Shuffling tests
===============

By default, every time you launch the testrunner it will run the tests in a
specific order.  However, if you want to ensure that your tests are well
isolated then running them in a varying order can be helpful. (Proper
isolation meaning your tests don't depend on each others outcomes or side
effects and thus the setup and tear down methods are written properly.)

The ``--shuffle`` option tells the test runner to shuffle the list of tests
before running them:

    >>> import os.path, sys
    >>> directory_with_tests = os.path.join(this_directory, 'testrunner-ex')
    >>> defaults = [
    ...     '--path', directory_with_tests,
    ...     '--tests-pattern', '^sampletestsf?$',
    ...     ]

    >>> from zope.testing import testrunner
    >>> default_argv = 'test -u -m sample1 -t test_y0 --list-tests '

Running shuffled tests
----------------------

When specifying the ``--shuffle`` option tests are ordered differently each
time you run the tests:

    >>> argv = (default_argv + '--shuffle').split()
    >>> testrunner.run_internal(defaults, argv)
    Tests were shuffled using seed number ...
    Listing zope.testing.testrunner.layer.UnitTests tests:
    ...
    False

Note: The runner prints out a new piece of information which is the seed number used
to generate the shuffled list of tests. This seed number can later be used to
re-run the tests in exactly the same order to support debugging.

Specifying a seed number to control tests shuffling
---------------------------------------------------

Along with the ``--shuffle`` option comes the ``--shuffle-seed`` option which
takes a seed number as an argument. If you want to reproduce test failures
that happened during a randomly shuffled test run then simply write down the
seed number that was printed out and run it again using the ``--shuffle-seed``
option. The order is guaranteed to be the same.

For example, using the seed number 0 will give us the following, stable, list of
tests:

    >>> argv = (default_argv + '--shuffle --shuffle-seed 0').split()
    >>> testrunner.run_internal(defaults, argv)
    Tests were shuffled using seed number 0.
    Listing zope.testing.testrunner.layer.UnitTests tests:
      test_y0 (sample1.sampletestsf.TestA)
      test_y0 (sample1.sampletests.test_one)
      test_y0 (sample1.sampletests.test1.TestA)
      test_y0 (sample1.sampletestsf)
      test_y0 (sample1.sampletests.test_one.TestA)
      test_y0 (sample1.sample13.sampletests)
      test_y0 (sample1.sample13.sampletests.TestA)
      test_y0 (sample1.sample11.sampletests)
      test_y0 (sample1.sample11.sampletests.TestA)
      test_y0 (sample1.sampletests.test1)
    False
    >>> testrunner.run_internal(defaults, argv)
    Tests were shuffled using seed number 0.
    Listing zope.testing.testrunner.layer.UnitTests tests:
      test_y0 (sample1.sampletestsf.TestA)
      test_y0 (sample1.sampletests.test_one)
      test_y0 (sample1.sampletests.test1.TestA)
      test_y0 (sample1.sampletestsf)
      test_y0 (sample1.sampletests.test_one.TestA)
      test_y0 (sample1.sample13.sampletests)
      test_y0 (sample1.sample13.sampletests.TestA)
      test_y0 (sample1.sample11.sampletests)
      test_y0 (sample1.sample11.sampletests.TestA)
      test_y0 (sample1.sampletests.test1)
    False

Whereas using the seed number 42 will give us the following, different but
stable, list of tests:

    >>> argv = (default_argv + '--shuffle --shuffle-seed 42').split()
    >>> testrunner.run_internal(defaults, argv)
    Tests were shuffled using seed number 42.
    Listing zope.testing.testrunner.layer.UnitTests tests:
      test_y0 (sample1.sample13.sampletests.TestA)
      test_y0 (sample1.sample13.sampletests)
      test_y0 (sample1.sampletests.test1)
      test_y0 (sample1.sampletests.test1.TestA)
      test_y0 (sample1.sample11.sampletests.TestA)
      test_y0 (sample1.sampletestsf)
      test_y0 (sample1.sampletests.test_one)
      test_y0 (sample1.sample11.sampletests)
      test_y0 (sample1.sampletestsf.TestA)
      test_y0 (sample1.sampletests.test_one.TestA)
    False
    >>> testrunner.run_internal(defaults, argv)
    Tests were shuffled using seed number 42.
    Listing zope.testing.testrunner.layer.UnitTests tests:
      test_y0 (sample1.sample13.sampletests.TestA)
      test_y0 (sample1.sample13.sampletests)
      test_y0 (sample1.sampletests.test1)
      test_y0 (sample1.sampletests.test1.TestA)
      test_y0 (sample1.sample11.sampletests.TestA)
      test_y0 (sample1.sampletestsf)
      test_y0 (sample1.sampletests.test_one)
      test_y0 (sample1.sample11.sampletests)
      test_y0 (sample1.sampletestsf.TestA)
      test_y0 (sample1.sampletests.test_one.TestA)
    False

Selecting a seed number without ``--shuffle``
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Note that the ``--shuffle-seed`` option must be used along with ``--shuffle``
option or tests will not be re-ordered:

    >>> argv = (default_argv + '--shuffle-seed 42').split()
    >>> testrunner.run_internal(defaults, argv)
    Listing zope.testing.testrunner.layer.UnitTests tests:
      test_y0 (sample1.sampletestsf.TestA)
      test_y0 (sample1.sampletestsf)
      test_y0 (sample1.sample11.sampletests.TestA)
      test_y0 (sample1.sample11.sampletests)
      test_y0 (sample1.sample13.sampletests.TestA)
      test_y0 (sample1.sample13.sampletests)
      test_y0 (sample1.sampletests.test1.TestA)
      test_y0 (sample1.sampletests.test1)
      test_y0 (sample1.sampletests.test_one.TestA)
      test_y0 (sample1.sampletests.test_one)
    False