148dc8991e
flexget: remove python311 fallback logic |
||
---|---|---|
.. | ||
applications | ||
build-support | ||
by-name | ||
common-updater | ||
data | ||
desktops | ||
development | ||
games | ||
kde | ||
misc | ||
os-specific | ||
pkgs-lib | ||
servers | ||
shells | ||
stdenv | ||
test | ||
tools | ||
top-level | ||
README.md |
Contributing to Nixpkgs packages
This document is for people wanting to contribute specifically to the package collection in Nixpkgs. See the CONTRIBUTING.md document for more general information.
Overview
top-level
: Entrypoints, package set aggregationsimpure.nix
,default.nix
,config.nix
: Definitions for the evaluation entry point ofimport <nixpkgs>
stage.nix
,all-packages.nix
,by-name-overlay.nix
,splice.nix
: Definitions for the top-level attribute set made available throughimport <nixpkgs> {…}
*-packages.nix
,linux-kernels.nix
,unixtools.nix
: Aggregations of nested package sets defined indevelopment
aliases.nix
,python-aliases.nix
: Aliases for package definitions that have been renamed or removedrelease*.nix
,make-tarball.nix
,packages-config.nix
,metrics.nix
,nixpkgs-basic-release-checks.nix
: Entry-points and utilities used by Hydra for continuous integration
development
*-modules
,*-packages
,*-pkgs
: Package definitions for nested package sets- All other directories loosely categorise top-level package definitions, see category hierarchy
build-support
: Buildersfetch*
: Fetchers
stdenv
: Standard environmentpkgs-lib
: Definitions for utilities that need packages but are not needed for packagestest
: Tests not directly associated with any specific packagesby-name
: Top-level packages organised by name (docs)- All other directories loosely categorise top-level packages definitions, see category hierarchy
Quick Start to Adding a Package
We welcome new contributors of new packages to Nixpkgs, arguably the greatest software database known. However, each new package comes with a cost for the maintainers, Continuous Integration, caching servers and users downloading Nixpkgs.
Before adding a new package, please consider the following questions:
- Is the package ready for general use? We don't want to include projects that are too immature or are going to be abandoned immediately. In case of doubt, check with upstream.
- Does the project have a clear license statement? Remember that software is unfree by default (all rights reserved), and merely providing access to the source code does not imply its redistribution. In case of doubt, ask upstream.
- How realistic is it that it will be used by other people? It's good that nixpkgs caters to various niches, but if it's a niche of 5 people it's probably too small.
- Are you willing to maintain the package? You should care enough about the package to be willing to keep it up and running for at least one complete Nixpkgs' release life-cycle.
- In case you are not able to maintain the package you wrote, you can seek someone to fill that role, effectively adopting the package.
If any of these questions' answer is no, then you should probably not add the package.
This is section describes a general framework of understanding and exceptions might apply.
Luckily it's pretty easy to maintain your own package set with Nix, which can then be added to the Nix User Repository project.
Now that this is out of the way. To add a package to Nixpkgs:
-
Checkout the Nixpkgs source tree:
$ git clone https://github.com/NixOS/nixpkgs $ cd nixpkgs
-
Create a package directory
pkgs/by-name/so/some-package
wheresome-package
is the package name andso
is the lowercased 2-letter prefix of the package name:$ mkdir -p pkgs/by-name/so/some-package
For more detailed information, see here.
-
Create a
package.nix
file in the package directory, containing a Nix expression — a piece of code that describes how to build the package. In this case, it should be a function that is called with the package dependencies as arguments, and returns a build of the package in the Nix store.$ emacs pkgs/by-name/so/some-package/package.nix $ git add pkgs/by-name/so/some-package/package.nix
If the package is written in a language other than C, you should use the corresponding language framework.
You can have a look at the existing Nix expressions under
pkgs/
to see how it’s done, some of which are also using the category hierarchy. Here are some good ones:-
GNU Hello:
pkgs/by-name/he/hello/package.nix
. Trivial package, which specifies somemeta
attributes which is good practice. -
GNU cpio:
pkgs/tools/archivers/cpio/default.nix
. Also a simple package. The generic builder instdenv
does everything for you. It has no dependencies beyondstdenv
. -
GNU Multiple Precision arithmetic library (GMP):
pkgs/development/libraries/gmp/5.1.x.nix
. Also done by the generic builder, but has a dependency onm4
. -
Pan, a GTK-based newsreader:
pkgs/applications/networking/newsreaders/pan/default.nix
. Has an optional dependency ongtkspell
, which is only built ifspellCheck
istrue
. -
Apache HTTPD:
pkgs/servers/http/apache-httpd/2.4.nix
. A bunch of optional features, variable substitutions in the configure flags, a post-install hook, and miscellaneous hackery. -
buildMozillaMach:
pkgs/applications/networking/browser/firefox/common.nix
. A reusable build function for Firefox, Thunderbird and Librewolf. -
JDiskReport, a Java utility:
pkgs/tools/misc/jdiskreport/default.nix
. Nixpkgs doesn’t have a decentstdenv
for Java yet so this is pretty ad-hoc. -
XML::Simple, a Perl module:
pkgs/top-level/perl-packages.nix
(search for theXMLSimple
attribute). Most Perl modules are so simple to build that they are defined directly inperl-packages.nix
; no need to make a separate file for them. -
Adobe Reader:
pkgs/applications/misc/adobe-reader/default.nix
. Shows how binary-only packages can be supported. In particular the builder usespatchelf
to set the RUNPATH and ELF interpreter of the executables so that the right libraries are found at runtime.
Some notes:
-
Add yourself as the maintainer of the package.
-
All other
meta
attributes are optional, but it’s still a good idea to provide at least thedescription
,homepage
andlicense
. -
The exact syntax and semantics of the Nix expression language, including the built-in functions, are Nix language reference.
-
-
To test whether the package builds, run the following command from the root of the nixpkgs source tree:
$ nix-build -A some-package
where
some-package
should be the package name. You may want to add the flag-K
to keep the temporary build directory in case something fails. If the build succeeds, a symlink./result
to the package in the Nix store is created. -
If you want to install the package into your profile (optional), do
$ nix-env -f . -iA libfoo
-
Optionally commit the new package and open a pull request to nixpkgs, or use the Patches category on Discourse for sending a patch without a GitHub account.
Commit conventions
-
Make sure you read about the commit conventions common to Nixpkgs as a whole.
-
Format the commit messages in the following way:
(pkg-name): (from -> to | init at version | refactor | etc) (Motivation for change. Link to release notes. Additional information.)
Examples:
-
nginx: init at 2.0.1
-
firefox: 54.0.1 -> 55.0
-
Category Hierarchy
Most top-level packages are organised in a loosely-categorised directory hierarchy in this directory. See the overview for which directories are part of this.
This category hierarchy is partially deprecated and will be migrated away over time.
The new pkgs/by-name
directory (docs) should be preferred instead.
The category hierarchy may still be used for packages that should be imported using an alternate callPackage
, such as python3Packages.callPackage
or libsForQt5.callPackage
.
If that is the case for a new package, here are some rules for picking the right category.
Many packages fall under several categories; what matters is the primary purpose of a package.
For example, the libxml2
package builds both a library and some tools; but it’s a library foremost, so it goes under pkgs/development/libraries
.
Categories
If it’s used to support software development:
-
If it’s a library used by other packages:
development/libraries
(e.g.libxml2
)
-
If it’s a compiler:
development/compilers
(e.g.gcc
)
-
If it’s an interpreter:
development/interpreters
(e.g.guile
)
-
If it’s a (set of) development tool(s):
-
If it’s a parser generator (including lexers):
development/tools/parsing
(e.g.bison
,flex
)
-
If it’s a build manager:
development/tools/build-managers
(e.g.gnumake
)
-
If it’s a language server:
development/tools/language-servers
(e.g.ccls
ornil
)
-
Else:
development/tools/misc
(e.g.binutils
)
-
-
Else:
development/misc
If it’s a (set of) tool(s):
(A tool is a relatively small program, especially one intended to be used non-interactively.)
-
If it’s for networking:
tools/networking
(e.g.wget
)
-
If it’s for text processing:
tools/text
(e.g.diffutils
)
-
If it’s a system utility, i.e., something related or essential to the operation of a system:
tools/system
(e.g.cron
)
-
If it’s an archiver (which may include a compression function):
tools/archivers
(e.g.zip
,tar
)
-
If it’s a compression program:
tools/compression
(e.g.gzip
,bzip2
)
-
If it’s a security-related program:
tools/security
(e.g.nmap
,gnupg
)
-
Else:
tools/misc
If it’s a shell:
shells
(e.g.bash
)
If it’s a server:
-
If it’s a web server:
servers/http
(e.g.apache-httpd
)
-
If it’s an implementation of the X Windowing System:
servers/x11
(e.g.xorg
— this includes the client libraries and programs)
-
Else:
servers/misc
If it’s a desktop environment:
desktops
(e.g.kde
,gnome
,enlightenment
)
If it’s a window manager:
applications/window-managers
(e.g.awesome
,stumpwm
)
If it’s an application:
A (typically large) program with a distinct user interface, primarily used interactively.
-
If it’s a version management system:
applications/version-management
(e.g.subversion
)
-
If it’s a terminal emulator:
applications/terminal-emulators
(e.g.alacritty
orrxvt
ortermite
)
-
If it’s a file manager:
applications/file-managers
(e.g.mc
orranger
orpcmanfm
)
-
If it’s for video playback / editing:
applications/video
(e.g.vlc
)
-
If it’s for graphics viewing / editing:
applications/graphics
(e.g.gimp
)
-
If it’s for networking:
-
If it’s a mailreader:
applications/networking/mailreaders
(e.g.thunderbird
)
-
If it’s a newsreader:
applications/networking/newsreaders
(e.g.pan
)
-
If it’s a web browser:
applications/networking/browsers
(e.g.firefox
)
-
Else:
applications/networking/misc
-
-
Else:
applications/misc
If it’s data (i.e., does not have a straight-forward executable semantics):
-
If it’s a font:
data/fonts
-
If it’s an icon theme:
data/icons
-
If it’s related to SGML/XML processing:
-
If it’s an XML DTD:
data/sgml+xml/schemas/xml-dtd
(e.g.docbook
)
-
If it’s an XSLT stylesheet:
(Okay, these are executable...)
data/sgml+xml/stylesheets/xslt
(e.g.docbook-xsl
)
-
-
If it’s a theme for a desktop environment, a window manager or a display manager:
data/themes
If it’s a game:
games
Else:
misc
Conventions
Package naming
The key words must, must not, required, shall, shall not, should, should not, recommended, may, and optional in this section are to be interpreted as described in RFC 2119. Only emphasized words are to be interpreted in this way.
In Nixpkgs, there are generally three different names associated with a package:
-
The
pname
attribute of the derivation. This is what most users see, in particular when usingnix-env
. -
The variable name used for the instantiated package in
all-packages.nix
, and when passing it as a dependency to other functions. Typically this is called the package attribute name. This is what Nix expression authors see. It can also be used when installing usingnix-env -iA
. -
The filename for (the directory containing) the Nix expression.
Most of the time, these are the same. For instance, the package e2fsprogs
has a pname
attribute "e2fsprogs"
, is bound to the variable name e2fsprogs
in all-packages.nix
, and the Nix expression is in pkgs/os-specific/linux/e2fsprogs/default.nix
.
There are a few naming guidelines:
-
The
pname
attribute should be identical to the upstream package name. -
The
pname
and theversion
attribute must not contain uppercase letters — e.g.,"mplayer"
instead of"MPlayer"
. -
The
version
attribute must start with a digit e.g.,"0.3.1rc2"
. -
If a package is a commit from a repository without a version assigned, then the
version
attribute should be the latest upstream version preceding that commit, followed by-unstable-
and the date of the (fetched) commit. The date must be in"YYYY-MM-DD"
format.
Example: Given a project had its latest releases 2.2
in November 2021, and 3.0
in January 2022, a commit authored on March 15, 2022 for an upcoming bugfix release 2.2.1
would have version = "2.2-unstable-2022-03-15"
.
- If a project has no suitable preceding releases - e.g., no versions at all, or an incompatible versioning / tagging schema - then the latest upstream version in the above schema should be
0
.
Example: Given a project that has no tags / released versions at all, or applies versionless tags like latest
or YYYY-MM-DD-Build
, a commit authored on March 15, 2022 would have version = "0-unstable-2022-03-15"
.
-
Dashes in the package
pname
should be preserved in new variable names, rather than converted to underscores or camel cased — e.g.,http-parser
instead ofhttp_parser
orhttpParser
. The hyphenated style is preferred in all three package names. -
If there are multiple versions of a package, this should be reflected in the variable names in
all-packages.nix
, e.g.json-c_0_9
andjson-c_0_11
. If there is an obvious “default” version, make an attribute likejson-c = json-c_0_9;
. See also versioning.
Versioning
Because every version of a package in Nixpkgs creates a potential maintenance burden, old versions of a package should not be kept unless there is a good reason to do so. For instance, Nixpkgs contains several versions of GCC because other packages don’t build with the latest version of GCC. Other examples are having both the latest stable and latest pre-release version of a package, or to keep several major releases of an application that differ significantly in functionality.
If there is only one version of a package, its Nix expression should be named e2fsprogs/default.nix
. If there are multiple versions, this should be reflected in the filename, e.g. e2fsprogs/1.41.8.nix
and e2fsprogs/1.41.9.nix
. The version in the filename should leave out unnecessary detail. For instance, if we keep the latest Firefox 2.0.x and 3.5.x versions in Nixpkgs, they should be named firefox/2.0.nix
and firefox/3.5.nix
, respectively (which, at a given point, might contain versions 2.0.0.20
and 3.5.4
). If a version requires many auxiliary files, you can use a subdirectory for each version, e.g. firefox/2.0/default.nix
and firefox/3.5/default.nix
.
All versions of a package must be included in all-packages.nix
to make sure that they evaluate correctly.
Meta attributes
meta.description
must:- Be short, just one sentence.
- Be capitalized.
- Not start with the definite or an indefinite article.
- Not start with the package name.
- More generally, it should not refer to the package name.
- Not end with a period (or any punctuation for that matter).
- Provide factual information.
- Avoid subjective language.
meta.license
must be set and match the upstream license.- If there is no upstream license,
meta.license
should default tolib.licenses.unfree
. - If in doubt, try to contact the upstream developers for clarification.
- If there is no upstream license,
meta.mainProgram
must be set to the name of the executable which facilitates the primary function or purpose of the package, if there is such an executable in$bin/bin/
(or$out/bin/
, if there is no"bin"
output).- Packages that only have a single executable in the applicable directory above should set
meta.mainProgram
. For example, the packageripgrep
only has a single executablerg
under$out/bin/
, soripgrep.meta.mainProgram
is set to"rg"
. - Packages like
polkit_gnome
that have no executables in the applicable directory should not setmeta.mainProgram
. - Packages like
e2fsprogs
that have multiple executables, none of which can be considered the main program, should not setmeta.mainProgram
. - Packages which are not primarily used for a single executable do not need to set
meta.mainProgram
. - Always prefer using a hardcoded string (don't use
pname
, for example). - When in doubt, ask for reviewer input.
- Packages that only have a single executable in the applicable directory above should set
meta.maintainers
must be set for new packages.
See the Nixpkgs manual for more details on standard meta-attributes.
Import From Derivation
Import From Derivation (IFD) is disallowed in Nixpkgs for performance reasons: Hydra evaluates the entire package set, and sequential builds during evaluation would increase evaluation times to become impractical.
Import From Derivation can be worked around in some cases by committing generated intermediate files to version control and reading those instead.
Sources
Always fetch source files using Nixpkgs fetchers. Use reproducible sources with a high degree of availability. Prefer protocols that support proxies.
A list of schemes for mirror://
URLs can be found in pkgs/build-support/fetchurl/mirrors.nix
, and is supported by fetchurl
.
Other fetchers which end up relying on fetchurl
may also support mirroring.
The preferred source hash type is sha256
.
Examples going from bad to best practices:
-
Bad: Uses
git://
which won't be proxied.{ src = fetchgit { url = "git://github.com/NixOS/nix.git"; rev = "1f795f9f44607cc5bec70d1300150bfefcef2aae"; hash = "sha256-7D4m+saJjbSFP5hOwpQq2FGR2rr+psQMTcyb1ZvtXsQ="; }; }
-
Better: This is ok, but an archive fetch will still be faster.
{ src = fetchgit { url = "https://github.com/NixOS/nix.git"; rev = "1f795f9f44607cc5bec70d1300150bfefcef2aae"; hash = "sha256-7D4m+saJjbSFP5hOwpQq2FGR2rr+psQMTcyb1ZvtXsQ="; }; }
-
Best: Fetches a snapshot archive for the given revision.
{ src = fetchFromGitHub { owner = "NixOS"; repo = "nix"; rev = "1f795f9f44607cc5bec70d1300150bfefcef2aae"; hash = "sha256-7D4m+saJjbSFP5hOwpQq2FGR2rr+psQMTcyb1ZvtXsQ="; }; }
Note
When fetching from GitHub, always reference revisions by their full commit hash. GitHub shares commit hashes among all forks and returns
404 Not Found
when a short commit hash is ambiguous. It already happened in Nixpkgs for short, 6-character commit hashes.Pushing large amounts of auto generated commits into forks is a practical vector for a denial-of-service attack, and was already demonstrated against GitHub Actions Beta.
Patches
Sometimes, changes are needed to the source to allow building a derivation in nixpkgs, or to get earlier access to an upstream fix or improvement.
When using the patches
parameter to mkDerivation
, make sure the patch name clearly describes the reason for the patch, or add a comment.
Patches already merged upstream or published elsewhere should be retrieved using fetchpatch
.
{
patches = [
(fetchpatch {
name = "fix-check-for-using-shared-freetype-lib.patch";
url = "http://git.ghostscript.com/?p=ghostpdl.git;a=patch;h=8f5d285";
hash = "sha256-uRcxaCjd+WAuGrXOmGfFeu79cUILwkRdBu48mwcBE7g=";
})
];
}
Otherwise, you can add a .patch
file to the nixpkgs
repository.
In the interest of keeping our maintenance burden and the size of nixpkgs to a minimum, only do this for patches that are unique to nixpkgs
or that have been proposed upstream but are not merged yet, cannot be easily fetched or have a high chance to disappear in the future due to unstable or unreliable URLs.
The latter avoids link rot when the upstream abandons, squashes or rebases their change, in which case the commit may get garbage-collected.
If a patch is available online but does not cleanly apply, it can be modified in some fixed ways by using additional optional arguments for fetchpatch
. Check the fetchpatch
reference for details.
{
patches = [ ./0001-changes.patch ];
}
If you do need to do create this sort of patch file, one way to do so is with git:
-
Move to the root directory of the source code you're patching.
$ cd the/program/source
-
If a git repository is not already present, create one and stage all of the source files.
$ git init $ git add .
-
Edit some files to make whatever changes need to be included in the patch.
-
Use git to create a diff, and pipe the output to a patch file:
$ git diff -a > nixpkgs/pkgs/the/package/0001-changes.patch
Deprecating/removing packages
There is currently no policy when to remove a package.
Before removing a package, one should try to find a new maintainer or fix smaller issues first.
Steps to remove a package from Nixpkgs
We use jbidwatcher as an example for a discontinued project here.
-
Have Nixpkgs checked out locally and up to date.
-
Create a new branch for your change, e.g.
git checkout -b jbidwatcher
-
Remove the actual package including its directory, e.g.
git rm -rf pkgs/applications/misc/jbidwatcher
-
Remove the package from the list of all packages (
pkgs/top-level/all-packages.nix
). -
Add an alias for the package name in
pkgs/top-level/aliases.nix
(There is alsopkgs/applications/editors/vim/plugins/aliases.nix
. Package sets typically do not have aliases, so we can't add them there.)For example in this case:
{ jbidwatcher = throw "jbidwatcher was discontinued in march 2021"; # added 2021-03-15 }
The throw message should explain in short why the package was removed for users that still have it installed.
-
Test if the changes introduced any issues by running
nix-env -qaP -f . --show-trace
. It should show the list of packages without errors. -
Commit the changes. Explain again why the package was removed. If it was declared discontinued upstream, add a link to the source.
$ git add pkgs/applications/misc/jbidwatcher/default.nix pkgs/top-level/all-packages.nix pkgs/top-level/aliases.nix $ git commit
Example commit message:
jbidwatcher: remove project was discontinued in march 2021. the program does not work anymore because ebay changed the login. https://web.archive.org/web/20210315205723/http://www.jbidwatcher.com/
-
Push changes to your GitHub fork with
git push
-
Create a pull request against Nixpkgs. Mention the package maintainer.
This is how the pull request looks like in this case: https://github.com/NixOS/nixpkgs/pull/116470
Package tests
To run the main types of tests locally:
- Run package-internal tests with
nix-build --attr pkgs.PACKAGE.passthru.tests
- Run NixOS tests with
nix-build --attr nixosTests.NAME
, whereNAME
is the name of the test listed innixos/tests/all-tests.nix
- Run global package tests with
nix-build --attr tests.PACKAGE
, wherePACKAGE
is the name of the test listed inpkgs/test/default.nix
- See
lib/tests/NAME.nix
for instructions on running specific library tests
Tests are important to ensure quality and make reviews and automatic updates easy.
The following types of tests exists:
- NixOS module tests, which spawn one or more NixOS VMs. They exercise both NixOS modules and the packaged programs used within them. For example, a NixOS module test can start a web server VM running the
nginx
module, and a client VM runningcurl
or a graphicalfirefox
, and test that they can talk to each other and display the correct content. - Nix package tests are a lightweight alternative to NixOS module tests. They should be used to create simple integration tests for packages, but cannot test NixOS services, and some programs with graphical user interfaces may also be difficult to test with them.
- The
checkPhase
of a package, which should execute the unit tests that are included in the source code of a package.
Here in the nixpkgs manual we describe mostly package tests; for module tests head over to the corresponding section in the NixOS manual.
Writing inline package tests
For very simple tests, they can be written inline:
{ /* ... , */ yq-go }:
buildGoModule rec {
# …
passthru.tests = {
simple = runCommand "${pname}-test" {} ''
echo "test: 1" | ${yq-go}/bin/yq eval -j > $out
[ "$(cat $out | tr -d $'\n ')" = '{"test":1}' ]
'';
};
}
Any derivaton can be specified as a test, even if it's in a different file.
Such a derivaton that implements a test can depend on the package under test, even in the presence of overrideAttrs
.
In the following example, (my-package.overrideAttrs f).passthru.tests
will work as expected, as long as the definition of tests
does not rely on the original my-package
or overrides all occurrences of my-package
:
# my-package/default.nix
{ stdenv, callPackage }:
stdenv.mkDerivation (finalAttrs: {
# ...
passthru.tests.example = callPackage ./example.nix { my-package = finalAttrs.finalPackage; };
})
# my-package/example.nix
{ runCommand, lib, my-package, ... }:
runCommand "my-package-test" {
nativeBuildInputs = [ my-package ];
src = lib.sources.sourcesByRegex ./. [ ".*.in" ".*.expected" ];
} ''
my-package --help
my-package <example.in >example.actual
diff -U3 --color=auto example.expected example.actual
mkdir $out
''
Writing larger package tests
This is an example using the phoronix-test-suite
package with the current best practices.
Add the tests in passthru.tests
to the package definition like this:
{ stdenv, lib, fetchurl, callPackage }:
stdenv.mkDerivation {
# …
passthru.tests = {
simple-execution = callPackage ./tests.nix { };
};
meta = { /* … */ };
}
Create tests.nix
in the package directory:
{ runCommand, phoronix-test-suite }:
let
inherit (phoronix-test-suite) pname version;
in
runCommand "${pname}-tests" { meta.timeout = 60; }
''
# automatic initial setup to prevent interactive questions
${phoronix-test-suite}/bin/phoronix-test-suite enterprise-setup >/dev/null
# get version of installed program and compare with package version
if [[ `${phoronix-test-suite}/bin/phoronix-test-suite version` != *"${version}"* ]]; then
echo "Error: program version does not match package version"
exit 1
fi
# run dummy command
${phoronix-test-suite}/bin/phoronix-test-suite dummy_module.dummy-command >/dev/null
# needed for Nix to register the command as successful
touch $out
''
Running package tests
You can run these tests with:
$ cd path/to/nixpkgs
$ nix-build -A phoronix-test-suite.tests
Examples of package tests
Here are examples of package tests:
- Jasmin compile test
- Lobster compile test
- Spacy annotation test
- Libtorch test
- Multiple tests for nanopb
Linking NixOS module tests to a package
Like package tests as shown above, NixOS module tests can also be linked to a package, so that the tests can be easily run when changing the related package.
For example, assuming we're packaging nginx
, we can link its module test via passthru.tests
:
{ stdenv, lib, nixosTests }:
stdenv.mkDerivation {
# ...
passthru.tests = {
nginx = nixosTests.nginx;
};
# ...
}
Automatic package updates
Nixpkgs periodically tries to update all packages that have a passthru.updateScript
attribute.
Note
A common pattern is to use the
nix-update-script
attribute provided in Nixpkgs, which runsnix-update
:{ stdenv, nix-update-script }: stdenv.mkDerivation { # ... passthru.updateScript = nix-update-script { }; }
For simple packages, this is often enough, and will ensure that the package is updated automatically by
nixpkgs-update
when a new version is released. The update bot runs periodically to attempt to automatically update packages, and will runpassthru.updateScript
if set. While not strictly necessary if the project is listed on Repology, usingnix-update-script
allows the package to update via many more sources (e.g. GitHub releases).
The passthru.updateScript
attribute can contain one of the following:
-
an executable file, either on the file system:
{ stdenv }: stdenv.mkDerivation { # ... passthru.updateScript = ./update.sh; }
or inside the expression itself:
{ stdenv, writeScript }: stdenv.mkDerivation { # ... passthru.updateScript = writeScript "update-zoom-us" '' #!/usr/bin/env nix-shell #!nix-shell -i bash -p curl pcre2 common-updater-scripts set -eu -o pipefail version="$(curl -sI https://zoom.us/client/latest/zoom_x86_64.tar.xz | grep -Fi 'Location:' | pcre2grep -o1 '/(([0-9]\.?)+)/')" update-source-version zoom-us "$version" ''; }
-
a list, a script file followed by arguments to be passed to it:
{ stdenv }: stdenv.mkDerivation { # ... passthru.updateScript = [ ../../update.sh pname "--requested-release=unstable" ]; }
-
an attribute set containing:
-
command
A string or list in the format expected by
passthru.updateScript
-
attrPath
(optional)A string containing the canonical attribute path for the package.
If present, it will be passed to the update script instead of the attribute path on which the package was discovered during Nixpkgs traversal.
-
supportedFeatures
(optional)A list of the [extra features the script supports][supported-features].
{ stdenv }: stdenv.mkDerivation rec { pname = "my-package"; # ... passthru.updateScript = { command = [ ../../update.sh pname ]; attrPath = pname; supportedFeatures = [ /* ... */ ]; }; }
-
How are update scripts executed?
Update scripts are to be invoked by the automatic package update script.
You can run nix-shell maintainers/scripts/update.nix
in the root of Nixpkgs repository for information on how to use it.
update.nix
offers several modes for selecting packages to update, and it will execute update scripts for all matched packages that have an updateScript
attribute.
Each update script will be passed the following environment variables:
- [
UPDATE_NIX_NAME
] – content of thename
attribute of the updated package - [
UPDATE_NIX_PNAME
] – content of thepname
attribute of the updated package - [
UPDATE_NIX_OLD_VERSION
] – content of theversion
attribute of the updated package - [
UPDATE_NIX_ATTR_PATH
] – attribute path theupdate.nix
discovered the package on (or the package's specifiedattrPath
when available). Example:pantheon.elementary-terminal
Note
An update script will be usually run from the root of the Nixpkgs repository, but you should not rely on that. Also note that
update.nix
executes update scripts in parallel by default, so you should avoid runninggit commit
or any other commands that cannot handle that.
While update scripts should not create commits themselves, update.nix
supports automatically creating commits when running it with --argstr commit true
.
If you need to customize commit message, you can have the update script implement the commit
feature.
Supported features
-
commit
This feature allows update scripts to ask
update.nix
to create Git commits.When support of this feature is declared, whenever the update script exits with
0
return status, it is expected to print a JSON list containing an object described below for each updated attribute to standard output. Example:[ { "attrPath": "volume_key", "oldVersion": "0.3.11", "newVersion": "0.3.12", "files": [ "/path/to/nixpkgs/pkgs/development/libraries/volume-key/default.nix" ] } ]
:::
When
update.nix
is run with--argstr commit true
, it will create a separate commit for each of the objects. An empty list can be returned when the script did not update any files; for example, when the package is already at the latest version.The commit object contains the following values:
attrPath
– a string containing the attribute patholdVersion
– a string containing the old versionnewVersion
– a string containing the new versionfiles
– a non-empty list of file paths (as strings) to add to the commitcommitBody
(optional) – a string with extra content to be appended to the default commit message (useful for adding changelog links)commitMessage
(optional) – a string to use instead of the default commit message
If the returned list contains exactly one object (e.g.
[{}]
), all values are optional and will be determined automatically.
Reviewing contributions
Package updates
A package update is the most trivial and common type of pull request. These pull requests mainly consist of updating the version part of the package name and the source hash.
It can happen that non-trivial updates include patches or more complex changes.
Reviewing process:
- Ensure that the package versioning fits the guidelines.
- Ensure that the commit text fits the guidelines.
- Ensure that the package maintainers are notified.
- CODEOWNERS will make GitHub notify users based on the submitted changes, but it can happen that it misses some of the package maintainers.
- Ensure that the meta field information fits the guidelines and is correct:
- License can change with version updates, so it should be checked to match the upstream license.
- If the package has no maintainer, a maintainer must be set. This can be the update submitter or a community member that accepts to take maintainership of the package.
- Ensure that the code contains no typos.
- Build the package locally.
-
Pull requests are often targeted to the master or staging branch, and building the pull request locally when it is submitted can trigger many source builds.
-
It is possible to rebase the changes on nixos-unstable or nixpkgs-unstable for easier review by running the following commands from a nixpkgs clone.
$ git fetch origin nixos-unstable $ git fetch origin pull/PRNUMBER/head $ git rebase --onto nixos-unstable BASEBRANCH FETCH_HEAD
- The first command fetches the nixos-unstable branch.
- The second command fetches the pull request changes,
PRNUMBER
is the number at the end of the pull request title andBASEBRANCH
the base branch of the pull request. - The third command rebases the pull request changes to the nixos-unstable branch.
-
The nixpkgs-review tool can be used to review a pull request content in a single command.
PRNUMBER
should be replaced by the number at the end of the pull request title. You can also provide the full github pull request url.$ nix-shell -p nixpkgs-review --run "nixpkgs-review pr PRNUMBER"
-
- Run every binary.
Sample template for a package update review is provided below.
##### Reviewed points
- [ ] package name fits guidelines
- [ ] package version fits guidelines
- [ ] package builds on ARCHITECTURE
- [ ] executables tested on ARCHITECTURE
- [ ] all depending packages build
- [ ] patches have a comment describing either the upstream URL or a reason why the patch wasn't upstreamed
- [ ] patches that are remotely available are fetched rather than vendored
##### Possible improvements
##### Comments
New packages
New packages are a common type of pull requests. These pull requests consists in adding a new nix-expression for a package.
Review process:
- Ensure that all file paths fit the guidelines.
- Ensure that the package name and version fits the guidelines.
- Ensure that the package versioning fits the guidelines.
- Ensure that the commit text fits the guidelines.
- Ensure that the meta fields fits the guidelines and contain the correct information:
- License must match the upstream license.
- Platforms should be set (or the package will not get binary substitutes).
- Maintainers must be set. This can be the package submitter or a community member that accepts taking up maintainership of the package.
- The
meta.mainProgram
must be set if a main executable exists.
- Report detected typos.
- Ensure the package source:
- Uses
mirror://
URLs when available. - Uses the most appropriate functions (e.g. packages from GitHub should use
fetchFromGitHub
).
- Uses
- Build the package locally.
- Run every binary.
Sample template for a new package review is provided below.
##### Reviewed points
- [ ] package path fits guidelines
- [ ] package name fits guidelines
- [ ] package version fits guidelines
- [ ] package builds on ARCHITECTURE
- [ ] executables tested on ARCHITECTURE
- [ ] `meta.description` is set and fits guidelines
- [ ] `meta.license` fits upstream license
- [ ] `meta.platforms` is set
- [ ] `meta.maintainers` is set
- [ ] `meta.mainProgram` is set, if applicable.
- [ ] build time only dependencies are declared in `nativeBuildInputs`
- [ ] source is fetched using the appropriate function
- [ ] the list of `phases` is not overridden
- [ ] when a phase (like `installPhase`) is overridden it starts with `runHook preInstall` and ends with `runHook postInstall`.
- [ ] patches have a comment describing either the upstream URL or a reason why the patch wasn't upstreamed
- [ ] patches that are remotely available are fetched rather than vendored
##### Possible improvements
##### Comments
Security
Submitting security fixes
Security fixes are submitted in the same way as other changes and thus the same guidelines apply.
-
If a new version fixing the vulnerability has been released, update the package;
-
If the security fix comes in the form of a patch and a CVE is available, then add the patch to the Nixpkgs tree, and apply it to the package. The name of the patch should be the CVE identifier, so e.g.
CVE-2019-13636.patch
; If a patch is fetched the name needs to be set as well, e.g.:(fetchpatch { name = "CVE-2019-11068.patch"; url = "https://gitlab.gnome.org/GNOME/libxslt/commit/e03553605b45c88f0b4b2980adfbbb8f6fca2fd6.patch"; hash = "sha256-SEKe/8HcW0UBHCfPTTOnpRlzmV2nQPPeL6HOMxBZd14="; })
If a security fix applies to both master and a stable release then, similar to regular changes, they are preferably delivered via master first and cherry-picked to the release branch.
Critical security fixes may by-pass the staging branches and be delivered directly to release branches such as master
and release-*
.
Vulnerability Roundup
Issues
Vulnerable packages in Nixpkgs are managed using issues. Currently opened ones can be found using the following:
github.com/NixOS/nixpkgs/issues?q=is:issue+is:open+"Vulnerability+roundup"
Each issue correspond to a vulnerable version of a package; As a consequence:
- One issue can contain several CVEs;
- One CVE can be shared across several issues;
- A single package can be concerned by several issues.
A "Vulnerability roundup" issue usually respects the following format:
<link to relevant package search on search.nix.gsc.io>, <link to relevant files in Nixpkgs on GitHub>
<list of related CVEs, their CVSS score, and the impacted NixOS version>
<list of the scanned Nixpkgs versions>
<list of relevant contributors>
Note that there can be an extra comment containing links to previously reported (and still open) issues for the same package.
Triaging and Fixing
Note: An issue can be a "false positive" (i.e. automatically opened, but without the package it refers to being actually vulnerable). If you find such a "false positive", comment on the issue an explanation of why it falls into this category, linking as much information as the necessary to help maintainers double check.
If you are investigating a "true positive":
- Find the earliest patched version or a code patch in the CVE details;
- Is the issue already patched (version up-to-date or patch applied manually) in Nixpkgs's
master
branch?- No:
- Submit a security fix;
- Once the fix is merged into
master
, submit the change to the vulnerable release branch(es);
- Yes: Backport the change to the vulnerable release branch(es).
- No:
- When the patch has made it into all the relevant branches (
master
, and the vulnerable releases), close the relevant issue(s).