https://github.com/hameerabbasi/arrayish

A library for downstream compatibility of Numpy-compatible arrays.
https://github.com/hameerabbasi/arrayish

Last synced: about 1 month ago
JSON representation

A library for downstream compatibility of Numpy-compatible arrays.

• Host: GitHub
• URL: https://github.com/hameerabbasi/arrayish
• Owner: hameerabbasi
• License: bsd-3-clause
• Created: 2018-04-20T05:13:33.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2019-06-14T16:44:30.000Z (over 5 years ago)
• Last Synced: 2024-10-05T00:52:06.289Z (about 1 month ago)
• Language: Python
• Homepage:
• Size: 24.4 KB
• Stars: 11
• Watchers: 8
• Forks: 0
• Open Issues: 2
• Metadata Files:
  • Readme: README.rst
  • License: LICENSE

Awesome Lists containing this project

README

        
Arrayish

========

This repository is being deprecated in favour of ``uarray`` and ``unumpy``.

A library for downstream compatibility of Numpy-compatible arrays.

Usage

-----

For an alternative array implementation

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.. code-block:: python

   import numpy as np

   import arrayish as ai

   class CustomArray(object):

       """Your custom ``ndarray`` compatible implementation."""

       pass

   def dot(a, b, out=None)

       """ Your custom implementation of ``np.dot``. """

       pass

   # Between your own arrays

   ai.dot.add((CustomArray, CustomArray), dot)

   # With Numpy Arrays, if you'd like. Can also pass another function

   ai.dot.add((np.ndarray, CustomArray), dot)

   ai.dot.add((CustomArray, np.ndarray), dot)

For a downstream library willing to support alternative arrays

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.. code-block:: python

   import arrayish as ai

   import numpy as np

   # Instead of this:

   np.dot(a, b)

   # Do this:

   ai.dot(a, b)