|Title:||Including mock in the Standard Library|
|Last-Modified:||2012-05-01 23:37:34 +1000 (Tue, 01 May 2012)|
|Author:||Michael Foord <michael at python.org>|
This PEP proposes adding the mock  testing library to the Python standard library as unittest.mock.
Creating mock objects for testing is a common need in Python. Many developers create ad-hoc mocks, as needed, in their test suites. This is currently what we do in the Python test suite, where a standardised mock object library would be helpful.
There are many mock object libraries available for Python . Of these, mock is overwhelmingly the most popular, with as many downloads on PyPI as the other mocking libraries combined.
An advantage of mock is that it is a mocking library and not a framework. It provides a configurable and flexible mock object, without being opinionated about how you write your tests. The mock api is now well battle-tested and stable.
mock also handles safely monkeypatching and unmonkeypatching objects during the scope of a test. This is hard to do safely and many developers / projects mimic this functionality (often incorrectly). A standardised way to do this, handling the complexity of patching in the presence of the descriptor protocol (etc) is useful. People are asking for a "patch"  feature to unittest. Doing this via mock.patch is preferable to re-implementing part of this functionality in unittest.
Addition of mock to the Python standard library was discussed and agreed to at the Python Language Summit 2012.
As of release 0.8, which is current at the time of writing, mock is compatible with Python 2.4-3.2. Moving into the Python standard library will allow for the removal of some Python 2 specific "compatibility hacks".
mock 0.8 introduced a new feature, "auto-speccing", obsoletes an older mock feature called "mocksignature". The "mocksignature" functionality can be removed from mock altogether prior to inclusion.
|||mock library on PyPI|
This document has been placed in the public domain.