external_python_setuptools/setuptools/tests/test_namespaces.py, branch 38.2.5

Tests: Run `python -m pip.main` to support Python 2.6

2017-09-20T12:52:48+00:00

See https://bugs.python.org/issue2751

Tests: Run `python -m pip` instead of plain `pip`

2017-09-20T12:03:29+00:00

This allows to run the testsuite on systems where `/usr/bin/pip`
is a specific Python version, but the tests are run on a different one.

For example in Fedora RPM, when the Python 3 tests are invoked,
they should use Python 3 pip, not Python 2 `/usr/bin/pip`.

Unlike other approaches, like using `pip2`/`pip3` which is currently
done in Fedora RPM (downstream patch), this way it Works Everywhere™
and the downstream patch can be dropped.

See https://src.fedoraproject.org/rpms/python-setuptools/blob/54eaa03a4dc97f93a5e4c92c55e580a4ab55a058/f/0001-Run-test-on-a-version-specific-pip.patch

Re-use test.paths_on_pythonpath to extend the PYTHONPATH variable rather than erasing it. When tests are run under pytest-runner (or other setup.py test invocations), the PYTHONPATH is carefully curated to include dependencies and the project under test. Overwriting PYTHONPATH will break tests in those environments. Fixes #884.

2016-12-22T14:31:40+00:00

Skip again on appveyor

2016-12-14T02:17:10+00:00

Add test capturing expectation when a package is both installed and in the current working directory. Ref #885.

2016-12-14T02:06:00+00:00

Add test attempting to capture failure, but it passes. Ref #885.

2016-12-13T00:47:05+00:00

Expect failure on Python 3.4 and earlier as module_from_spec isn't available. Ref #250.

2016-12-11T21:18:36+00:00

cast the value to a bool so pytest doesn't try to eval it

2016-12-01T16:30:26+00:00

Evaluate the expression directly. Workaround for #860.

2016-11-30T00:20:05+00:00

Skip failing test on appveyor until the cause can be uncovered. Ref #851.

2016-11-18T21:07:10+00:00