Tutorial: Writing Plugins

Plugins are the means to extend the existing functionality of DNF5. Two types of plugins are supported:

• DNF5 Plugins:

  • Active: Used to implement one or more commands.

• LIBDNF5 Plugins:

  • Passive: Used to implement additional logic into the library using hooks.

They are written in the native language of DNF5, which is C++.

However, a libdnf5 plugin can register other libdnf5 plugins, so it is possible to write a plugin that will load and register libdnf5 plugins written in different languages. The experimental python_plugins_loader does this for Python.

In addition, the actions libdnf5 plugin allows configuring any executable to be triggered by a selected plugin hook. See the actions plugin documentation for more details.

Note

Existing plugins from the preceding DNF project are not compatible with the new DNF5. A different API is now used, and they were written in Python, which is not a mandatory component in DNF5.

For detailed information on both types of plugins, refer to their respective pages:

• DNF5 Plugins
  • Command vs Plugin
  • Writing an Active Plugin
  • Building the Binary
  • Delivering to the User
• LIBDNF5 Plugins
  • Writing a Passive Plugin
  • Building the Binary
  • Delivering to the User

Debugging Tips

To test your freshly built DNF5 Plugin, redirect DNF5 to load it by setting the DNF5_PLUGINS_DIR environmental variable to your build directory (e.g., DNF5_PLUGINS_DIR=~/dnf5/build/dnf5-plugins/template_plugin).

Speaking about LIBDNF5 Plugins, utilize the LIBDNF_PLUGINS_CONFIG_DIR environmental variable to configure the directory with the plugin’s configuration. This can also be overridden by the pluginconfpath configuration option. Additionally, set the directory for the plugin binaries with the pluginpath configuration option.

Ensure effective debugging by building the project with debugging symbols (-DCMAKE_BUILD_TYPE=Debug).