cloudpickle

https://github.com/cloudpipe/cloudpickle/blob/v3.1.0/CHANGES.md#310

Drop support for Python 3.6 and 3.7, add official support for Python 3.12.

A large code base reorganization and simplification was possible after dropping support for older Python versions.

See the changelog for more details:

• https://github.com/cloudpipe/cloudpickle/blob/v3.0.0/CHANGES.md#300

Installation

pip install cloudpickle==0.4.4

Changes Since v0.4.3

• logging.RootLogger can be also pickled
  (PR #160).

Installation

pip install cloudpickle

Changes Since v0.5.2

• Fixed a crash in Python 2 when serializing non-hashable instancemethods of built-in
  types (issue #144).

• itertools objects can also pickled
  (PR #156).

• logging.RootLogger can be also pickled
  (PR #160).

Installation

pip install cloudpickle

Changes Since v0.4.2

• Fixed a regression: AttributeError when loading pickles that hold a
  reference to a dynamically defined class from the __main__ module.
  (issue #131).
• Fixed a crash in Python 2 when serializing non-hashable instancemethods of built-in
  types. (issue #144)

Installation

pip install cloudpickle

Changes Since v0.4.0

• Fixed a crash when pickling dynamic classes whose __dict__ attribute was defined as a property. Most notably, this affected dynamic namedtuples in Python 2. (https://github.com/cloudpipe/cloudpickle/pull/113)
• Cloudpickle now preserves the __module__ attribute of functions (https://github.com/cloudpipe/cloudpickle/pull/118/).
• Fixed a crash when pickling modules that don't have a __package__ attribute (https://github.com/cloudpipe/cloudpickle/pull/116).

Get it while it's briny with

pip install cloudpickle

Ch-ch-ch-changes

• Fix functions with empty cells (https://github.com/cloudpipe/cloudpickle/pull/91)
• Allow pickling Logger objects (https://github.com/cloudpipe/cloudpickle/pull/96)
• Fix crash when pickling dynamic class cycles (https://github.com/cloudpipe/cloudpickle/pull/102)
• Support WeakSets and ABCMeta instances (https://github.com/cloudpipe/cloudpickle/pull/104)
• Ignore "None" modules added to sys.modules (https://github.com/cloudpipe/cloudpickle/pull/107)
• Remove non-standard __transient__ support (https://github.com/cloudpipe/cloudpickle/pull/110)
• Catch exception from pickle.whichmodule() (https://github.com/cloudpipe/cloudpickle/pull/112)

Get it while it's hot with

pip install cloudpickle

Changes since v0.2.2

• Import submodules accessed by pickled functions (https://github.com/cloudpipe/cloudpickle/pull/80)
• Support recursive functions inside closures (https://github.com/cloudpipe/cloudpickle/pull/89, https://github.com/cloudpipe/cloudpickle/pull/90)
• Fix ResourceWarnings and DeprecationWarnings (https://github.com/cloudpipe/cloudpickle/pull/88)
• Assume modules with __file__ attribute are not dynamic (https://github.com/cloudpipe/cloudpickle/pull/85)

Get it while it's hot with

pip install cloudpickle

Changes

• Import submodules accessed by pickled functions (https://github.com/cloudpipe/cloudpickle/pull/80)
• Support recursive functions inside closures (https://github.com/cloudpipe/cloudpickle/pull/89, https://github.com/cloudpipe/cloudpickle/pull/90)
• Fix ResourceWarnings and DeprecationWarnings (https://github.com/cloudpipe/cloudpickle/pull/88)
• Assume modules with __file__ attribute are not dynamic (https://github.com/cloudpipe/cloudpickle/pull/85)

cloudpickle bug fix release v0.1.1

• fixed save_classmethod (#41)
• now allows users to import cloudpickle to dump and load pickled data (#37)
• no more pickling of closed files, was broken on Python 3 (#32)
• more tests!

• Pickle arbitrary functions, classes
• Python 3 support!

Many thanks to the originators of PiCloud and the maintainers of PySpark for their stewardship of cloudpickle over the years!

Note: This was released on an ✈️, in the literal ☁️s.