AppImage provides a way for developers to provide “native” binaries for Linux users. It allow packaging applications for any common Linux based operating system, including Ubuntu, Debian, Fedora, and more. AppImages contain all the dependencies that cannot be assumed to be part of each target system, and will run on most Linux distributions without further modifications. Briefcase uses linuxdeploy to build the AppImage in the correct format.

Packaging binaries for Linux is complicated, because of the inconsistent library versions present on each distribution. An AppImage can be executed on any Linux distribution with a version of libc greater than or equal the version of the distribution where the AppImage was created.

To simplify the packaging process, Briefcase provides a pre-compiled Python support library. This support library was compiled on Ubuntu 18.04, which means the AppImages build by Briefcase can be used on any Linux distribution of about the same age or newer - but those AppImages must be compiled on Ubuntu 18.04.

This means you have four options for using Briefcase to compile a Linux AppImage:

  1. Run the version-sensitive parts of the build process inside Docker. This is the default behavior of Briefcase. This also means that it is possible to build Linux binaries on any platform that can run Docker.

  2. Install Ubuntu 18.04 on your own machine.

  3. Find a cloud or CI provider that can provide you an Ubuntu 18.04 machine for build purposes. Github Actions, for example, provides Ubuntu 18.04 as a build option. Again, you’ll need to use the --no-docker command line option.

  4. Build your own version of the BeeWare Python support libraries. If you take this approach, be aware that your AppImage will only be as portable as the version of libc that is available on the distribution you use. If you build using Ubuntu 19.10, for example, you can expect that only people on the most recent versions of another distribution will be able to run your AppImage.

Icon format#

AppImages uses .png format icons. An application must provide icons in the following sizes:

  • 16px

  • 32px

  • 64px

  • 128px

  • 256px

  • 512px

Splash Image format#

AppImages do not support splash screens or installer images.

Additional options#

The following options can be provided at the command line when producing AppImages.


Use native execution, rather than using Docker to start a container.

Application configuration#

The following options can be added to the<appname>.linux.appimage section of your pyproject.toml file.


A list of operating system packages that must be installed for the AppImage build to succeed. If a Docker build is requested, this list will be passed to the Docker context when building the container for the app build. By default, entries should be Ubuntu 18.04 apt package requirements. For example,

system_requires = [‘libgirepository1.0-dev’, ‘libcairo2-dev’]

would make the GTK GI and Cairo operating system libraries available to your app.

If you see errors during briefcase build of the form:

Could not find dependency:

but the app works under briefcase dev, the problem may be an incomplete system_requires definition. The briefcase build process generates a new environment that is completely isolated from your development environment, so if your app has any operating system dependencies, they must be listed in your system_requires definition.


A list of linuxdeploy plugins that you wish to be included when building the AppImage. This is needed for applications that depend on libraries that have dependencies that cannot be automatically discovered by linuxdeploy. GTK and Qt both have complex runtime resource requirements that can be difficult for linuxdeploy to identify automatically.

The linuxdeploy_plugins declaration is a list of strings. Briefcase can take plugin definitions in three formats:

  1. The name of a plugin known by Briefcase. One of gtk or qt.

  2. A URL where a plugin can be downloaded

  3. A path to a local plugin file

If your plugin requires an environment variable for configuration, that environment variable can be provided as a prefix to the plugin declaration, similar to how environment variables can be defined for a shell command.

For example, the gtk plugin requires the DEPLOY_GTK_VERSION environment variable. To set this variable with the Briefcase-managed GTK linuxdeploy plugin, you would define:

linuxdeploy_plugins = ["DEPLOY_GTK_VERSION=3 gtk"]

Or, if you were using a plugin stored as a local file:

linuxdeploy_plugins = ["DEPLOY_GTK_VERSION=3 path/to/plugins/"]


Any additional Docker instructions that are required to configure the container used to build your Python app. For example, any dependencies that cannot be configured with apt-get could be installed. dockerfile_extra_content is string literal that will be added verbatim to the end of the project Dockerfile.

Any Dockerfile instructions added by dockerfile_extra_content will be executed as the brutus user, rather than the root user. If you need to perform container setup operations as root, switch the container’s user to root, perform whatever operations are required, then switch back to the brutus user - e.g.:

dockerfile_extra_content = """
RUN <first command run as brutus>

USER root
RUN <second command run as root>

USER brutus

Runtime issues with AppImages#

Packaging on Linux is a difficult problem - especially when it comes to binary libraries. The following are some common problems you may see, and ways that they can be mitigated.

Undefined symbol and Namespace not available errors#

If you get the error:

ValueError: Namespace Something not available


ImportError: /usr/lib/ undefined symbol: some_symbol

it is likely that one or more of the libraries you are using in your app requires a linuxdeploy plugin. GUI libraries, or libraries that do dynamic module loading are particularly prone to this problem.

ELF load command address/offset not properly aligned#

Briefcase uses a tool named linuxdeploy to build AppImages. linuxdeploy processes all the libraries used by an app so that they can be relocated into the final packaged binary. Building a manylinux binary wheel involves a tool named auditwheel that performs a very similar process. Unfortunately, processing a binary with linuxdeploy after it has been processed by auditwheel can result in a binary library that cannot be loaded at runtime.

This is particularly common when a module installed as a binary wheel has a dependency on external libraries. For example, Pillow is a Python library that contains a binary submodule; that submodule uses libpng, libtiff, and other system libraries for image manipulation. If you install Pillow from a manylinux wheel, you may see an error similar to the following at runtime:

Traceback (most recent call last):
File "/tmp/.mount_TestbewwDi98/usr/app/testbed/", line 54, in main
File "/tmp/.mount_TestbewwDi98/usr/app/testbed/", line 94, in test_pillow
   from PIL import Image
File "/tmp/.mount_TestbewwDi98/usr/app_packages/PIL/", line 132, in <module>
   from . import _imaging as core
ImportError: ELF load command address/offset not properly aligned

This indicates that one of the libraries that has been included in the AppImage has become corrupted as a result of double processing.

The solution is to ask Briefcase to install the affected library from source. This can be done by adding a "--no-binary" entry to the requires declaration for your app. For example, if your app includes Pillow as a requirement:

requires = ["pillow==9.1.0"]

You can force Briefcase to install Pillow from source by adding:

requires = [
    "--no-binary", "pillow",

Since the library will be installed from source, you also need to add any system requirements that are needed to compile the binary library. For example, Pillow requires the development libraries for the various image formats that it uses:

system_requires = [
    ... other system requirements ...

If you are missing a system requirement, the call to briefcase build will fail with an error:

error: subprocess-exited-with-error

× pip subprocess to install build dependencies did not run successfully.
│ exit code: 1
╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.
>>> Return code: 1

Unable to install requirements. This may be because one of your
requirements is invalid, or because pip was unable to connect
to the PyPI server.

You must add a separate --no-binary option for every binary library you want to install from source. For example, if your app also includes the cryptography library, and you want to install that library from source, you would add:

requires = [
    "--no-binary", "pillow",
    "--no-binary", "cryptography",

If you want to force all packages to be installed from source, you can add a single :all declaration:

requires = [
    "--no-binary", ":all:",

The --no-binary declaration doesn’t need to be added to the same requires declaration that defines the requirement. For example, if you have a library that is used on all platforms, the declaration will probably be in the top-level requires, not the platform-specific requires. If you add --no-binary in the top-level requires, the use of a binary wheel would be prevented on all platforms. To avoid this, you can add the requirement in the top-level requires, but add the --no-binary declaration to the linux-specific requirements:

formal_name = "Hello World"
requires = [

requires = [
    "--no-binary", "pillow"