pyside6-deploy: the deployment tool for Qt for Python#

pyside6-deploy is an easy to use tool for deploying PySide6 applications to different platforms. It is a wrapper around Nuitka, a Python compiler that compiles your Python code to C code, and links with libpython to produce the final executable.

The final executable produced has a .exe suffix on Windows, .bin on Linux and .app on macOS.

Note

Although using a virtual environment for Python is recommended for pyside6-deploy, do not add the virtual environment to the application directory you are trying to deploy. pyside6-deploy will try to package this venv folder and will eventually fail.

How to use it?#

There are 2 different ways with which you can deploy your PySide6 application using pyside6-deploy:

Approach 1: Using the main python entry point file#

In this approach, you point pyside6-deploy to the file containing the main Python entry point file of the project i.e. the file containing if __name__ == "__main__":. The command looks like this:

pyside6-deploy /path/to/main_file.py

On running the command, pyside6-deploy installs all the dependencies required for deployment into the Python environment.

If your main Python entry point file is named main.py, then you don’t have to point it to the filename. You can run pyside6-deploy without any options, and it will work.

Note

If your project contains a pysidedeploy.spec file, which is generated on the first run of pyside6-deploy on the project directory, then for any subsequent runs of pyside6-deploy you can run pyside6-deploy without specifying the main Python entry point file. It would take the path to the main file from the pysidedeploy.spec file. To know more about what deployment parameters are controlled by pysidedeploy.spec file, read pysidedeploy.

Approach 2: Using pysidedeploy.spec config file#

When you run pyside6-deploy for the first time, it creates a file called pysidedeploy.spec in the project directory. This file controls various parameters that influence the deployment process. Any subsequent runs of pyside6-deploy on the project directory, would not require additional parameters like the main Python entry point file. You can also point pyside6-deploy to the path of the pysidedeploy.spec file (in case it is not in the same directory), to take the parameters from that file. This can be done with the following command:

pyside6-deploy -c /path/to/pysidedeploy.spec

pysidedeploy.spec#

As mentioned in the Approach 2 above, you can use this file to control the various parameters of the deployment process. The file has multiple sections, with each section containing multiple keys (parameters being controlled) assigned to a value. The advantages of such a file are two folds:

  1. Using the command line, you can control the deployment parameters without specifying them each time. It is saved permanently in a file, and any subsequent runs much later in time would enable the user to be aware of their last deployment parameters.

  2. Since these parameters are saved into a file, they can be checked into version control. This gives the user more control of the deployment process. For example, when you decide to exclude more QML plugins, or want to include more Nuitka options into your executable.

This file is also used by the pyside6-android-deploy tool as a configuration file. The advantage here is that you can have one single file to control deployment to all platforms.

The relevant parameters for pyside6-deploy are:

app
  • title: The name of the application

  • project_dir: Project directory. The general assumption made is that the project directory is the parent directory of the main Python entry point file

  • input_file: Path to the main Python entry point file

  • project_file: If it exists, this points to the path to the Qt Creator Python Project File .pyproject file. Such a file makes sure that the deployment process never considers unnecessary files when bundling the executable.

  • exec_directory: The directory where the final executable is generated.

  • icon: The icon used for the application. For Windows, the icon image should be of .ico format, for macOS it should be of .icns format, and for linux all standard image formats are accepted.

python
  • python_path: Path to the Python executable. It is recommended to run the deployment process inside a virtual environment as certain python packages will be installed onto the Python environment.

  • packages: The Python packages installed into the Python environment for deployment to work. By default, the Python packages nuitka, ordered_set and zstandard are installed. If the deployment platform is Linux-based, then patchelf is also installed

qt
  • qml_files: Comma-separated paths to all the QML files bundled with the executable

  • excluded_qml_plugins: The problem with using Nuitka for QML deployment is that all the QML plugins are also bundled with the executable. When the plugins are bundled, the binaries of the plugin’s Qt module are also packaged. For example, size heavy module like QtWebEngine also gets added to your executable, even when you do not use it in your code. The excluded_qml_plugins parameter helps you to explicitly specify which all QML plugins are excluded. pyside6-deploy automatically checks the QML files against the various QML plugins and excludes the following Qt modules if they don’t exist:

    QtQuick, QtQuick3D, QtCharts, QtWebEngine, QtTest, QtSensors
    

    The reason why only the presence of the above 6 Qt modules is searched for is because they have the most size heavy binaries among all the Qt modules. With this, you can drastically reduce the size of your executables.

  • modules: Comma-separated list of all the Qt modules used by the application. Just like the other configuration options in pysidedeploy.spec, this option is also computed automatically by pyside6-deploy. However, if the user wants to explicitly include certain Qt modules, the module names can be appended to this list without the Qt prefix. e.g. Network instead of QtNetwork

  • plugins: Comma-separated list of all the Qt plugins used by the application. Just like the other configuration options in pysidedeploy.spec, this option is also computed automatically by pyside6-deploy. However, if the user wants to explicitly include certain Qt plugins, the plugin names can be appended to this list. To see all the plugins bundled with PySide6, see the plugins folder in the site-packages on your Python where PySide6 is installed. The plugin name correspond to their folder name.

nuitka
  • macos.permissions: Only relevant for macOS. This option lists the permissions used by the macOS application, as found in the Info.plist file of the macOS application bundle, using the so-called UsageDescription strings. The permissions are normally automatically found by pyside6-deploy. However the user can also explicitly specify them using the format <UsageDescriptionKey>:<Short Description>. For example, the Camera permission is specified as:

    NSCameraUsageDescription:CameraAccess
    
  • extra_args: Any extra Nuitka arguments specified. It is specified as space-separated command line arguments i.e. just like how you would specify it when you use Nuitka through the command line. By default, it contains the following arguments:

    --quiet --noinclude-qt-translations=True
    

Command Line Options#

The most important command line options are the path to the main Python entry point file and the pysidedeploy.spec file. If neither of these files exists or their command line options are given, then pyside6-deploy assumes that your current working directory does not contain a PySide6 project.

Here are all the command line options of pyside6-deploy:

  • main entry point file: This option does not have a name or a flag and is not restricted by it. This enables pyside6-deploy to be used like:

    pyside6-deploy /path/to/main_file.py
    
  • -c/–config-file: This option is used to specify the path to pysidedeploy.spec explicitly

  • –init: Used to only create the pysidedeploy.spec file Usage:

    pyside6-deploy /path/to/main --init
    
  • -v/–verbose: Runs pyside6-deploy in verbose mode.

  • –dry-run: Displays the final Nuitka command being run.

  • –keep-deployment-files: When this option is added, it retains the build folders created by

    Nuitka during the deployment process.

  • -f/–force: When this option is used, it forces through all the input prompts. pyside6-deploy prompts the user to create a Python virtual environment, if not already in one. With this option, the current Python environment is used irrespective of whether the current Python environment is a virtual environment or not.

  • –name: Application name.

  • –extra-ignore-dirs: Comma-separated directory names inside the project directory. These directories will be skipped when searching for Python files relevant to the project.

  • –extra-modules: Comma-separated list of Qt modules to be added to the application, in case they are not found automatically. The module name can either be specified by omitting the prefix of Qt or including it eg: both Network and QtNetwork works.

Considerations#

For deployment to work efficiently by bundling only the necessary plugins, the following utilities are required to be installed on the system:

OS

Dependencies

Installation

Windows

dumpbin

Shipped with MSVC. Run vcvarsall.bat to add it to PATH

Linux

readelf

Available by default

macOS

dyld_info

Available by default from macOS 12 and upwards