Arch package guidelines

From ArchWiki

When building packages for Arch Linux, adhere to the package guidelines below, especially if the intention is to contribute a new package to Arch Linux. You should also see the PKGBUILD(5) and makepkg(8) manpages.

PKGBUILD prototype

# Maintainer: Your Name <>
md5sums=() #autofill using updpkgsums

build() {
  cd "$pkgname-$pkgver"

  ./configure --prefix=/usr

package() {
  cd "$pkgname-$pkgver"

  make DESTDIR="$pkgdir/" install

Other prototypes are found in /usr/share/pacman/ from the pacman package.

Package etiquette

  • Packages should never be installed to /usr/local/
  • Do not introduce new variables or functions into PKGBUILD build scripts, unless the package cannot be built without doing so, as these could possibly conflict with variables and functions used in makepkg itself.
  • If a new variable or a new function is absolutely required, prefix its name with an underscore (_), e.g.
  • Avoid using /usr/libexec/ for anything. Use /usr/lib/$pkgname/ instead.
  • The packager field from the package meta file can be customized by the package builder by modifying the appropriate option in the /etc/makepkg.conf file, or alternatively override it by creating ~/.makepkg.conf.
  • Do not use makepkg subroutines (e.g. error, msg, msg2, plain, warning) as they might change at any time. To print data, use printf or echo.
  • All important messages should be echoed during install using an .install file. For example, if a package needs extra setup to work, directions should be included.
  • Dependencies are the most common packaging error. Please take the time to verify them carefully, for example by running ldd on dynamic executables, checking tools required by scripts or looking at the documentation of the software. The namcap utility can help you in this regard. This tool can analyze both PKGBUILD and the resulting package tarball and will warn you about bad permissions, missing dependencies, redundant dependencies, and other common mistakes.
  • Any optional dependencies that are not needed to run the package or have it generally function should not be included in the depends array; instead the information should be added to the optdepends array:
optdepends=('cups: printing support'
            'sane: scanners support'
            'libgphoto2: digital cameras support'
            'alsa-lib: sound support'
            'giflib: GIF images support'
            'libjpeg: JPEG images support'
            'libpng: PNG images support')
The above example is taken from the wine package. The optdepends information is automatically printed out on installation/upgrade so one should not keep this kind of information in .install files.
  • When creating a package description for a package, do not include the package name in a self-referencing way. For example, "Nedit is a text editor for X11" could be simplified to "A text editor for X11". Also try to keep the descriptions to ~80 characters or less.
  • Try to keep the line length in the PKGBUILD below ~100 characters.
  • Where possible, remove empty lines from the PKGBUILD (provides, replaces, etc.)
  • It is common practice to preserve the order of the PKGBUILD fields as shown above. However, this is not mandatory, as the only requirement in this context is correct bash syntax.
  • Quote variables which may contain spaces, such as "$pkgdir" and "$srcdir".
  • To ensure the integrity of packages, make sure that the integrity variables contain correct values. These can be updated using the updpkgsums(8) tool.

Package naming

  • Package names can contain only alphanumeric characters and any of @, ., _, +, -. Names are not allowed to start with hyphens or dots. All letters should be lowercase.
  • Package names should not be suffixed with the upstream major release version number (e.g. we do not want libfoo2 if upstream calls it libfoo v2.3.4) in case the library and its dependencies are expected to be able to keep using the most recent library version with each respective upstream release. However, for some software or dependencies, this can not be assumed. In the past this has been especially true for widget toolkits such as GTK and Qt. Software that depends on such toolkits can usually not be trivially ported to a new major version. As such, in cases where software can not trivially keep rolling alongside its dependencies, package names should carry the major version suffix (e.g. gtk2, gtk3, qt4, qt5). For cases where most dependencies can keep rolling along the newest release but some cannot (for instance closed source that needs libpng12 or similar), a deprecated version of that package might be called libfoo1 while the current version is just libfoo.

Package versioning

  • Package versions (i.e. PKGBUILD#pkgver) should be the same as the version released by the author. Versions can include letters if need be (eg, nmap's version is 2.54BETA32). Version tags may not include hyphens! Letters, numbers, and periods only.
  • Package releases (i.e. PKGBUILD#pkgrel) are specific to Arch Linux packages. These allow users to differentiate between newer and older package builds. When a new package version is first released, the release count starts at 1. Then as fixes and optimizations are made, the package will be re-released to the Arch Linux public and the release number will increment. When a new version comes out, the release count resets to 1. Package release tags follow the same naming restrictions as version tags.

Package dependencies

Package relations

Package sources

  • HTTPS sources (https:// for tarballs, git+https:// for git sources) should be used wherever possible
  • Sources should be verified using PGP signatures wherever possible (this might entail building from a git tag instead of a source tarball, if upstream signs commits and tags but not the tarballs)

The factual accuracy of this article or section is disputed.

Reason: commit# is not required in recent pacman as proper checksum is supported for git sources. See[1]. gitea package also has been updated to use below approach (Discuss in Talk:Arch package guidelines)
  • When building from a git tag, use its object hash obtained from git rev-parse instead of the tag name:
_tag=1234567890123456789012345678901234567890 # git rev-parse "v$pkgver"

pkgver() {
    cd "$pkgname"
    git describe
An example for this approach can be found in the gitea package. The reason for this practice is that tags can be force pushed to change the commit that they are pointing to, which would alter the built package. Using the tag object hash ensures the integrity of the sources because force pushing the tag changes its hash. Using a pkgver() function prevents accidentally bumping pkgver without updating _tag as well.
  • Do not diminish the security or validity of a package (e.g. by removing a checksum check or by removing PGP signature verification), because an upstream release is broken or suddenly lacks a certain feature (e.g. PGP signature missing for a new release)
  • Sources have to be unique in srcdir (this might require renaming them when downloading, e.g. "${pkgname}-${pkgver}.tar.gz::https://${pkgname}.tld/download/${pkgver}.tar.gz")
  • Avoid using specific mirrors (e.g. on sourceforge) to download, as they might become unavailable

Working with upstream

It is considered best-practice to work closely with upstream wherever possible. This entails reporting problems about building and testing a package.

  • Report problems to upstream right away.
  • Upstream patches wherever possible.
  • Add comments with links to relevant (upstream) bug tracker tickets in the PKGBUILD (this is particularly important, as it ensures, that other packagers can understand changes and work with a package as well).

It is recommended to track upstream with tools such as nvchecker or urlwatch to be informed about new stable releases.


  • Configuration files should be placed in the /etc directory. If there is more than one configuration file, it is customary to use a subdirectory in order to keep the /etc area as clean as possible. Use /etc/pkg where pkg is the name of the package (or a suitable alternative, eg, apache uses /etc/httpd/).
  • Package files should follow these general directory guidelines:
/etc System-essential configuration files
/usr/bin Binaries
/usr/lib Libraries
/usr/include Header files
/usr/lib/pkg Modules, plugins, etc.
/usr/share/doc/pkg Application documentation
/usr/share/info GNU Info system files
/usr/share/licenses/pkg Application licenses
/usr/share/man Manpages
/usr/share/pkg Application data
/var/lib/pkg Persistent application storage
/etc/pkg Configuration files for pkg
/opt/pkg Large self-contained packages
  • Packages should not contain any of the following directories:
    • /bin
    • /sbin
    • /dev
    • /home
    • /srv
    • /media
    • /mnt
    • /proc
    • /root
    • /selinux
    • /sys
    • /tmp
    • /var/tmp
    • /run

Makepkg duties

When makepkg is used to build a package, it does the following automatically:

  1. Checks if package dependencies and makedepends are installed
  2. Downloads source files from servers
  3. Checks the integrity of source files
  4. Unpacks source files
  5. Does any necessary patching
  6. Builds the software and installs it in a fake root
  7. Strips symbols from binaries
  8. Strips debugging symbols from libraries
  9. Compresses manual and/or info pages
  10. Generates the package meta file which is included with each package
  11. Compresses the fake root into the package file
  12. Stores the package file in the configured destination directory (i.e. the current working directory by default)


The arch array should contain 'x86_64' if the compiled package is architecture-specific. Otherwise, use 'any' for architecture independent packages.


See PKGBUILD#license.

Reproducible builds

Arch is working on making all packages reproducible. A packager can check if a package is reproducible with makerepropkg from devtools or repro from archlinux-repro.

$ makerepropkg $pkgname-1-1-any.pkg.tar.zst


$ repro -f $pkgname-1-1-any.pkg.tar.zst

If the timestamp is required at build-time, use the environment variable SOURCE_DATE_EPOCH. The format is documented upstream.

Additional guidelines

Be sure to read the above guidelines first - important points are listed on this page that will not be repeated in the following guideline pages. These specific guidelines are intended as an addition to the standards listed on this page.

Arch package guidelines

32-bitCLRCMakeCrossDKMSEclipseElectronFontFree PascalGNOMEGoHaskellJavaKDEKernel modulesLispMesonMinGWNode.jsNonfreeOCamlPerlPHPPythonRRubyRustShellVCSWebWine

Packages submitted to the AUR must additionally comply with AUR submission guidelines.