Run py.test test in different process

Question

I'm trying to test tensorflow program. I'm setting up tensorflow session using parametrized py.test fixture:

@pytest.fixture(scope="session", params=configuration)
def session(request):
    if request.param == 'tensorflow':
        return tf.Session()
    elif request.param == 'tensorflow-eager':
        tfe.enable_eager_execution()
        return tf.Session()
    elif ...

Tensorflow has global state, thus several test launches can pollute it. For example there is no way to disable eager execution after it has been enabled. Is there a way to instruct py.test to create a new process for each test? Or another way to configure environment for the test besides using parametrized fixture? Example usage:

@pytest.mark.parametrize("bias_type", ['variable', 'ndarray', 'list', 'tuple'])
@pytest.mark.parametrize("kernel_type", ['variable', 'ndarray', 'list', 'tuple'])
@pytest.mark.parametrize("input_type", ['variable', 'ndarray', 'list', 'tuple'])
def test_convolution(session, input_type, kernel_type, bias_type):
    ...

Answer 1

As suggested in the comments, using pytest-xdist will be the solution. The plugin is designed for parallel or distributed execution of tests (even a multi-platform execution is possible), but is well suited to serve your request of running each test in a separate process - you can achieve this with the --forked argument.

Disclaimer

--forked argument will not work on Windows, because Windows doesn't support the fork-exec model and doesn't ship any replacement for fork().

A quick demo

Let's define a fixture that will attempt to turn the eager execution on before running each test:

from tensorflow.contrib.eager.python import tfe

import pytest


@pytest.fixture(scope='function', autouse=True)
def eager(request):
    tfe.enable_eager_execution()

This fixture will obviously fail all tests but the first one since the eager execution can be turned only once. With some dummy tests:

def test_spam():
    assert True

def test_eggs():
    assert True

def test_bacon():
    assert True

Running plain pytest fails as expected:

$ pytest -v
============================== test session starts ================================
platform darwin -- Python 3.6.3, pytest-3.3.1, py-1.5.2, pluggy-0.6.0 -- /Users/hoefling/.virtualenvs/stackoverflow/bin/python3.6
cachedir: .cache
rootdir: /Users/hoefling/projects/private/stackoverflow/so-48234032, inifile:
plugins: forked-0.2, mock-1.6.3, hypothesis-3.44.4
collected 3 items

test_spam.py::test_spam PASSED                                                [ 33%]
test_spam.py::test_eggs ERROR                                                 [ 66%]
test_spam.py::test_bacon ERROR                                                [100%]

...
E       ValueError: Do not call tfe.enable_eager_execution more than once in the
same process. Note eager-mode methods such as tfe.run() also call 
tfe.enable_eager_execution.
...

Now install pytest-xdist:

$ pip install pytest-xdist

and rerun the tests:

$ pytest -v --forked
============================== test session starts ================================
platform darwin -- Python 3.6.3, pytest-3.3.1, py-1.5.2, pluggy-0.6.0 -- /Users/hoefling/.virtualenvs/stackoverflow/bin/python3.6
cachedir: .cache
rootdir: /Users/hoefling/projects/private/stackoverflow/so-48234032, inifile:
plugins: forked-0.2, xdist-1.22.0, mock-1.6.3, hypothesis-3.44.4
collected 3 items

test_spam.py::test_spam PASSED                                                [ 33%]
test_spam.py::test_eggs PASSED                                                [ 66%]
test_spam.py::test_bacon PASSED                                               [100%]

============================= 3 passed in 6.09 seconds ============================

The tests still run sequentially, but each one in an own subprocess, so none of them fails.

Now you can start experimenting with parallel execution, e.g.

$ pytest -v --forked --numprocesses=auto

etc. Refer to plugin docs for more info and more usage examples.