generate a Makefile. You can set the variable makeMakerFlags to pass flags to Makefile.PL

  2. It adds the contents of the PERL5LIB environment variable to #! .../bin/perl line of Perl scripts as -Idir flags. This ensures that a script can find its dependencies. (This can cause this shebang line to become too long for Darwin to handle; see the note below.)

  3. In the fixup phase, it writes the propagated build inputs (propagatedBuildInputs) to the file $out/nix-support/propagated-user-env-packages. nix-env recursively installs all packages listed in this file when you install a package that has it. This ensures that a Perl package can find its dependencies.

buildPerlPackage is built on top of stdenv, so everything can be customised in the usual way. For instance, the BerkeleyDB module has a preConfigure hook to generate a configuration file used by Makefile.PL:

{ buildPerlPackage, fetchurl, db }:

buildPerlPackage rec {
  name = "BerkeleyDB-0.36";

  src = fetchurl {
    url = "mirror://cpan/authors/id/P/PM/PMQS/${name}.tar.gz";
    sha256 = "07xf50riarb60l1h6m2dqmql8q5dij619712fsgw7ach04d8g3z1";

  preConfigure = ''
    echo "LIB = ${db.out}/lib" >
    echo "INCLUDE = ${}/include" >>

Dependencies on other Perl packages can be specified in the buildInputs and propagatedBuildInputs attributes. If something is exclusively a build-time dependency, use buildInputs; if it’s (also) a runtime dependency, use propagatedBuildInputs. For instance, this builds a Perl module that has runtime dependencies on a bunch of other modules:

ClassC3Componentised = buildPerlPackage rec {
  name = "Class-C3-Componentised-1.0004";
  src = fetchurl {
    url = "mirror://cpan/authors/id/A/AS/ASH/${name}.tar.gz";
    sha256 = "0xql73jkcdbq4q9m0b0rnca6nrlvf5hyzy8is0crdk65bynvs8q1";
  propagatedBuildInputs = [
    ClassC3 ClassInspector TestException MROCompat

On Darwin, if a script has too many -Idir flags in its first line (its “shebang line”), it will not run. This can be worked around by calling the shortenPerlShebang function from the postInstall phase:

{ lib, stdenv, buildPerlPackage, fetchurl, shortenPerlShebang }:

ImageExifTool = buildPerlPackage {
  pname = "Image-ExifTool";
  version = "11.50";

  src = fetchurl {
    url = "";
    sha256 = "0d8v48y94z8maxkmw1rv7v9m0jg2dc8xbp581njb6yhr7abwqdv3";

  buildInputs = lib.optional stdenv.isDarwin shortenPerlShebang;
  postInstall = lib.optionalString stdenv.isDarwin ''
    shortenPerlShebang $out/bin/exiftool

This will remove the -I flags from the shebang line, rewrite them in the use lib form, and put them on the next line instead. This function can be given any number of Perl scripts as arguments; it will modify them in-place. Generation from CPAN

Nix expressions for Perl packages can be generated (almost) automatically from CPAN. This is done by the program nix-generate-from-cpan, which can be installed as follows:

$ nix-env -i nix-generate-from-cpan

This program takes a Perl module name, looks it up on CPAN, fetches and unpacks the corresponding package, and prints a Nix expression on standard output. For example:

$ nix-generate-from-cpan XML::Simple
  XMLSimple = buildPerlPackage rec {
    name = "XML-Simple-2.22";
    src = fetchurl {
      url = "mirror://cpan/authors/id/G/GR/GRANTM/${name}.tar.gz";
      sha256 = "b9450ef22ea9644ae5d6ada086dc4300fa105be050a2030ebd4efd28c198eb49";
    propagatedBuildInputs = [ XMLNamespaceSupport XMLSAX XMLSAXExpat ];
    meta = {
      description = "An API for simple XML files";
      license = with lib.licenses; [ artistic1 gpl1Plus ];

The output can be pasted into pkgs/top-level/perl-packages.nix or wherever else you need it. Cross-compiling modules

Nixpkgs has experimental support for cross-compiling Perl modules. In many cases, it will just work out of the box, even for modules with native extensions. Sometimes, however, the Makefile.PL for a module may (indirectly) import a native module. In that case, you will need to make a stub for that module that will satisfy the Makefile.PL and install it into lib/perl5/site_perl/cross_perl/${perl.version}. See the postInstall for DBI for an example.

15.24. PHP

15.24.1. User Guide Overview

Several versions of PHP are available on Nix, each of which having a wide variety of extensions and libraries available.

The different versions of PHP that nixpkgs provides are located under attributes named based on major and minor version number; e.g., php74 is PHP 7.4.

Only versions of PHP that are supported by upstream for the entirety of a given NixOS release will be included in that release of NixOS. See PHP Supported Versions.

The attribute php refers to the version of PHP considered most stable and thoroughly tested in nixpkgs for any given release of NixOS - not necessarily the latest major release from upstream.

All available PHP attributes are wrappers around their respective binary PHP package and provide commonly used extensions this way. The real PHP 7.4 package, i.e. the unwrapped one, is available as php74.unwrapped; see the next section for more details.

Interactive tools built on PHP are put in php.packages; composer is for example available at php.packages.composer.

Most extensions that come with PHP, as well as some popular third-party ones, are available in php.extensions; for example, the opcache extension shipped with PHP is available at php.extensions.opcache and the third-party ImageMagick extension at php.extensions.imagick. Installing PHP with extensions

A PHP package with specific extensions enabled can be built using php.withExtensions. This is a function which accepts an anonymous function as its only argument; the function should accept two named parameters: enabled - a list of currently enabled extensions and all - the set of all extensions, and return a list of wanted extensions. For example, a PHP package with all default extensions and ImageMagick enabled:

php.withExtensions ({ enabled, all }:
  enabled ++ [ all.imagick ])

To exclude some, but not all, of the default extensions, you can filter the enabled list like this:

php.withExtensions ({ enabled, all }:
  (lib.filter (e: e != php.extensions.opcache) enabled)
  ++ [ all.imagick ])

To build your list of extensions from the ground up, you can simply ignore enabled:

php.withExtensions ({ all, ... }: with all; [ imagick opcache ])

php.withExtensions provides extensions by wrapping a minimal php base package, providing a php.ini file listing all extensions to be loaded. You can access this package through the php.unwrapped attribute; useful if you, for example, need access to the dev output. The generated php.ini file can be accessed through the php.phpIni attribute.

If you want a PHP build with extra configuration in the php.ini file, you can use php.buildEnv. This function takes two named and optional parameters: extensions and extraConfig. extensions takes an extension specification equivalent to that of php.withExtensions, extraConfig a string of additional php.ini configuration parameters. For example, a PHP package with the opcache and ImageMagick extensions enabled, and memory_limit set to 256M:

php.buildEnv {
  extensions = { all, ... }: with all; [ imagick opcache ];
  extraConfig = "memory_limit=256M";
} Example setup for phpfpm

You can use the previous examples in a phpfpm pool called foo as follows:

  myPhp = php.withExtensions ({ all, ... }: with all; [ imagick opcache ]);
in {
  services.phpfpm.pools."foo".phpPackage = myPhp;
  myPhp = php.buildEnv {
    extensions = { all, ... }: with all; [ imagick opcache ];
    extraConfig = "memory_limit=256M";
in {
  services.phpfpm.pools."foo".phpPackage = myPhp;
}; Example usage with nix-shell

This brings up a temporary environment that contains a PHP interpreter with the extensions imagick and opcache enabled:

nix-shell -p 'php.withExtensions ({ all, ... }: with all; [ imagick opcache ])' Installing PHP packages with extensions

All interactive tools use the PHP package you get them from, so all packages at php.packages.* use the php package with its default extensions. Sometimes this default set of extensions isn’t enough and you may want to extend it. A common case of this is the composer package: a project may depend on certain extensions and composer won’t work with that project unless those extensions are loaded.

Example of building composer with additional extensions:

(php.withExtensions ({ all, enabled }:
  enabled ++ (with all; [ imagick redis ]))
).packages.composer Overriding PHP packages

php-packages.nix form a scope, allowing us to override the packages defined within. For example, to apply a patch to a mysqlnd extension, you can simply pass an overlay-style function to php’s packageOverrides argument:

php.override {
  packageOverrides = final: prev: {
    extensions = prev.extensions // {
      mysqlnd = prev.extensions.mysqlnd.overrideAttrs (attrs: {
        patches = attrs.patches or [] ++ [

15.25. Python

15.25.1. User Guide Using Python Overview

Several versions of the Python interpreter are available on Nix, as well as a high amount of packages. The attribute python3 refers to the default interpreter, which is currently CPython 3.9. The attribute python refers to CPython 2.7 for backwards-compatibility. It is also possible to refer to specific versions, e.g. python38 refers to CPython 3.8, and pypy refers to the default PyPy interpreter.

Python is used a lot, and in different ways. This affects also how it is packaged. In the case of Python on Nix, an important distinction is made between whether the package is considered primarily an application, or whether it should be used as a library, i.e., of primary interest are the modules in site-packages that should be importable.

In the Nixpkgs tree Python applications can be found throughout, depending on what they do, and are called from the main package set. Python libraries, however, are in separate sets, with one set per interpreter version.

The interpreters have several common attributes. One of these attributes is pkgs, which is a package set of Python libraries for this specific interpreter. E.g., the toolz package corresponding to the default interpreter is python.pkgs.toolz, and the CPython 3.8 version is python38.pkgs.toolz. The main package set contains aliases to these package sets, e.g. pythonPackages refers to python.pkgs and python38Packages to python38.pkgs. Installing Python and packages

The Nix and NixOS manuals explain how packages are generally installed. In the case of Python and Nix, it is important to make a distinction between whether the package is considered an application or a library.

Applications on Nix are typically installed into your user profile imperatively using nix-env -i, and on NixOS declaratively by adding the package name to environment.systemPackages in /etc/nixos/configuration.nix. Dependencies such as libraries are automatically installed and should not be installed explicitly.

The same goes for Python applications. Python applications can be installed in your profile, and will be wrapped to find their exact library dependencies, without impacting other applications or polluting your user environment.

But Python libraries you would like to use for development cannot be installed, at least not individually, because they won’t be able to find each other resulting in import errors. Instead, it is possible to create an environment with python.buildEnv or python.withPackages where the interpreter and other executables are wrapped to be able to find each other and all of the modules.

In the following examples we will start by creating a simple, ad-hoc environment with a nix-shell that has numpy and toolz in Python 3.8; then we will create a re-usable environment in a single-file Python script; then we will create a full Python environment for development with this same environment.

Philosphically, this should be familiar to users who are used to a venv style of development: individual projects create their own Python environments without impacting the global environment or each other. Ad-hoc temporary Python environment with nix-shell

The simplest way to start playing with the way nix wraps and sets up Python environments is with nix-shell at the cmdline. These environments create a temporary shell session with a Python and a precise list of packages (plus their runtime dependencies), with no other Python packages in the Python interpreter’s scope.

To create a Python 3.8 session with numpy and toolz available, run:

$ nix-shell -p 'python38.withPackages(ps: with ps; [ numpy toolz ])'

By default nix-shell will start a bash session with this interpreter in our PATH, so if we then run:

[nix-shell:~/src/nixpkgs]$ python3
Python 3.8.1 (default, Dec 18 2019, 19:06:26)
[GCC 9.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy; import toolz

Note that no other modules are in scope, even if they were imperatively installed into our user environment as a dependency of a Python application:

>>> import requests
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'requests'

We can add as many additional modules onto the nix-shell as we need, and we will still get 1 wrapped Python interpreter. We can start the interpreter directly like so:

$ nix-shell -p 'python38.withPackages(ps: with ps; [ numpy toolz requests ])' --run python3
these derivations will be built:
building '/nix/store/xbdsrqrsfa1yva5s7pzsra8k08gxlbz1-python3-3.8.1-env.drv'...
created 277 symlinks in user environment
Python 3.8.1 (default, Dec 18 2019, 19:06:26)
[GCC 9.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import requests

Notice that this time it built a new Python environment, which now includes requests. Building an environment just creates wrapper scripts that expose the selected dependencies to the interpreter while re-using the actual modules. This means if any other env has installed requests or numpy in a different context, we don’t need to recompile them – we just recompile the wrapper script that sets up an interpreter pointing to them. This matters much more for big modules like pytorch or tensorflow.

Module names usually match their names on, but you can use the Nixpkgs search website to find them as well (along with non-python packages).

At this point we can create throwaway experimental Python environments with arbitrary dependencies. This is a good way to get a feel for how the Python interpreter and dependencies work in Nix and NixOS, but to do some actual development, we’ll want to make it a bit more persistent. Running Python scripts and using nix-shell as shebang

Sometimes, we have a script whose header looks like this:

#!/usr/bin/env python3
import numpy as np
a = np.array([1,2])
b = np.array([3,4])
print(f"The dot product of {a} and {b} is: {, b)}")

Executing this script requires a python3 that has numpy. Using what we learned in the previous section, we could startup a shell and just run it like so:

$ nix-shell -p 'python38.withPackages(ps: with ps; [ numpy ])' --run 'python3'
The dot product of [1 2] and [3 4] is: 11

But if we maintain the script ourselves, and if there are more dependencies, it may be nice to encode those dependencies in source to make the script re-usable without that bit of knowledge. That can be done by using nix-shell as a shebang, like so:

#!/usr/bin/env nix-shell
#!nix-shell -i python3 -p "python3.withPackages(ps: [ ps.numpy ])"
import numpy as np
a = np.array([1,2])
b = np.array([3,4])
print(f"The dot product of {a} and {b} is: {, b)}")

Then we simply execute it, without requiring any environment setup at all!

$ ./
The dot product of [1 2] and [3 4] is: 11

If the dependencies are not available on the host where is executed, it will build or download them from a Nix binary cache prior to starting up, prior that it is executed on a machine with a multi-user nix installation.

This provides a way to ship a self bootstrapping Python script, akin to a statically linked binary, where it can be run on any machine (provided nix is installed) without having to assume that numpy is installed globally on the system.

By default it is pulling the import checkout of Nixpkgs itself from our nix channel, which is nice as it cache aligns with our other package builds, but we can make it fully reproducible by pinning the nixpkgs import:

#!/usr/bin/env nix-shell
#!nix-shell -i python3 -p "python3.withPackages(ps: [ ps.numpy ])"
#!nix-shell -I nixpkgs=
import numpy as np
a = np.array([1,2])
b = np.array([3,4])
print(f"The dot product of {a} and {b} is: {, b)}")

This will execute with the exact same versions of Python 3.8, numpy, and system dependencies a year from now as it does today, because it will always use exactly git commit d373d80b1207d52621961b16aa4a3438e4f98167 of Nixpkgs for all of the package versions.

This is also a great way to ensure the script executes identically on different servers. Load environment from .nix expression

We’ve now seen how to create an ad-hoc temporary shell session, and how to create a single script with Python dependencies, but in the course of normal development we’re usually working in an entire package repository.

As explained in the Nix manual, nix-shell can also load an expression from a .nix file. Say we want to have Python 3.8, numpy and toolz, like before, in an environment. We can add a shell.nix file describing our dependencies:

with import <nixpkgs> {};
(python38.withPackages (ps: [ps.numpy ps.toolz])).env

And then at the command line, just typing nix-shell produces the same environment as before. In a normal project, we’ll likely have many more dependencies; this can provide a way for developers to share the environments with each other and with CI builders.

What’s happening here?

  1. We begin with importing the Nix Packages collections. import <nixpkgs> imports the <nixpkgs> function, {} calls it and the with statement brings all attributes of nixpkgs in the local scope. These attributes form the main package set.

  2. Then we create a Python 3.8 environment with the withPackages function, as before.

  3. The withPackages function expects us to provide a function as an argument that takes the set of all Python packages and returns a list of packages to include in the environment. Here, we select the packages numpy and toolz from the package set.

To combine this with mkShell you can:

with import <nixpkgs> {};
  pythonEnv = python38.withPackages (ps: [
in mkShell {
  packages = [



This will create a unified environment that has not just our Python interpreter and its Python dependencies, but also tools like black or mypy and libraries like libffi the openssl in scope. This is generic and can span any number of tools or languages across the Nixpkgs ecosystem. Installing environments globally on the system

Up to now, we’ve been creating environments scoped to an ad-hoc shell session, or a single script, or a single project. This is generally advisable, as it avoids pollution across contexts.

However, sometimes we know we will often want a Python with some basic packages, and want this available without having to enter into a shell or build context. This can be useful to have things like vim/emacs editors and plugins or shell tools just work without having to set them up, or when running other software that expects packages to be installed globally.

To create your own custom environment, create a file in ~/.config/nixpkgs/overlays/ that looks like this:

# ~/.config/nixpkgs/overlays/myEnv.nix
self: super: {
  myEnv = super.buildEnv {
    name = "myEnv";
    paths = [
      # A Python 3 interpreter with some packages
      (self.python3.withPackages (
        ps: with ps; [

      # Some other packages we'd like as part of this env

You can then build and install this to your profile with:

nix-env -iA myEnv

One limitation of this is that you can only have 1 Python env installed globally, since they conflict on the python to load out of your PATH.

If you get a conflict or prefer to keep the setup clean, you can have nix-env atomically uninstall all other imperatively installed packages and replace your profile with just myEnv by using the --replace flag. Environment defined in /etc/nixos/configuration.nix

For the sake of completeness, here’s how to install the environment system-wide on NixOS.

{ # ...

  environment.systemPackages = with pkgs; [
    (python38.withPackages(ps: with ps; [ numpy toolz ]))
} Developing with Python

Above, we were mostly just focused on use cases and what to do to get started creating working Python environments in nix.

Now that you know the basics to be up and running, it is time to take a step back and take a deeper look at how Python packages are packaged on Nix. Then, we will look at how you can use development mode with your code. Python library packages in Nixpkgs

With Nix all packages are built by functions. The main function in Nix for building Python libraries is buildPythonPackage. Let’s see how we can build the toolz package.

{ lib, buildPythonPackage, fetchPypi }:

buildPythonPackage rec {
  pname = "toolz";
  version = "0.10.0";

  src = fetchPypi {
    inherit pname version;
    sha256 = "08fdd5ef7c96480ad11c12d472de21acd32359996f69a5259299b540feba4560";

  doCheck = false;

  meta = with lib; {
    homepage = "";
    description = "List processing tools and functional utilities";
    license = licenses.bsd3;
    maintainers = with maintainers; [ fridh ];

What happens here? The function buildPythonPackage is called and as argument it accepts a set. In this case the set is a recursive set, rec. One of the arguments is the name of the package, which consists of a basename (generally following the name on PyPi) and a version. Another argument, src specifies the source, which in this case is fetched from PyPI using the helper function fetchPypi. The argument doCheck is used to set whether tests should be run when building the package. Furthermore, we specify some (optional) meta information. The output of the function is a derivation.

An expression for toolz can be found in the Nixpkgs repository. As explained in the introduction of this Python section, a derivation of toolz is available for each interpreter version, e.g. python38.pkgs.toolz refers to the toolz derivation corresponding to the CPython 3.8 interpreter.

The above example works when you’re directly working on pkgs/top-level/python-packages.nix in the Nixpkgs repository. Often though, you will want to test a Nix expression outside of the Nixpkgs tree.

The following expression creates a derivation for the toolz package, and adds it along with a numpy package to a Python environment.

with import <nixpkgs> {};

( let
    my_toolz = python38.pkgs.buildPythonPackage rec {
      pname = "toolz";
      version = "0.10.0";

      src = python38.pkgs.fetchPypi {
        inherit pname version;
        sha256 = "08fdd5ef7c96480ad11c12d472de21acd32359996f69a5259299b540feba4560";

      doCheck = false;

      meta = {
        homepage = "";
        description = "List processing tools and functional utilities";

  in python38.withPackages (ps: [ps.numpy my_toolz])

Executing nix-shell will result in an environment in which you can use Python 3.8 and the toolz package. As you can see we had to explicitly mention for which Python version we want to build a package.

So, what did we do here? Well, we took the Nix expression that we used earlier to build a Python environment, and said that we wanted to include our own version of toolz, named my_toolz. To introduce our own package in the scope of withPackages we used a let expression. You can see that we used ps.numpy to select numpy from the nixpkgs package set (ps). We did not take toolz from the Nixpkgs package set this time, but instead took our own version that we introduced with the let expression. Handling dependencies

Our example, toolz, does not have any dependencies on other Python packages or system libraries. According to the manual, buildPythonPackage uses the arguments buildInputs and propagatedBuildInputs to specify dependencies. If something is exclusively a build-time dependency, then the dependency should be included in buildInputs, but if it is (also) a runtime dependency, then it should be added to propagatedBuildInputs. Test dependencies are considered build-time dependencies and passed to checkInputs.

The following example shows which arguments are given to buildPythonPackage in order to build datashape.

{ lib, buildPythonPackage, fetchPypi, numpy, multipledispatch, python-dateutil, pytest }:

buildPythonPackage rec {
  pname = "datashape";
  version = "0.4.7";

  src = fetchPypi {
    inherit pname version;
    sha256 = "14b2ef766d4c9652ab813182e866f493475e65e558bed0822e38bf07bba1a278";

  checkInputs = [ pytest ];
  propagatedBuildInputs = [ numpy multipledispatch python-dateutil ];

  meta = with lib; {
    homepage = "";
    description = "A data description language";
    license = licenses.bsd2;
    maintainers = with maintainers; [ fridh ];

We can see several runtime dependencies, numpy, multipledispatch, and python-dateutil. Furthermore, we have one checkInputs, i.e. pytest. pytest is a test runner and is only used during the checkPhase and is therefore not added to propagatedBuildInputs.

In the previous case we had only dependencies on other Python packages to consider. Occasionally you have also system libraries to consider. E.g., lxml provides Python bindings to libxml2 and libxslt. These libraries are only required when building the bindings and are therefore added as buildInputs.

{ lib, pkgs, buildPythonPackage, fetchPypi }:

buildPythonPackage rec {
  pname = "lxml";
  version = "3.4.4";

  src = fetchPypi {
    inherit pname version;
    sha256 = "16a0fa97hym9ysdk3rmqz32xdjqmy4w34ld3rm3jf5viqjx65lxk";

  buildInputs = [ pkgs.libxml2 pkgs.libxslt ];

  meta = with lib; {
    description = "Pythonic binding for the libxml2 and libxslt libraries";
    homepage = "";
    license = licenses.bsd3;
    maintainers = with maintainers; [ sjourdois ];

In this example lxml and Nix are able to work out exactly where the relevant files of the dependencies are. This is not always the case.

The example below shows bindings to The Fastest Fourier Transform in the West, commonly known as FFTW. On Nix we have separate packages of FFTW for the different types of floats ("single", "double", "long-double"). The bindings need all three types, and therefore we add all three as buildInputs. The bindings don’t expect to find each of them in a different folder, and therefore we have to set LDFLAGS and CFLAGS.

{ lib, pkgs, buildPythonPackage, fetchPypi, numpy, scipy }:

buildPythonPackage rec {
  pname = "pyFFTW";
  version = "0.9.2";

  src = fetchPypi {
    inherit pname version;
    sha256 = "f6bbb6afa93085409ab24885a1a3cdb8909f095a142f4d49e346f2bd1b789074";

  buildInputs = [ pkgs.fftw pkgs.fftwFloat pkgs.fftwLongDouble];

  propagatedBuildInputs = [ numpy scipy ];

  # Tests cannot import pyfftw. pyfftw works fine though.
  doCheck = false;

  preConfigure = ''
    export LDFLAGS="-L${}/lib -L${pkgs.fftwFloat.out}/lib -L${pkgs.fftwLongDouble.out}/lib"
    export CFLAGS="-I${}/include -I${}/include -I${}/include"

  meta = with lib; {
    description = "A pythonic wrapper around FFTW, the FFT library, presenting a unified interface for all the supported transforms";
    homepage = "";
    license = with licenses; [ bsd2 bsd3 ];
    maintainers = with maintainers; [ fridh ];

Note also the line doCheck = false;, we explicitly disabled running the test-suite. Testing Python Packages

It is highly encouraged to have testing as part of the package build. This helps to avoid situations where the package was able to build and install, but is not usable at runtime. Currently, all packages will use the test command provided by the (i.e. python test). However, this is currently deprecated and your package should provide its own checkPhase.

NOTE: The checkPhase for python maps to the installCheckPhase on a normal derivation. This is due to many python packages not behaving well to the pre-installed version of the package. Version info, and natively compiled extensions generally only exist in the install directory, and thus can cause issues when a test suite asserts on that behavior.

NOTE: Tests should only be disabled if they don’t agree with nix (e.g. external dependencies, network access, flakey tests), however, as many tests should be enabled as possible. Failing tests can still be a good indication that the package is not in a valid state. Using pytest

Pytest is the most common test runner for python repositories. A trivial test run would be:

  checkInputs = [ pytest ];
  checkPhase = "pytest";

However, many repositories’ test suites do not translate well to nix’s build sandbox, and will generally need many tests to be disabled.

To filter tests using pytest, one can do the following:

  checkInputs = [ pytest ];
  # avoid tests which need additional data or touch network
  checkPhase = ''
    pytest tests/ --ignore=tests/integration -k 'not download and not update'

--ignore will tell pytest to ignore that file or directory from being collected as part of a test run. This is useful is a file uses a package which is not available in nixpkgs, thus skipping that test file is much easier than having to create a new package.

-k is used to define a predicate for test names. In this example, we are filtering out tests which contain download or update in their test case name. Only one -k argument is allowed, and thus a long predicate should be concatenated with “\” and wrapped to the next line.

NOTE: In pytest==6.0.1, the use of “\” to continue a line (e.g. -k 'not download \') has been removed, in this case, it’s recommended to use pytestCheckHook. Using pytestCheckHook

pytestCheckHook is a convenient hook which will substitute the setuptools test command for a checkPhase which runs pytest. This is also beneficial when a package may need many items disabled to run the test suite.

Using the example above, the analagous pytestCheckHook usage would be:

  checkInputs = [ pytestCheckHook ];

  # requires additional data
  pytestFlagsArray = [ "tests/" "--ignore=tests/integration" ];

  disabledTests = [
    # touches network

  disabledTestPaths = [

This is expecially useful when tests need to be conditionallydisabled, for example:

  disabledTests = [
    # touches network
  ] ++ lib.optionals (pythonAtLeast "3.8") [
    # broken due to python3.8 async changes
  ] ++ lib.optionals stdenv.isDarwin [
    # can fail when building with other packages

Trying to concatenate the related strings to disable tests in a regular checkPhase would be much harder to read. This also enables us to comment on why specific tests are disabled. Using pythonImportsCheck

Although unit tests are highly prefered to validate correctness of a package, not all packages have test suites that can be ran easily, and some have none at all. To help ensure the package still works, pythonImportsCheck can attempt to import the listed modules.

  pythonImportsCheck = [ "requests" "urllib" ];

roughly translates to:

  postCheck = ''
    python -c "import requests; import urllib"

However, this is done in it’s own phase, and not dependent on whether doCheck = true;

This can also be useful in verifying that the package doesn’t assume commonly present packages (e.g. setuptools) Develop local package

As a Python developer you’re likely aware of development mode (python develop); instead of installing the package this command creates a special link to the project code. That way, you can run updated code without having to reinstall after each and every change you make. Development mode is also available. Let’s see how you can use it.

In the previous Nix expression the source was fetched from an url. We can also refer to a local source instead using src = ./path/to/source/tree;

If we create a shell.nix file which calls buildPythonPackage, and if src is a local source, and if the local source has a, then development mode is activated.

In the following example we create a simple environment that has a Python 3.8 version of our package in it, as well as its dependencies and other packages we like to have in the environment, all specified with propagatedBuildInputs. Indeed, we can just add any package we like to have in our environment to propagatedBuildInputs.

with import <nixpkgs> {};
with python38Packages;

buildPythonPackage rec {
  name = "mypackage";
  src = ./path/to/package/source;
  propagatedBuildInputs = [ pytest numpy pkgs.libsndfile ];

It is important to note that due to how development mode is implemented on Nix it is not possible to have multiple packages simultaneously in development mode. Organising your packages

So far we discussed how you can use Python on Nix, and how you can develop with it. We’ve looked at how you write expressions to package Python packages, and we looked at how you can create environments in which specified packages are available.

At some point you’ll likely have multiple packages which you would like to be able to use in different projects. In order to minimise unnecessary duplication we now look at how you can maintain a repository with your own packages. The important functions here are import and callPackage. Including a derivation using callPackage

Earlier we created a Python environment using withPackages, and included the toolz package via a let expression. Let’s split the package definition from the environment definition.

We first create a function that builds toolz in ~/path/to/toolz/release.nix

{ lib, buildPythonPackage }:

buildPythonPackage rec {
  pname = "toolz";
  version = "0.10.0";

  src = fetchPypi {
    inherit pname version;
    sha256 = "08fdd5ef7c96480ad11c12d472de21acd32359996f69a5259299b540feba4560";

  meta = with lib; {
    homepage = "";
    description = "List processing tools and functional utilities";
    license = licenses.bsd3;
    maintainers = with maintainers; [ fridh ];

It takes an argument buildPythonPackage. We now call this function using callPackage in the definition of our environment

with import <nixpkgs> {};

( let
    toolz = callPackage /path/to/toolz/release.nix {
      buildPythonPackage = python38Packages.buildPythonPackage;
  in python38.withPackages (ps: [ ps.numpy toolz ])

Important to remember is that the Python version for which the package is made depends on the python derivation that is passed to buildPythonPackage. Nix tries to automatically pass arguments when possible, which is why generally you don’t explicitly define which python derivation should be used. In the above example we use buildPythonPackage that is part of the set python38Packages, and in this case the python38 interpreter is automatically used.

15.25.2. Reference Interpreters

Versions 2.7, 3.6, 3.7, 3.8 and 3.9 of the CPython interpreter are available as respectively python27, python37, python38 and python39. The aliases python2 and python3 correspond to respectively python27 and python39. The attribute python maps to python2. The PyPy interpreters compatible with Python 2.7 and 3 are available as pypy27 and pypy3, with aliases pypy2 mapping to pypy27 and pypy mapping to pypy2. The Nix expressions for the interpreters can be found in pkgs/development/interpreters/python.

All packages depending on any Python interpreter get appended out/{python.sitePackages} to $PYTHONPATH if such directory exists. Missing tkinter module standard library

To reduce closure size the Tkinter/tkinter is available as a separate package, pythonPackages.tkinter. Attributes on interpreters packages

Each interpreter has the following attributes:

  • libPrefix. Name of the folder in ${python}/lib/ for corresponding interpreter.

  • interpreter. Alias for ${python}/bin/${executable}.

  • buildEnv. Function to build python interpreter environments with extra packages bundled together. See section python.buildEnv function for usage and documentation.

  • withPackages. Simpler interface to buildEnv. See section python.withPackages function for usage and documentation.

  • sitePackages. Alias for lib/${libPrefix}/site-packages.

  • executable. Name of the interpreter executable, e.g. python3.8.

  • pkgs. Set of Python packages for that specific interpreter. The package set can be modified by overriding the interpreter and passing packageOverrides. Optimizations

The Python interpreters are by default not build with optimizations enabled, because the builds are in that case not reproducible. To enable optimizations, override the interpreter of interest, e.g using

  pkgs = import ./. {};
  mypython = pkgs.python3.override {
    enableOptimizations = true;
    reproducibleBuild = false;
    self = mypython;
in mypython Building packages and applications

Python libraries and applications that use setuptools or distutils are typically built with respectively the buildPythonPackage and buildPythonApplication functions. These two functions also support installing a wheel.

All Python packages reside in pkgs/top-level/python-packages.nix and all applications elsewhere. In case a package is used as both a library and an application, then the package should be in pkgs/top-level/python-packages.nix since only those packages are made available for all interpreter versions. The preferred location for library expressions is in pkgs/development/python-modules. It is important that these packages are called from pkgs/top-level/python-packages.nix and not elsewhere, to guarantee the right version of the package is built.

Based on the packages defined in pkgs/top-level/python-packages.nix an attribute set is created for each available Python interpreter. The available sets are

  • pkgs.python27Packages

  • pkgs.python37Packages

  • pkgs.python38Packages

  • pkgs.python39Packages

  • pkgs.python310Packages

  • pkgs.pypyPackages

and the aliases

  • pkgs.python2Packages pointing to pkgs.python27Packages

  • pkgs.python3Packages pointing to pkgs.python39Packages

  • pkgs.pythonPackages pointing to pkgs.python2Packages buildPythonPackage function

The buildPythonPackage function is implemented in pkgs/development/interpreters/python/mk-python-derivation using setup hooks.

The following is an example:

{ lib, buildPythonPackage, fetchPypi, hypothesis, setuptools-scm, attrs, py, setuptools, six, pluggy }:

buildPythonPackage rec {
  pname = "pytest";
  version = "3.3.1";

  src = fetchPypi {
    inherit pname version;
    sha256 = "cf8436dc59d8695346fcd3ab296de46425ecab00d64096cebe79fb51ecb2eb93";

  postPatch = ''
    # don't test bash builtins
    rm testing/

  checkInputs = [ hypothesis ];
  nativeBuildInputs = [ setuptools-scm ];
  propagatedBuildInputs = [ attrs py setuptools six pluggy ];

  meta = with lib; {
    maintainers = with maintainers; [ domenkozar lovek323 madjar lsix ];
    description = "Framework for writing tests";

The buildPythonPackage mainly does four things:

  • In the buildPhase, it calls ${python.interpreter} bdist_wheel to build a wheel binary zipfile.

  • In the installPhase, it installs the wheel file using pip install *.whl.

  • In the postFixup phase, the wrapPythonPrograms bash function is called to wrap all programs in the $out/bin/* directory to include $PATH environment variable and add dependent libraries to script’s sys.path.

  • In the installCheck phase, ${python.interpreter} test is ran.

By default tests are run because doCheck = true. Test dependencies, like e.g. the test runner, should be added to checkInputs.

By default meta.platforms is set to the same value as the interpreter unless overridden otherwise. buildPythonPackage parameters

All parameters from stdenv.mkDerivation function are still supported. The following are specific to buildPythonPackage:

  • catchConflicts ? true: If true, abort package build if a package name appears more than once in dependency tree. Default is true.

  • disabled ? false: If true, package is not built for the particular Python interpreter version.

  • dontWrapPythonPrograms ? false: Skip wrapping of Python programs.

  • permitUserSite ? false: Skip setting the PYTHONNOUSERSITE environment variable in wrapped programs.

  • format ? "setuptools": Format of the source. Valid options are "setuptools", "pyproject", "flit", "wheel", and "other". "setuptools" is for when the source has a and setuptools is used to build a wheel, flit, in case flit should be used to build a wheel, and wheel in case a wheel is provided. Use other when a custom buildPhase and/or installPhase is needed.

  • makeWrapperArgs ? []: A list of strings. Arguments to be passed to makeWrapper, which wraps generated binaries. By default, the arguments to makeWrapper set PATH and PYTHONPATH environment variables before calling the binary. Additional arguments here can allow a developer to set environment variables which will be available when the binary is run. For example, makeWrapperArgs = ["--set FOO BAR" "--set BAZ QUX"].

  • namePrefix: Prepends text to ${name} parameter. In case of libraries, this defaults to "python3.8-" for Python 3.8, etc., and in case of applications to "".

  • pipInstallFlags ? []: A list of strings. Arguments to be passed to pip install. To pass options to python install, use --install-option. E.g., pipInstallFlags=["--install-option='--cpp_implementation'"].

  • pythonPath ? []: List of packages to be added into $PYTHONPATH. Packages in pythonPath are not propagated (contrary to propagatedBuildInputs).

  • preShellHook: Hook to execute commands before shellHook.

  • postShellHook: Hook to execute commands after shellHook.

  • removeBinByteCode ? true: Remove bytecode from /bin. Bytecode is only created when the filenames end with .py.

  • setupPyGlobalFlags ? []: List of flags passed to command.

  • setupPyBuildFlags ? []: List of flags passed to build_ext command.

The stdenv.mkDerivation function accepts various parameters for describing build inputs (see Specifying dependencies). The following are of special interest for Python packages, either because these are primarily used, or because their behaviour is different:

  • nativeBuildInputs ? []: Build-time only dependencies. Typically executables as well as the items listed in setup_requires.

  • buildInputs ? []: Build and/or run-time dependencies that need to be compiled for the host machine. Typically non-Python libraries which are being linked.

  • checkInputs ? []: Dependencies needed for running the checkPhase. These are added to nativeBuildInputs when doCheck = true. Items listed in tests_require go here.

  • propagatedBuildInputs ? []: Aside from propagating dependencies, buildPythonPackage also injects code into and wraps executables with the paths included in this list. Items listed in install_requires go here. Overriding Python packages

The buildPythonPackage function has a overridePythonAttrs method that can be used to override the package. In the following example we create an environment where we have the blaze package using an older version of pandas. We override first the Python interpreter and pass packageOverrides which contains the overrides for packages in the package set.

with import <nixpkgs> {};

  python = let
    packageOverrides = self: super: {
      pandas = super.pandas.overridePythonAttrs(old: rec {
        version = "0.19.1";
        src =  super.fetchPypi {
          pname = "pandas";
          inherit version;
          sha256 = "08blshqj9zj1wyjhhw3kl2vas75vhhicvv72flvf1z3jvapgw295";
  in pkgs.python3.override {inherit packageOverrides; self = python;};

in python.withPackages(ps: [ps.blaze])).env buildPythonApplication function

The buildPythonApplication function is practically the same as buildPythonPackage. The main purpose of this function is to build a Python package where one is interested only in the executables, and not importable modules. For that reason, when adding this package to a python.buildEnv, the modules won’t be made available.

Another difference is that buildPythonPackage by default prefixes the names of the packages with the version of the interpreter. Because this is irrelevant for applications, the prefix is omitted.

When packaging a Python application with buildPythonApplication, it should be called with callPackage and passed python or pythonPackages (possibly specifying an interpreter version), like this:

{ lib, python3Packages }:

python3Packages.buildPythonApplication rec {
  pname = "luigi";
  version = "2.7.9";

  src = python3Packages.fetchPypi {
    inherit pname version;
    sha256 = "035w8gqql36zlan0xjrzz9j4lh9hs0qrsgnbyw07qs7lnkvbdv9x";

  propagatedBuildInputs = with python3Packages; [ tornado_4 python-daemon ];

  meta = with lib; {

This is then added to all-packages.nix just as any other application would be.

luigi = callPackage ../applications/networking/cluster/luigi { };

Since the package is an application, a consumer doesn’t need to care about Python versions or modules, which is why they don’t go in pythonPackages. toPythonApplication function

A distinction is made between applications and libraries, however, sometimes a package is used as both. In this case the package is added as a library to python-packages.nix and as an application to all-packages.nix. To reduce duplication the toPythonApplication can be used to convert a library to an application.

The Nix expression shall use buildPythonPackage and be called from python-packages.nix. A reference shall be created from all-packages.nix to the attribute in python-packages.nix, and the toPythonApplication shall be applied to the reference:

youtube-dl = with pythonPackages; toPythonApplication youtube-dl; toPythonModule function

In some cases, such as bindings, a package is created using stdenv.mkDerivation and added as attribute in all-packages.nix. The Python bindings should be made available from python-packages.nix. The toPythonModule function takes a derivation and makes certain Python-specific modifications.

opencv = toPythonModule (pkgs.opencv.override {
  enablePython = true;
  pythonPackages = self;

Do pay attention to passing in the right Python version! python.buildEnv function

Python environments can be created using the low-level pkgs.buildEnv function. This example shows how to create an environment that has the Pyramid Web Framework. Saving the following as default.nix

with import <nixpkgs> {};

python.buildEnv.override {
  extraLibs = [ pythonPackages.pyramid ];
  ignoreCollisions = true;

and running nix-build will create


with wrapped binaries in bin/.

You can also use the env attribute to create local environments with needed packages installed. This is somewhat comparable to virtualenv. For example, running nix-shell with the following shell.nix

with import <nixpkgs> {};

(python3.buildEnv.override {
  extraLibs = with python3Packages; [ numpy requests ];

will drop you into a shell where Python will have the specified packages in its path. python.buildEnv arguments
  • extraLibs: List of packages installed inside the environment.

  • postBuild: Shell command executed after the build of environment.

  • ignoreCollisions: Ignore file collisions inside the environment (default is false).

  • permitUserSite: Skip setting the PYTHONNOUSERSITE environment variable in wrapped binaries in the environment. python.withPackages function

The python.withPackages function provides a simpler interface to the python.buildEnv functionality. It takes a function as an argument that is passed the set of python packages and returns the list of the packages to be included in the environment. Using the withPackages function, the previous example for the Pyramid Web Framework environment can be written like this:

with import <nixpkgs> {};

python.withPackages (ps: [ps.pyramid])

withPackages passes the correct package set for the specific interpreter version as an argument to the function. In the above example, ps equals pythonPackages. But you can also easily switch to using python3:

with import <nixpkgs> {};

python3.withPackages (ps: [ps.pyramid])

Now, ps is set to python3Packages, matching the version of the interpreter.

As python.withPackages simply uses python.buildEnv under the hood, it also supports the env attribute. The shell.nix file from the previous section can thus be also written like this:

with import <nixpkgs> {};

(python38.withPackages (ps: [ps.numpy ps.requests])).env

In contrast to python.buildEnv, python.withPackages does not support the more advanced options such as ignoreCollisions = true or postBuild. If you need them, you have to use python.buildEnv.

Python 2 namespace packages may provide that collide. In that case python.buildEnv should be used with ignoreCollisions = true. Setup hooks

The following are setup hooks specifically for Python packages. Most of these are used in buildPythonPackage.

  • eggUnpackhook to move an egg to the correct folder so it can be installed with the eggInstallHook

  • eggBuildHook to skip building for eggs.

  • eggInstallHook to install eggs.

  • flitBuildHook to build a wheel using flit.

  • pipBuildHook to build a wheel using pip and PEP 517. Note a build system (e.g. setuptools or flit) should still be added as nativeBuildInput.

  • pipInstallHook to install wheels.

  • pytestCheckHook to run tests with pytest. See example usage.

  • pythonCatchConflictsHook to check whether a Python package is not already existing.

  • pythonImportsCheckHook to check whether importing the listed modules works.

  • pythonRemoveBinBytecode to remove bytecode from the /bin folder.

  • setuptoolsBuildHook to build a wheel using setuptools.

  • setuptoolsCheckHook to run tests with python test.

  • venvShellHook to source a Python 3 venv at the venvDir location. A venv is created if it does not yet exist. postVenvCreation can be used to to run commands only after venv is first created.

  • wheelUnpackHook to move a wheel to the correct folder so it can be installed with the pipInstallHook. Development mode

Development or editable mode is supported. To develop Python packages buildPythonPackage has additional logic inside shellPhase to run pip install -e . --prefix $TMPDIR/for the package.

Warning: shellPhase is executed only if exists.

Given a default.nix:

with import <nixpkgs> {};

pythonPackages.buildPythonPackage {
  name = "myproject";
  buildInputs = with pythonPackages; [ pyramid ];

  src = ./.;

Running nix-shell with no arguments should give you the environment in which the package would be built with nix-build.

Shortcut to setup environments with C headers/libraries and Python packages:

nix-shell -p pythonPackages.pyramid zlib libjpeg git

Note: There is a boolean value lib.inNixShell set to true if nix-shell is invoked. Tools

Packages inside nixpkgs are written by hand. However many tools exist in community to help save time. No tool is preferred at the moment. Deterministic builds

The Python interpreters are now built deterministically. Minor modifications had to be made to the interpreters in order to generate deterministic bytecode. This has security implications and is relevant for those using Python in a nix-shell.

When the environment variable DETERMINISTIC_BUILD is set, all bytecode will have timestamp 1. The buildPythonPackage function sets DETERMINISTIC_BUILD=1 and PYTHONHASHSEED=0. Both are also exported in nix-shell. Automatic tests

It is recommended to test packages as part of the build process. Source distributions (sdist) often include test files, but not always.

By default the command python test is run as part of the checkPhase, but often it is necessary to pass a custom checkPhase. An example of such a situation is when py.test is used. Common issues
  • Non-working tests can often be deselected. By default buildPythonPackage runs python test. Most Python modules follows the standard test protocol where the pytest runner can be used instead. py.test supports a -k parameter to ignore test methods or classes:

    buildPythonPackage {
      # ...
      # assumes the tests are located in tests
      checkInputs = [ pytest ];
      checkPhase = ''
        py.test -k 'not function_name and not other_function' tests
  • Tests that attempt to access $HOME can be fixed by using the following work-around before running tests (e.g. preCheck): export HOME=$(mktemp -d)

15.25.3. FAQ How to solve circular dependencies?

Consider the packages A and B that depend on each other. When packaging B, a solution is to override package A not to depend on B as an input. The same should also be done when packaging A. How to override a Python package?

We can override the interpreter and pass packageOverrides. In the following example we rename the pandas package and build it.

with import <nixpkgs> {};

  python = let
    packageOverrides = self: super: {
      pandas = super.pandas.overridePythonAttrs(old: {name="foo";});
  in pkgs.python38.override {inherit packageOverrides;};

in python.withPackages(ps: [ps.pandas])).env

Using nix-build on this expression will build an environment that contains the package pandas but with the new name foo.

All packages in the package set will use the renamed package. A typical use case is to switch to another version of a certain package. For example, in the Nixpkgs repository we have multiple versions of django and scipy. In the following example we use a different version of scipy and create an environment that uses it. All packages in the Python package set will now use the updated scipy version.

with import <nixpkgs> {};

( let
    packageOverrides = self: super: {
      scipy = super.scipy_0_17;
  in (pkgs.python38.override {inherit packageOverrides;}).withPackages (ps: [ps.blaze])

The requested package blaze depends on pandas which itself depends on scipy.

If you want the whole of Nixpkgs to use your modifications, then you can use overlays as explained in this manual. In the following example we build a inkscape using a different version of numpy.

  pkgs = import <nixpkgs> {};
  newpkgs = import pkgs.path { overlays = [ (self: super: {
    python38 = let
      packageOverrides = python-self: python-super: {
        numpy = python-super.numpy_1_18;
    in super.python38.override {inherit packageOverrides;};
  } ) ]; };
in newpkgs.inkscape python bdist_wheel cannot create .whl

Executing python bdist_wheel in a nix-shell fails with

ValueError: ZIP does not support timestamps before 1980

This is because files from the Nix store (which have a timestamp of the UNIX epoch of January 1, 1970) are included in the .ZIP, but .ZIP archives follow the DOS convention of counting timestamps from 1980.

The command bdist_wheel reads the SOURCE_DATE_EPOCH environment variable, which nix-shell sets to 1. Unsetting this variable or giving it a value corresponding to 1980 or later enables building wheels.

Use 1980 as timestamp:

nix-shell --run "SOURCE_DATE_EPOCH=315532800 python3 bdist_wheel"

or the current time:

nix-shell --run "SOURCE_DATE_EPOCH=$(date +%s) python3 bdist_wheel"


nix-shell --run "unset SOURCE_DATE_EPOCH; python3 bdist_wheel" install_data / data_files problems

If you get the following error:

could not create '/nix/store/6l1bvljpy8gazlsw2aw9skwwp4pmvyxw-python-2.7.8/etc':
Permission denied

This is a known bug in setuptools. Setuptools install_data does not respect --prefix. An example of such package using the feature is pkgs/tools/X11/xpra/default.nix.

As workaround install it as an extra preInstall step:

${python.interpreter} install_data --install-dir=$out --root=$out
sed -i '/ = data\_files/d' Rationale of non-existent global site-packages

On most operating systems a global site-packages is maintained. This however becomes problematic if you want to run multiple Python versions or have multiple versions of certain libraries for your projects. Generally, you would solve such issues by creating virtual environments using virtualenv.

On Nix each package has an isolated dependency tree which, in the case of Python, guarantees the right versions of the interpreter and libraries or packages are available. There is therefore no need to maintain a global site-packages.

If you want to create a Python environment for development, then the recommended method is to use nix-shell, either with or without the python.buildEnv function. How to consume Python modules using pip in a virtual environment like I am used to on other Operating Systems?

While this approach is not very idiomatic from Nix perspective, it can still be useful when dealing with pre-existing projects or in situations where it’s not feasible or desired to write derivations for all required dependencies.

This is an example of a default.nix for a nix-shell, which allows to consume a virtual environment created by venv, and install Python modules through pip the traditional way.

Create this default.nix file, together with a requirements.txt and simply execute nix-shell.

with import <nixpkgs> { };

  pythonPackages = python3Packages;
in pkgs.mkShell rec {
  name = "impurePythonEnv";
  venvDir = "./.venv";
  buildInputs = [
    # A Python interpreter including the 'venv' module is required to bootstrap
    # the environment.

    # This execute some shell code to initialize a venv in $venvDir before
    # dropping into the shell

    # Those are dependencies that we would like to use from nixpkgs, which will
    # add them to PYTHONPATH and thus make them accessible from within the venv.

    # In this particular example, in order to compile any binary extensions they may
    # require, the Python modules listed in the hypothetical requirements.txt need
    # the following packages to be installed locally:

  # Run this command, only after creating the virtual environment
  postVenvCreation = ''
    pip install -r requirements.txt

  # Now we can execute any commands within the virtual environment.
  # This is optional and can be left out to run pip manually.
  postShellHook = ''
    # allow pip to install wheels


In case the supplied venvShellHook is insufficient, or when Python 2 support is needed, you can define your own shell hook and adapt to your needs like in the following example:

with import <nixpkgs> { };

  venvDir = "./.venv";
  pythonPackages = python3Packages;
in pkgs.mkShell rec {
  name = "impurePythonEnv";
  buildInputs = [
    # Needed when using python 2.7
    # pythonPackages.virtualenv
    # ...

  # This is very close to how venvShellHook is implemented, but
  # adapted to use 'virtualenv'
  shellHook = ''
    SOURCE_DATE_EPOCH=$(date +%s)

    if [ -d "${venvDir}" ]; then
      echo "Skipping venv creation, '${venvDir}' already exists"
      echo "Creating new venv environment in path: '${venvDir}'"
      # Note that the module venv was only introduced in python 3, so for 2.7
      # this needs to be replaced with a call to virtualenv
      ${pythonPackages.python.interpreter} -m venv "${venvDir}"

    # Under some circumstances it might be necessary to add your virtual
    # environment to PYTHONPATH, which you can do here too;
    # PYTHONPATH=$PWD/${venvDir}/${pythonPackages.python.sitePackages}/:$PYTHONPATH

    source "${venvDir}/bin/activate"

    # As in the previous example, this is optional.
    pip install -r requirements.txt

Note that the pip install is an imperative action. So every time nix-shell is executed it will attempt to download the Python modules listed in requirements.txt. However these will be cached locally within the virtualenv folder and not downloaded again. How to override a Python package from configuration.nix?

If you need to change a package’s attribute(s) from configuration.nix you could do:

  nixpkgs.config.packageOverrides = super: {
    python = super.python.override {
      packageOverrides = python-self: python-super: {
        twisted = python-super.twisted.overrideAttrs (oldAttrs: {
          src = super.fetchPypi {
            pname = "twisted";
            version = "19.10.0";
            sha256 = "7394ba7f272ae722a74f3d969dcf599bc4ef093bc392038748a490f1724a515d";
            extension = "tar.bz2";

pythonPackages.twisted is now globally overridden. All packages and also all NixOS services that reference twisted (such as services.buildbot-worker) now use the new definition. Note that python-super refers to the old package set and python-self to the new, overridden version.

To modify only a Python package set instead of a whole Python derivation, use this snippet:

  myPythonPackages = pythonPackages.override {
    overrides = self: super: {
      twisted = ...;
  } How to override a Python package using overlays?

Use the following overlay template:

self: super: {
  python = super.python.override {
    packageOverrides = python-self: python-super: {
      twisted = python-super.twisted.overrideAttrs (oldAttrs: {
        src = super.fetchPypi {
          pname = "twisted";
          version = "19.10.0";
          sha256 = "7394ba7f272ae722a74f3d969dcf599bc4ef093bc392038748a490f1724a515d";
          extension = "tar.bz2";
} How to use Intel’s MKL with numpy and scipy?

MKL can be configured using an overlay. See the section Using overlays to configure alternatives. What inputs do setup_requires, install_requires and tests_require map to?

In a or setup.cfg it is common to declare dependencies:

  • setup_requires corresponds to nativeBuildInputs

  • install_requires corresponds to propagatedBuildInputs

  • tests_require corresponds to checkInputs

15.25.4. Contributing Contributing guidelines

The following rules are desired to be respected:

  • Python libraries are called from python-packages.nix and packaged with buildPythonPackage. The expression of a library should be in pkgs/development/python-modules/<name>/default.nix.

  • Python applications live outside of python-packages.nix and are packaged with buildPythonApplication.

  • Make sure libraries build for all Python interpreters.

  • By default we enable tests. Make sure the tests are found and, in the case of libraries, are passing for all interpreters. If certain tests fail they can be disabled individually. Try to avoid disabling the tests altogether. In any case, when you disable tests, leave a comment explaining why.

  • Commit names of Python libraries should reflect that they are Python libraries, so write for example pythonPackages.numpy: 1.11 -> 1.12.

  • Attribute names in python-packages.nix as well as pnames should match the library’s name on PyPI, but be normalized according to PEP 0503. This means that characters should be converted to lowercase and . and _ should be replaced by a single - (foo-bar-baz instead of Foo__Bar.baz). If necessary, pname has to be given a different value within fetchPypi.

  • Attribute names in python-packages.nix should be sorted alphanumerically to avoid merge conflicts and ease locating attributes.

15.25.5. Package set maintenance

The whole Python package set has a lot of packages that do not see regular updates, because they either are a very fragile component in the Python ecosystem, like for example the hypothesis package, or packages that have no maintainer, so maintenance falls back to the package set maintainers. Updating packages in bulk

There is a tool to update alot of python libraries in bulk, it exists at maintainers/scripts/update-python-libraries with this repository.

It can quickly update minor or major versions for all packages selected and create update commits, and supports the fetchPypi, fetchurl and fetchFromGitHub fetchers. When updating lots of packages that are hosted on GitHub, exporting a GITHUB_API_TOKEN is highly recommended.

Updating packages in bulk leads to lots of breakages, which is why a stabilization period on the python-unstable branch is required.

Once the branch is sufficiently stable it should normally be merged into the staging branch.

An exemplary call to update all python libraries between minor versions would be:

$ maintainers/scripts/update-python-libraries --target minor --commit --use-pkgs-prefix pkgs/development/python-modules/**/default.nix

15.26. Qt

Writing Nix expressions for Qt libraries and applications is largely similar as for other C++ software. This section assumes some knowledge of the latter. There are two problems that the Nixpkgs Qt infrastructure addresses, which are not shared by other C++ software:

  1. There are usually multiple supported versions of Qt in Nixpkgs. All of a package’s dependencies must be built with the same version of Qt. This is similar to the version constraints imposed on interpreted languages like Python.

  2. Qt makes extensive use of runtime dependency detection. Runtime dependencies are made into build dependencies through wrappers.

15.26.1. Nix expression for a Qt package (default.nix)

    { stdenv, lib, qtbase, wrapQtAppsHook }: 1

    stdenv.mkDerivation {
      pname = "myapp";
      version = "1.0";

      buildInputs = [ qtbase ];
      nativeBuildInputs = [ wrapQtAppsHook ]; 2


Import Qt modules directly, that is: qtbase, qtdeclarative, etc. Do not import Qt package sets such as qt5 because the Qt versions of dependencies may not be coherent, causing build and runtime failures.


All Qt packages must include wrapQtAppsHook in nativeBuildInputs, or you must explicitly set dontWrapQtApps.

15.26.2. Locating runtime dependencies

Qt applications must be wrapped to find runtime dependencies. Include wrapQtAppsHook in nativeBuildInputs:

{ stdenv, wrapQtAppsHook }:

stdenv.mkDerivation {
  # ...
  nativeBuildInputs = [ wrapQtAppsHook ];

Add entries to qtWrapperArgs are to modify the wrappers created by wrapQtAppsHook:

{ stdenv, wrapQtAppsHook }:

stdenv.mkDerivation {
  # ...
  nativeBuildInputs = [ wrapQtAppsHook ];
  qtWrapperArgs = [ ''--prefix PATH : /path/to/bin'' ];

The entries are passed as arguments to wrapProgram.

Set dontWrapQtApps to stop applications from being wrapped automatically. Wrap programs manually with wrapQtApp, using the syntax of wrapProgram:

{ stdenv, lib, wrapQtAppsHook }:

stdenv.mkDerivation {
  # ...
  nativeBuildInputs = [ wrapQtAppsHook ];
  dontWrapQtApps = true;
  preFixup = ''
      wrapQtApp "$out/bin/myapp" --prefix PATH : /path/to/bin
Note: wrapQtAppsHook ignores files that are non-ELF executables. This means that scripts won’t be automatically wrapped so you’ll need to manually wrap them as previously mentioned. An example of when you’d always need to do this is with Python applications that use PyQt.

15.26.3. Adding a library to Nixpkgs

Add Qt libraries to qt5-packages.nix to make them available for every supported Qt version. Example adding a Qt library

The following represents the contents of qt5-packages.nix.

  # ...

  mylib = callPackage ../path/to/mylib {};

  # ...

Libraries are built with every available version of Qt. Use the meta.broken attribute to disable the package for unsupported Qt versions:

{ stdenv, lib, qtbase }:

stdenv.mkDerivation {
  # ...
  # Disable this library with Qt < 5.9.0
  meta.broken = lib.versionOlder qtbase.version "5.9.0";

15.26.4. Adding an application to Nixpkgs

Add Qt applications to qt5-packages.nix. Add an alias to all-packages.nix to select the Qt 5 version used for the application. Example adding a Qt application

The following represents the contents of qt5-packages.nix.

  # ...

  myapp = callPackage ../path/to/myapp {};

  # ...

The following represents the contents of all-packages.nix.

  # ...

  myapp = libsForQt5.myapp;

  # ...

15.27. R

15.27.1. Installation

Define an environment for R that contains all the libraries that you’d like to use by adding the following snippet to your $HOME/.config/nixpkgs/config.nix file:

    packageOverrides = super: let self = super.pkgs; in

        rEnv = super.rWrapper.override {
            packages = with self.rPackages; [

Then you can use nix-env -f "<nixpkgs>" -iA rEnv to install it into your user profile. The set of available libraries can be discovered by running the command nix-env -f "<nixpkgs>" -qaP -A rPackages. The first column from that output is the name that has to be passed to rWrapper in the code snipped above.

However, if you’d like to add a file to your project source to make the environment available for other contributors, you can create a default.nix file like so:

with import <nixpkgs> {};
  myProject = stdenv.mkDerivation {
    name = "myProject";
    version = "1";
    src = if lib.inNixShell then null else nix;

    buildInputs = with rPackages; [

and then run nix-shell . to be dropped into a shell with those packages available.

15.27.2. RStudio

RStudio uses a standard set of packages and ignores any custom R environments or installed packages you may have. To create a custom environment, see rstudioWrapper, which functions similarly to rWrapper:

    packageOverrides = super: let self = super.pkgs; in

        rstudioEnv = super.rstudioWrapper.override {
            packages = with self.rPackages; [

Then like above, nix-env -f "<nixpkgs>" -iA rstudioEnv will install this into your user profile.

Alternatively, you can create a self-contained shell.nix without the need to modify any configuration files:

{ pkgs ? import <nixpkgs> {}

pkgs.rstudioWrapper.override {
  packages = with pkgs.rPackages; [ dplyr ggplot2 reshape2 ];

Executing nix-shell will then drop you into an environment equivalent to the one above. If you need additional packages just add them to the list and re-enter the shell.

15.27.3. Updating the package set

There is a script and associated environment for regenerating the package sets and synchronising the rPackages tree to the current CRAN and matching BIOC release. These scripts are found in the pkgs/development/r-modules directory and executed as follows:

nix-shell generate-shell.nix

Rscript generate-r-packages.R cran  >
mv cran-packages.nix

Rscript generate-r-packages.R bioc  >
mv bioc-packages.nix

Rscript generate-r-packages.R bioc-annotation >
mv bioc-annotation-packages.nix

Rscript generate-r-packages.R bioc-experiment >
mv bioc-experiment-packages.nix

generate-r-packages.R <repo> reads <repo>-packages.nix, therefore the renaming.

Some packages require overrides to specify external dependencies or other patches and special requirements. These overrides are specified in the pkgs/development/r-modules/default.nix file. As the *-packages.nix contents are automatically generated it should not be edited and broken builds should be addressed using overrides.

15.28. Ruby

15.28.1. Using Ruby

Several versions of Ruby interpreters are available on Nix, as well as over 250 gems and many applications written in Ruby. The attribute ruby refers to the default Ruby interpreter, which is currently MRI 2.6. It’s also possible to refer to specific versions, e.g. ruby_2_y, jruby, or mruby.

In the Nixpkgs tree, Ruby packages can be found throughout, depending on what they do, and are called from the main package set. Ruby gems, however are separate sets, and there’s one default set for each interpreter (currently MRI only).

There are two main approaches for using Ruby with gems. One is to use a specifically locked Gemfile for an application that has very strict dependencies. The other is to depend on the common gems, which we’ll explain further down, and rely on them being updated regularly.

The interpreters have common attributes, namely gems, and withPackages. So you can refer to ruby.gems.nokogiri, or ruby_2_7.gems.nokogiri to get the Nokogiri gem already compiled and ready to use.

Since not all gems have executables like nokogiri, it’s usually more convenient to use the withPackages function like this: ruby.withPackages (p: with p; [ nokogiri ]). This will also make sure that the Ruby in your environment will be able to find the gem and it can be used in your Ruby code (for example via ruby or irb executables) via require "nokogiri" as usual. Temporary Ruby environment with nix-shell

Rather than having a single Ruby environment shared by all Ruby development projects on a system, Nix allows you to create separate environments per project. nix-shell gives you the possibility to temporarily load another environment akin to a combined chruby or rvm and bundle exec.

There are two methods for loading a shell with Ruby packages. The first and recommended method is to create an environment with ruby.withPackages and load that.

$ nix-shell -p "ruby.withPackages (ps: with ps; [ nokogiri pry ])"

The other method, which is not recommended, is to create an environment and list all the packages directly.

$ nix-shell -p ruby.gems.nokogiri ruby.gems.pry

Again, it’s possible to launch the interpreter from the shell. The Ruby interpreter has the attribute gems which contains all Ruby gems for that specific interpreter. Load Ruby environment from .nix expression

As explained in the Nix manual, nix-shell can also load an expression from a .nix file. Say we want to have Ruby 2.6, nokogori, and pry. Consider a shell.nix file with:

with import <nixpkgs> {};
ruby.withPackages (ps: with ps; [ nokogiri pry ])

What’s happening here?

  1. We begin with importing the Nix Packages collections. import <nixpkgs> imports the <nixpkgs> function, {} calls it and the with statement brings all attributes of nixpkgs in the local scope. These attributes form the main package set.

  2. Then we create a Ruby environment with the withPackages function.

  3. The withPackages function expects us to provide a function as an argument that takes the set of all ruby gems and returns a list of packages to include in the environment. Here, we select the packages nokogiri and pry from the package set. Execute command with --run

A convenient flag for nix-shell is --run. It executes a command in the nix-shell. We can e.g. directly open a pry REPL:

$ nix-shell -p "ruby.withPackages (ps: with ps; [ nokogiri pry ])" --run "pry"

Or immediately require nokogiri in pry:

$ nix-shell -p "ruby.withPackages (ps: with ps; [ nokogiri pry ])" --run "pry -rnokogiri"

Or run a script using this environment:

$ nix-shell -p "ruby.withPackages (ps: with ps; [ nokogiri pry ])" --run "ruby example.rb" Using nix-shell as shebang

In fact, for the last case, there is a more convenient method. You can add a shebang to your script specifying which dependencies nix-shell needs. With the following shebang, you can just execute ./example.rb, and it will run with all dependencies.

#! /usr/bin/env nix-shell
#! nix-shell -i ruby -p "ruby.withPackages (ps: with ps; [ nokogiri rest-client ])"

require 'nokogiri'
require 'rest-client'

body = RestClient.get('').body
puts Nokogiri::HTML(body).at('h1').text

15.28.2. Developing with Ruby Using an existing Gemfile

In most cases, you’ll already have a Gemfile.lock listing all your dependencies. This can be used to generate a gemset.nix which is used to fetch the gems and combine them into a single environment. The reason why you need to have a separate file for this, is that Nix requires you to have a checksum for each input to your build. Since the Gemfile.lock that bundler generates doesn’t provide us with checksums, we have to first download each gem, calculate its SHA256, and store it in this separate file.

So the steps from having just a Gemfile to a gemset.nix are:

$ bundle lock
$ bundix

If you already have a Gemfile.lock, you can simply run bundix and it will work the same.

To update the gems in your Gemfile.lock, you may use the bundix -l flag, which will create a new Gemfile.lock in case the Gemfile has a more recent time of modification.

Once the gemset.nix is generated, it can be used in a bundlerEnv derivation. Here is an example you could use for your shell.nix:

# ...
  gems = bundlerEnv {
    name = "gems-for-some-project";
    gemdir = ./.;
in mkShell { packages = [ gems gems.wrappedRuby ]; }

With this file in your directory, you can run nix-shell to build and use the gems. The important parts here are bundlerEnv and wrappedRuby.

The bundlerEnv is a wrapper over all the gems in your gemset. This means that all the /lib and /bin directories will be available, and the executables of all gems (even of indirect dependencies) will end up in your $PATH. The wrappedRuby provides you with all executables that come with Ruby itself, but wrapped so they can easily find the gems in your gemset.

One common issue that you might have is that you have Ruby 2.6, but also bundler in your gemset. That leads to a conflict for /bin/bundle and /bin/bundler. You can resolve this by wrapping either your Ruby or your gems in a lowPrio call. So in order to give the bundler from your gemset priority, it would be used like this:

# ...
mkShell { buildInputs = [ gems (lowPrio gems.wrappedRuby) ]; } Gem-specific configurations and workarounds

In some cases, especially if the gem has native extensions, you might need to modify the way the gem is built.

This is done via a common configuration file that includes all of the workarounds for each gem.

This file lives at /pkgs/development/ruby-modules/gem-config/default.nix, since it already contains a lot of entries, it should be pretty easy to add the modifications you need for your needs.

In the meanwhile, or if the modification is for a private gem, you can also add the configuration to only your own environment.

Two places that allow this modification are the ruby derivation, or bundlerEnv.

Here’s the ruby one:

{ pg_version ? "10", pkgs ? import <nixpkgs> { } }:
  myRuby = pkgs.ruby.override {
    defaultGemConfig = pkgs.defaultGemConfig // {
      pg = attrs: {
        buildFlags =
        [ "--with-pg-config=${pkgs."postgresql_${pg_version}"}/bin/pg_config" ];
in myRuby.withPackages (ps: with ps; [ pg ])

And an example with bundlerEnv:

{ pg_version ? "10", pkgs ? import <nixpkgs> { } }:
  gems = pkgs.bundlerEnv {
    name = "gems-for-some-project";
    gemdir = ./.;
    gemConfig = pkgs.defaultGemConfig // {
      pg = attrs: {
        buildFlags =
        [ "--with-pg-config=${pkgs."postgresql_${pg_version}"}/bin/pg_config" ];
in mkShell { buildInputs = [ gems gems.wrappedRuby ]; }

And finally via overlays:

{ pg_version ? "10" }:
  pkgs = import <nixpkgs> {
    overlays = [
      (self: super: {
        defaultGemConfig = super.defaultGemConfig // {
          pg = attrs: {
            buildFlags = [
in pkgs.ruby.withPackages (ps: with ps; [ pg ])

Then we can get whichever postgresql version we desire and the pg gem will always reference it correctly:

$ nix-shell --argstr pg_version 9_4 --run 'ruby -rpg -e "puts PG.library_version"'

$ nix-shell --run 'ruby -rpg -e "puts PG.library_version"'

Of course for this use-case one could also use overlays since the configuration for pg depends on the postgresql alias, but for demonstration purposes this has to suffice. Adding a gem to the default gemset

Now that you know how to get a working Ruby environment with Nix, it’s time to go forward and start actually developing with Ruby. We will first have a look at how Ruby gems are packaged on Nix. Then, we will look at how you can use development mode with your code.

All gems in the standard set are automatically generated from a single Gemfile. The dependency resolution is done with bundler and makes it more likely that all gems are compatible to each other.

In order to add a new gem to nixpkgs, you can put it into the /pkgs/development/ruby-modules/with-packages/Gemfile and run ./maintainers/scripts/update-ruby-packages.

To test that it works, you can then try using the gem with:

NIX_PATH=nixpkgs=$PWD nix-shell -p "ruby.withPackages (ps: with ps; [ name-of-your-gem ])" Packaging applications

A common task is to add a ruby executable to nixpkgs, popular examples would be chef, jekyll, or sass. A good way to do that is to use the bundlerApp function, that allows you to make a package that only exposes the listed executables, otherwise the package may cause conflicts through common paths like bin/rake or bin/bundler that aren’t meant to be used.

The absolute easiest way to do that is to write a Gemfile along these lines:

source '' do
  gem 'mdl'

If you want to package a specific version, you can use the standard Gemfile syntax for that, e.g. gem 'mdl', '0.5.0', but if you want the latest stable version anyway, it’s easier to update by simply running the bundle lock and bundix steps again.

Now you can also make a default.nix that looks like this:

{ bundlerApp }:

bundlerApp {
  pname = "mdl";
  gemdir = ./.;
  exes = [ "mdl" ];

All that’s left to do is to generate the corresponding Gemfile.lock and gemset.nix as described above in the Using an existing Gemfile section. Packaging executables that require wrapping

Sometimes your app will depend on other executables at runtime, and tries to find it through the PATH environment variable.

In this case, you can provide a postBuild hook to bundlerApp that wraps the gem in another script that prefixes the PATH.

Of course you could also make a custom gemConfig if you know exactly how to patch it, but it’s usually much easier to maintain with a simple wrapper so the patch doesn’t have to be adjusted for each version.

Here’s another example:

{ lib, bundlerApp, makeWrapper, git, gnutar, gzip }:

bundlerApp {
  pname = "r10k";
  gemdir = ./.;
  exes = [ "r10k" ];

  buildInputs = [ makeWrapper ];

  postBuild = ''
    wrapProgram $out/bin/r10k --prefix PATH : ${lib.makeBinPath [ git gnutar gzip ]}

15.29. Rust

To install the rust compiler and cargo put

environment.systemPackages = [

into your configuration.nix or bring them into scope with nix-shell -p rustc cargo.

For other versions such as daily builds (beta and nightly), use either rustup from nixpkgs (which will manage the rust installation in your home directory), or use a community maintained Rust overlay.

15.29.1. Compiling Rust applications with Cargo

Rust applications are packaged by using the buildRustPackage helper from rustPlatform:

{ lib, fetchFromGitHub, rustPlatform }:

rustPlatform.buildRustPackage rec {
  pname = "ripgrep";
  version = "12.1.1";

  src = fetchFromGitHub {
    owner = "BurntSushi";
    repo = pname;
    rev = version;
    sha256 = "1hqps7l5qrjh9f914r5i6kmcz6f1yb951nv4lby0cjnp5l253kps";

  cargoSha256 = "03wf9r2csi6jpa7v5sw5lpxkrk4wfzwmzx7k3991q3bdjzcwnnwp";

  meta = with lib; {
    description = "A fast line-oriented regex search tool, similar to ag and ack";
    homepage = "";
    license = licenses.unlicense;
    maintainers = [ maintainers.tailhook ];

buildRustPackage requires either the cargoSha256 or the cargoHash attribute which is computed over all crate sources of this package. cargoHash256 is used for traditional Nix SHA-256 hashes, such as the one in the example above. cargoHash should instead be used for SRI hashes. For example:

  cargoHash = "sha256-l1vL2ZdtDRxSGvP0X/l3nMw8+6WF67KPutJEzUROjg8=";

Both types of hashes are permitted when contributing to nixpkgs. The Cargo hash is obtained by inserting a fake checksum into the expression and building the package once. The correct checksum can then be taken from the failed build. A fake hash can be used for cargoSha256 as follows:

  cargoSha256 = lib.fakeSha256;

For cargoHash you can use:

  cargoHash = lib.fakeHash;

Per the instructions in the Cargo Book best practices guide, Rust applications should always commit the Cargo.lock file in git to ensure a reproducible build. However, a few packages do not, and Nix depends on this file, so if it is missing you can use cargoPatches to apply it in the patchPhase. Consider sending a PR upstream with a note to the maintainer describing why it’s important to include in the application.

The fetcher will verify that the Cargo.lock file is in sync with the src attribute, and fail the build if not. It will also will compress the vendor directory into a tar.gz archive.

The tarball with vendored dependencies contains a directory with the package’s name, which is normally composed of pname and version. This means that the vendored dependencies hash (cargoSha256/cargoHash) is dependent on the package name and version. The cargoDepsName attribute can be used to use another name for the directory of vendored dependencies. For example, the hash can be made invariant to the version by setting cargoDepsName to pname:

rustPlatform.buildRustPackage rec {
  pname = "broot";
  version = "1.2.0";

  src = fetchCrate {
    inherit pname version;
    sha256 = "1mqaynrqaas82f5957lx31x80v74zwmwmjxxlbywajb61vh00d38";

  cargoHash = "sha256-JmBZcDVYJaK1cK05cxx5BrnGWp4t8ca6FLUbvIot67s=";
  cargoDepsName = pname;

  # ...
} Importing a Cargo.lock file

Using cargoSha256 or cargoHash is tedious when using buildRustPackage within a project, since it requires that the hash is updated after every change to Cargo.lock. Therefore, buildRustPackage also supports vendoring dependencies directly from a Cargo.lock file using the cargoLock argument. For example:

rustPlatform.buildRustPackage {
  pname = "myproject";
  version = "1.0.0";

  cargoLock = {
    lockFile = ./Cargo.lock;

  # ...

This will retrieve the dependencies using fixed-output derivations from the specified lockfile.

One caveat is that Cargo.lock cannot be patched in the patchPhase because it runs after the dependencies have already been fetched. If you need to patch or generate the lockfile you can alternatively set cargoLock.lockFileContents to a string of its contents:

rustPlatform.buildRustPackage {
  pname = "myproject";
  version = "1.0.0";

  cargoLock = let
    fixupLockFile = path: f (builtins.readFile path);
  in {
    lockFileContents = fixupLockFile ./Cargo.lock;

  # ...

Note that setting cargoLock.lockFile or cargoLock.lockFileContents doesn’t add a Cargo.lock to your src, and a Cargo.lock is still required to build a rust package. A simple fix is to use:

postPatch = ''
  cp ${./Cargo.lock} Cargo.lock

The output hash of each dependency that uses a git source must be specified in the outputHashes attribute. For example:

rustPlatform.buildRustPackage rec {
  pname = "myproject";
  version = "1.0.0";

  cargoLock = {
    lockFile = ./Cargo.lock;
    outputHashes = {
      "finalfusion-0.14.0" = "17f4bsdzpcshwh74w5z119xjy2if6l2wgyjy56v621skr2r8y904";

  # ...

If you do not specify an output hash for a git dependency, building the package will fail and inform you of which crate needs to be added. To find the correct hash, you can first use lib.fakeSha256 or lib.fakeHash as a stub hash. Building the package (and thus the vendored dependencies) will then inform you of the correct hash. Cargo features

You can disable default features using buildNoDefaultFeatures, and extra features can be added with buildFeatures.

If you want to use different features for check phase, you can use checkNoDefaultFeatures and checkFeatures. They are only passed to cargo test and not cargo build. If left unset, they default to buildNoDefaultFeatures and buildFeatures.

For example:

rustPlatform.buildRustPackage rec {
  pname = "myproject";
  version = "1.0.0";

  buildNoDefaultFeatures = true;
  buildFeatures = [ "color" "net" ];

  # disable network features in tests
  checkFeatures = [ "color" ];

  # ...
} Cross compilation

By default, Rust packages are compiled for the host platform, just like any other package is. The --target passed to rust tools is computed from this. By default, it takes the stdenv.hostPlatform.config and replaces components where they are known to differ. But there are ways to customize the argument:

  • To choose a different target by name, define stdenv.hostPlatform.rustc.config as that name (a string), and that name will be used instead.

    For example:

    import <nixpkgs> {
      crossSystem = (import <nixpkgs/lib>).systems.examples.armhf-embedded // {
        rustc.config = "thumbv7em-none-eabi";

    will result in:

    --target thumbv7em-none-eabi
  • To pass a completely custom target, define stdenv.hostPlatform.rustc.config with its name, and stdenv.hostPlatform.rustc.platform with the value. The value will be serialized to JSON in a file called ${stdenv.hostPlatform.rustc.config}.json, and the path of that file will be used instead.

    For example:

    import <nixpkgs> {
      crossSystem = (import <nixpkgs/lib>).systems.examples.armhf-embedded // {
        rustc.config = "thumb-crazy";
        rustc.platform = { foo = ""; bar = ""; };

    will result in:

    --target /nix/store/asdfasdfsadf-thumb-crazy.json # contains {"foo":"","bar":""}

Note that currently custom targets aren’t compiled with std, so cargo test will fail. This can be ignored by adding doCheck = false; to your derivation. Running package tests

When using buildRustPackage, the checkPhase is enabled by default and runs cargo test on the package to build. To make sure that we don’t compile the sources twice and to actually test the artifacts that will be used at runtime, the tests will be ran in the release mode by default.

However, in some cases the test-suite of a package doesn’t work properly in the release mode. For these situations, the mode for checkPhase can be changed like so:

rustPlatform.buildRustPackage {
  /* ... */
  checkType = "debug";

Please note that the code will be compiled twice here: once in release mode for the buildPhase, and again in debug mode for the checkPhase.

Test flags, e.g., --package foo, can be passed to cargo test via the cargoTestFlags attribute.

Another attribute, called checkFlags, is used to pass arguments to the test binary itself, as stated (here)[]. Tests relying on the structure of the target/ directory

Some tests may rely on the structure of the target/ directory. Those tests are likely to fail because we use cargo --target during the build. This means that the artifacts are stored in target/<architecture>/release/, rather than in target/release/.

This can only be worked around by patching the affected tests accordingly. Disabling package-tests

In some instances, it may be necessary to disable testing altogether (with doCheck = false;):

  • If no tests exist – the checkPhase should be explicitly disabled to skip unnecessary build steps to speed up the build.

  • If tests are highly impure (e.g. due to network usage).

There will obviously be some corner-cases not listed above where it’s sensible to disable tests. The above are just guidelines, and exceptions may be granted on a case-by-case basis.

However, please check if it’s possible to disable a problematic subset of the test suite and leave a comment explaining your reasoning. Setting test-threads

buildRustPackage will use parallel test threads by default, sometimes it may be necessary to disable this so the tests run consecutively.

rustPlatform.buildRustPackage {
  /* ... */
  dontUseCargoParallelTests = true;
} Building a package in debug mode

By default, buildRustPackage will use release mode for builds. If a package should be built in debug mode, it can be configured like so:

rustPlatform.buildRustPackage {
  /* ... */
  buildType = "debug";

In this scenario, the checkPhase will be ran in debug mode as well. Custom build/install-procedures

Some packages may use custom scripts for building/installing, e.g. with a Makefile. In these cases, it’s recommended to override the buildPhase/installPhase/checkPhase.

Otherwise, some steps may fail because of the modified directory structure of target/. Building a crate with an absent or out-of-date Cargo.lock file

buildRustPackage needs a Cargo.lock file to get all dependencies in the source code in a reproducible way. If it is missing or out-of-date one can use the cargoPatches attribute to update or add it.

rustPlatform.buildRustPackage rec {
  cargoPatches = [
    # a patch file to add/update Cargo.lock in the source code

15.29.2. Compiling non-Rust packages that include Rust code

Several non-Rust packages incorporate Rust code for performance- or security-sensitive parts. rustPlatform exposes several functions and hooks that can be used to integrate Cargo in non-Rust packages. Vendoring of dependencies

Since network access is not allowed in sandboxed builds, Rust crate dependencies need to be retrieved using a fetcher. rustPlatform provides the fetchCargoTarball fetcher, which vendors all dependencies of a crate. For example, given a source path src containing Cargo.toml and Cargo.lock, fetchCargoTarball can be used as follows:

cargoDeps = rustPlatform.fetchCargoTarball {
  inherit src;
  hash = "sha256-BoHIN/519Top1NUBjpB/oEMqi86Omt3zTQcXFWqrek0=";

The src attribute is required, as well as a hash specified through one of the sha256 or hash attributes. The following optional attributes can also be used:

  • name: the name that is used for the dependencies tarball. If name is not specified, then the name cargo-deps will be used.

  • sourceRoot: when the Cargo.lock/Cargo.toml are in a subdirectory, sourceRoot specifies the relative path to these files.

  • patches: patches to apply before vendoring. This is useful when the Cargo.lock/Cargo.toml files need to be patched before vendoring.

If a Cargo.lock file is available, you can alternatively use the importCargoLock function. In contrast to fetchCargoTarball, this function does not require a hash (unless git dependencies are used) and fetches every dependency as a separate fixed-output derivation. importCargoLock can be used as follows:

cargoDeps = rustPlatform.importCargoLock {
  lockFile = ./Cargo.lock;

If the Cargo.lock file includes git dependencies, then their output hashes need to be specified since they are not available through the lock file. For example:

cargoDeps = rustPlatform.importCargoLock {
  lockFile = ./Cargo.lock;
  outputHashes = {
    "rand-0.8.3" = "0ya2hia3cn31qa8894s3av2s8j5bjwb6yq92k0jsnlx7jid0jwqa";

If you do not specify an output hash for a git dependency, building cargoDeps will fail and inform you of which crate needs to be added. To find the correct hash, you can first use lib.fakeSha256 or lib.fakeHash as a stub hash. Building cargoDeps will then inform you of the correct hash. Hooks

rustPlatform provides the following hooks to automate Cargo builds:

  • cargoSetupHook: configure Cargo to use dependencies vendored through fetchCargoTarball. This hook uses the cargoDeps environment variable to find the vendored dependencies. If a project already vendors its dependencies, the variable cargoVendorDir can be used instead. When the Cargo.toml/Cargo.lock files are not in sourceRoot, then the optional cargoRoot is used to specify the Cargo root directory relative to sourceRoot.

  • cargoBuildHook: use Cargo to build a crate. If the crate to be built is a crate in e.g. a Cargo workspace, the relative path to the crate to build can be set through the optional buildAndTestSubdir environment variable. Features can be specified with cargoBuildNoDefaultFeatures and cargoBuildFeatures. Additional Cargo build flags can be passed through cargoBuildFlags.

  • maturinBuildHook: use Maturin to build a Python wheel. Similar to cargoBuildHook, the optional variable buildAndTestSubdir can be used to build a crate in a Cargo workspace. Additional Maturin flags can be passed through maturinBuildFlags.

  • cargoCheckHook: run tests using Cargo. The build type for checks can be set using cargoCheckType. Features can be specified with cargoCheckNoDefaultFeaatures and cargoCheckFeatures. Additional flags can be passed to the tests using checkFlags and checkFlagsArray. By default, tests are run in parallel. This can be disabled by setting dontUseCargoParallelTests.

  • cargoInstallHook: install binaries and static/shared libraries that were built using cargoBuildHook. Examples Python package using setuptools-rust

For Python packages using setuptools-rust, you can use fetchCargoTarball and cargoSetupHook to retrieve and set up Cargo dependencies. The build itself is then performed by buildPythonPackage.

The following example outlines how the tokenizers Python package is built. Since the Python package is in the source/bindings/python directory of the tokenizers project’s source archive, we use sourceRoot to point the tooling to this directory:

{ fetchFromGitHub
, buildPythonPackage
, rustPlatform
, setuptools-rust

buildPythonPackage rec {
  pname = "tokenizers";
  version = "0.10.0";

  src = fetchFromGitHub {
    owner = "huggingface";
    repo = pname;
    rev = "python-v${version}";
    hash = "sha256-rQ2hRV52naEf6PvRsWVCTN7B1oXAQGmnpJw4iIdhamw=";

  cargoDeps = rustPlatform.fetchCargoTarball {
    inherit src sourceRoot;
    name = "${pname}-${version}";
    hash = "sha256-BoHIN/519Top1NUBjpB/oEMqi86Omt3zTQcXFWqrek0=";

  sourceRoot = "source/bindings/python";

  nativeBuildInputs = [ setuptools-rust ] ++ (with rustPlatform; [

  # ...

In some projects, the Rust crate is not in the main Python source directory. In such cases, the cargoRoot attribute can be used to specify the crate’s directory relative to sourceRoot. In the following example, the crate is in src/rust, as specified in the cargoRoot attribute. Note that we also need to specify the correct path for fetchCargoTarball.

{ buildPythonPackage
, fetchPypi
, rustPlatform
, setuptools-rust
, openssl

buildPythonPackage rec {
  pname = "cryptography";
  version = "3.4.2"; # Also update the hash in vectors.nix

  src = fetchPypi {
    inherit pname version;
    sha256 = "1i1mx5y9hkyfi9jrrkcw804hmkcglxi6rmf7vin7jfnbr2bf4q64";

  cargoDeps = rustPlatform.fetchCargoTarball {
    inherit src;
    sourceRoot = "${pname}-${version}/${cargoRoot}";
    name = "${pname}-${version}";
    hash = "sha256-PS562W4L1NimqDV2H0jl5vYhL08H9est/pbIxSdYVfo=";

  cargoRoot = "src/rust";

  # ...
} Python package using maturin

Python packages that use Maturin can be built with fetchCargoTarball, cargoSetupHook, and maturinBuildHook. For example, the following (partial) derivation builds the retworkx Python package. fetchCargoTarball and cargoSetupHook are used to fetch and set up the crate dependencies. maturinBuildHook is used to perform the build.

{ lib
, buildPythonPackage
, rustPlatform
, fetchFromGitHub

buildPythonPackage rec {
  pname = "retworkx";
  version = "0.6.0";

  src = fetchFromGitHub {
    owner = "Qiskit";
    repo = "retworkx";
    rev = version;
    sha256 = "11n30ldg3y3y6qxg3hbj837pnbwjkqw3nxq6frds647mmmprrd20";

  cargoDeps = rustPlatform.fetchCargoTarball {
    inherit src;
    name = "${pname}-${version}";
    hash = "sha256-heOBK8qi2nuc/Ib+I/vLzZ1fUUD/G/KTw9d7M4Hz5O0=";

  format = "pyproject";

  nativeBuildInputs = with rustPlatform; [ cargoSetupHook maturinBuildHook ];

  # ...

15.29.3. Compiling Rust crates using Nix instead of Cargo Simple operation

When run, cargo build produces a file called Cargo.lock, containing pinned versions of all dependencies. Nixpkgs contains a tool called carnix (nix-env -iA nixos.carnix), which can be used to turn a Cargo.lock into a Nix expression.

That Nix expression calls rustc directly (hence bypassing Cargo), and can be used to compile a crate and all its dependencies. Here is an example for a minimal hello crate:

$ cargo new hello
$ cd hello
$ cargo build
     Compiling hello v0.1.0 (file:///tmp/hello)
     Finished dev [unoptimized + debuginfo] target(s) in 0.20 secs
$ carnix -o hello.nix --src ./. Cargo.lock --standalone
$ nix-build hello.nix -A hello_0_1_0

Now, the file produced by the call to carnix, called hello.nix, looks like:

# Generated by carnix 0.6.5: carnix -o hello.nix --src ./. Cargo.lock --standalone
{ stdenv, buildRustCrate, fetchgit }:
let kernel =;
    # ... (content skipped)
rec {
  hello = f: hello_0_1_0 { features = hello_0_1_0_features { hello_0_1_0 = f; }; };
  hello_0_1_0_ = { dependencies?[], buildDependencies?[], features?[] }: buildRustCrate {
    crateName = "hello";
    version = "0.1.0";
    authors = [ " <>" ];
    src = ./.;
    inherit dependencies buildDependencies features;
  hello_0_1_0 = { features?(hello_0_1_0_features {}) }: hello_0_1_0_ {};
  hello_0_1_0_features = f: updateFeatures f (rec {
        hello_0_1_0.default = (f.hello_0_1_0.default or true);
    }) [ ];

In particular, note that the argument given as --src is copied verbatim to the source. If we look at a more complicated dependencies, for instance by adding a single line libc="*" to our Cargo.toml, we first need to run cargo build to update the Cargo.lock. Then, carnix needs to be run again, and produces the following nix file:

# Generated by carnix 0.6.5: carnix -o hello.nix --src ./. Cargo.lock --standalone
{ stdenv, buildRustCrate, fetchgit }:
let kernel =;
    # ... (content skipped)
rec {
  hello = f: hello_0_1_0 { features = hello_0_1_0_features { hello_0_1_0 = f; }; };
  hello_0_1_0_ = { dependencies?[], buildDependencies?[], features?[] }: buildRustCrate {
    crateName = "hello";
    version = "0.1.0";
    authors = [ " <>" ];
    src = ./.;
    inherit dependencies buildDependencies features;
  libc_0_2_36_ = { dependencies?[], buildDependencies?[], features?[] }: buildRustCrate {
    crateName = "libc";
    version = "0.2.36";
    authors = [ "The Rust Project Developers" ];
    sha256 = "01633h4yfqm0s302fm0dlba469bx8y6cs4nqc8bqrmjqxfxn515l";
    inherit dependencies buildDependencies features;
  hello_0_1_0 = { features?(hello_0_1_0_features {}) }: hello_0_1_0_ {
    dependencies = mapFeatures features ([ libc_0_2_36 ]);
  hello_0_1_0_features = f: updateFeatures f (rec {
    hello_0_1_0.default = (f.hello_0_1_0.default or true);
    libc_0_2_36.default = true;
  }) [ libc_0_2_36_features ];
  libc_0_2_36 = { features?(libc_0_2_36_features {}) }: libc_0_2_36_ {
    features = mkFeatures (features.libc_0_2_36 or {});
  libc_0_2_36_features = f: updateFeatures f (rec {
    libc_0_2_36.default = (f.libc_0_2_36.default or true);
    libc_0_2_36.use_std =
      (f.libc_0_2_36.use_std or false) ||
      (f.libc_0_2_36.default or false) ||
      (libc_0_2_36.default or false);
  }) [];

Here, the libc crate has no src attribute, so buildRustCrate will fetch it from A sha256 attribute is still needed for Nix purity. Handling external dependencies

Some crates require external libraries. For crates from, such libraries can be specified in defaultCrateOverrides package in nixpkgs itself.

Starting from that file, one can add more overrides, to add features or build inputs by overriding the hello crate in a separate file.

with import <nixpkgs> {};
((import ./hello.nix).hello {}).override {
  crateOverrides = defaultCrateOverrides // {
    hello = attrs: { buildInputs = [ openssl ]; };

Here, crateOverrides is expected to be a attribute set, where the key is the crate name without version number and the value a function. The function gets all attributes passed to buildRustCrate as first argument and returns a set that contains all attribute that should be overwritten.

For more complicated cases, such as when parts of the crate’s derivation depend on the crate’s version, the attrs argument of the override above can be read, as in the following example, which patches the derivation:

with import <nixpkgs> {};
((import ./hello.nix).hello {}).override {
  crateOverrides = defaultCrateOverrides // {
    hello = attrs: lib.optionalAttrs (lib.versionAtLeast attrs.version "1.0")  {
      postPatch = ''
        substituteInPlace lib/ \
          --replace "/usr/share/zoneinfo" "${tzdata}/share/zoneinfo"

Another situation is when we want to override a nested dependency. This actually works in the exact same way, since the crateOverrides parameter is forwarded to the crate’s dependencies. For instance, to override the build inputs for crate libc in the example above, where libc is a dependency of the main crate, we could do:

with import <nixpkgs> {};
((import hello.nix).hello {}).override {
  crateOverrides = defaultCrateOverrides // {
    libc = attrs: { buildInputs = []; };
} Options and phases configuration

Actually, the overrides introduced in the previous section are more general. A number of other parameters can be overridden:

  • The version of rustc used to compile the crate:

    (hello {}).override { rust = pkgs.rust; };
  • Whether to build in release mode or debug mod