Compiled metadata system

Starting with version 2.4 PyObjC provides a more efficient, lazy loading metadata system. This can greatly reduce the memory use and startup time for PyObjC based applications while still providing full access to Cocoa APIs.

Creating a framework wrapper

A framework wrapper with the new metadata system is always a python package. The package contains an “” file that creates the lazy loader, a “” file with the compiled metadata and optionally other modules and extensions.

The general structure of the “” file is:

import sys, objc
import Foundation
from . import _metadata

sys.modules['FrameworkName'] = objc.ObjCLazyModule('FrameworkName',
     _metadata.__dict__, None, {
         '__doc__': __doc__,
         'objc': objc,
         '__path__': __path__,
     }, (Foundation,))

The framework name, identifier and path should be replaced by the correct values for the wrapped framework. The import of “Foundation” can be replaced by imports of other framework this framework relies on (also add those to the last argument of objc.ObjCLazyModule).

Contents of the “_metadata” module

The exact contents of the “_metadata” module will be described later.

Generating the “_metadata” module

The “objective-metadata” project contains a tool for collecting information about frameworks and compiling that information and manual additions into a “_metadata” module.

That project currently is not a stable as I’d like, this documentation will be updated with more information when that changes.

API description

class objc.ObjCLazyModule(name, frameworkIdentifier, frameworkPath, metadict[, inline_list[, initialdict[, parents]]])

A subclass of the built-in module type that adds lazy-loading of values defined in PyObjC metadata.

  • frameworkIdentifier – the bundle_identifier argument for a call to loadBundle()

  • frameworkPath – the bundle_path argument for a call to loadBundle()

  • metadict – the dictionary with metadata, usually the __dict__ of a module generated by the metadata compiler.

  • inline_list – a capsule object with function definitions, see loadFunctionList() for more information.

  • initial_dict – additional values to add to the module dictionary

  • parents – a list of parent modules, the module behaves as if those modules were imported using from parent parent import *, but lazily fetches definitions on first access.


This is the primary entry point for the framework wrappers shipped with PyObjC.