Release |version|
-----------------

:Release: |version|
:Date: TBD

New Features
------------

None

New Types
---------

* datashape now supports a ``categorical`` type (:issue:`150`).

Experimental Types
------------------

.. warning::

   Experimental types are subject to change.

* Records can now be constructed with the new syntax:
  ``R['field0':type0, 'field1':type1, ...]``
  where each slice object represents a field in the record.
  ``R`` acts as an alias for ``Record`` to make it more pleasant to
  construct these literal types (:issue:`186`).

API Changes
-----------

* ``datashape`` no longer supports Python 2.6 (:issue:`189`).
* ``datashape`` no longer support Python 3.3 (:issue:`191`).
* The default ``repr`` of ``Mono`` subclass now prints out the slot
  names as keyword arguments next to their values (:issue:`188`).
  For example

  Instead of

  .. code-block:: python

     >>> from datashape import Decimal
     >>> Decimal(precision=11, scale=2)
     Decimal(11, 2)

  we have

  .. code-block:: python

     >>> Decimal(precision=11, scale=2)
     Decimal(precision=11, scale=2)
* Fields are now always constructed with ``str`` in Record datashapes
  (:issue:`197`).

Bug Fixes
---------

* Makes the parser recognize ``null`` and ``void`` (:issue:`183`).
* Cache the datashape hash value to avoid potentially expensive recomputation
  during memoization (:issue:`184`).
* Fix discovery of strings that start with things that look like numbers
  (:issue:`190`).
* Makes the parser recognize ``object`` (:issue:`193`).
* Make string field names in Record types have the same string type
  (:issue:`200`).
* Fix the reprs for :class:`~datashape.coretypes.Function` objects
  (:issue:`194`).

Miscellaneous
-------------

None
