Jsonpickle

======
* Remove accidental pin of setuptools to versions below 59. This allows
jsonpickle to build with CPython 3.11 and 3.12 alphas. (424)
* Remove accidental dependency on pytz in pandas tests. (+421)
* Fix issue with decoding bson.bson.Int64 objects (422)

======
* Drop support for CPython<3.7. CPython 3.6 and below have reached EOL
and no longer receive security updates. (375)
* Add support for CPython 3.11. (395) (+396)
* Remove jsonlib and yajl backends (py2 only)
* Add ``include_properties`` option to the pickler. This should only
be used if analyzing generated json outside of Python. (297) (+387)
* Allow the ``classes`` argument to ``jsonpickle.decode`` to be a dict
of class name to class object. This lets you decode arbitrary dumps
into different classes. (148) (+392)
* Fix bug with deserializing `numpy.poly1d`. (391)
* Allow frozen dataclasses to be deserialized. (240)
* Fixed a bug where pickling a function could return a ``None`` module. (399)
* Removed old bytes/quopri and ref decoding abaility from the unpickler.
These were last used in jsonpickle<1. Removing them causes a slight speedup
in unpickling lists (~5%). (+403)
* Fixed a bug with namedtuples encoding on CPython 3.11. (411)
* When using the ``sort_keys`` option for the ``simplejson`` backend,
jsonpickle now produces correct object references with py/id tags. (408)
* Speed up the internal method ``_restore_tags`` by ~10%. This should speed
up unpickling of almost every object.

======

* Classes with a custom ``__getitem__()`` and ``append()``
now pickle properly. (362) (+379)
* Remove the demjson backend, as demjson hasn't been maintained
for 5 years. (+379)
* Added new handler for numpy objects when using unpickleable=False.
(381) (+382)
* Added exception handling for class attributes that can't be accessed.
(301) (+383)
* Added a long-requested on_missing attribute to the Unpickler class.
This lets you choose behavior for when jsonpickle can't find a class
to deserialize to. (190) (193) (+384)
* Private members of ``__slots__`` are no longer skipped when encoding.
Any objects encoded with versions prior to 2.2.0 should still decode
properly. (318) (+385)

======

* Python 3.10 is now officially supported. (+376)
* Benchmarks were added to aid in optimization efforts. (350) (+352)
* ``is_reducible()`` was sped up by ~80%. (+353) (+354)
* ``_restore_tags()`` was sped up by ~100%. Unpickling items
with a lot of tuples and sets will benefit most. Python 2 users
and users deserializing pickles from jsonpickle <= 0.9.6 may see
a slight performance decrease if using a lot of bytes, ref,
and/or repr objects. (+354)
* ``is_iterator()`` was sped up by ~20% by removing an unnecessary
variable assignment. (+354)
* ``jsonpickle.decode`` has a new option, ``v1_decode`` to assist in
decoding objects created in jsonpickle version 1. (364)
* The ``encode()`` documentation has been updated to help sklearn users.
* ``demjson`` has been removed from the test suite. (+374)
* ``SQLALchemy<1.2`` is no longer being tested by jsonpickle.
Users of sqlalchemy + jsonpickle can always use 1.2 or 1.3.
When jsonpickle v3 is released we will add SQLAlchemy 1.4 to
the test suite alongside removal of support for Python 3.5 and earlier.

======
* Major release: the serialized JSON format now preserves dictionary
identity, which is a subtle change in the serialized format. (351)
* Dictionary identity is now preserved. For example, if the same
dictionary appears twice in a list, the reconstituted list
will now contain two references to the same dictionary. (255) (+332)

======
* Patch release to avoid the change in behavior from the preservation
of dict identity. The next release will be v2.0.0. (351)
* This release does *not* include the performance improvements
from v1.5.1.
* Pandas DataFrame objects with multilevel columns are now supported.
(346) (+347)
* Numpy 1.20 is now officially supported. (336)
* Python 3.9 is now officially supported. (+348)
* Achieved a small speedup for _get_flattener by merging type checks. (+349)