Fix PEX environment setup.

[jsirois]

Previously, if a PEX had a custom interpreter, it was not threaded
through to the environments it activated. Add a test that the
designated interpreter is used.

Fixes #522

[jsirois]

With the new test added but the fix not applied, the test failed like so:

$ tox -epy27
...
tests/test_pex.py::test_activate_interpreter_different_from_current Failed to execute PEX file, missing linux_x86_64-cp-27-cp27mu compatible dependencies for:
my-project
...

[jsirois]

Hrm. Locally I was seeing the errors of the form seen in CI - randomly :(:

tests/test_pex.py::test_activate_interpreter_different_from_current **** Failed to install tmpiEkmjW (caused by: NonZeroExit("received exit code 1 during execution of `['/home/travis/build/pantsbuild/pex/.pyenv_test/versions/3.6.3/bin/python3.6', '-', 'bdist_wheel', '--dist-dir=/tmp/tmpnPulvz']` while trying to execute `['/home/travis/build/pantsbuild/pex/.pyenv_test/versions/3.6.3/bin/python3.6', '-', 'bdist_wheel', '--dist-dir=/tmp/tmpnPulvz']`",)
):
stdout:
stderr:
usage: setup.py [global_opts] cmd1 [cmd1_opts] [cmd2 [cmd2_opts] ...]
   or: setup.py --help [cmd1 cmd2 ...]
   or: setup.py --help-commands
   or: setup.py cmd --help
error: invalid command 'bdist_wheel'

On successful runs I could copy the PYTHONHASHSEED and then re-run with tox --hashseed=<value> ... and get consistent success. I'll dig into the randomness in the am.

[jsirois]

OK - this exposes a longstanding bug with extras - order is important when importing setuptools. If wheel is not on the sys.path when it is imported, no wheel distutils commands will be discovered. I'll try a fix to get CI green and then break that out as a seperate PR.

[CMLivingston]

This LGTM, however I hit a test failure on the py36-requests environment:

**** Failed to install tmps18ib62q (caused by: NonZeroExit("received exit code 1 during execution of `['/Users/clivingston/workspace/pex/.pyenv_test/versions/2.7.10/bin/python2.7', '-', 'bdist_wheel', '--dist-dir=/var/folders/7n/x4k4tqh94k9gwm61g046m2340000gn/T/tmp39eob7fs']` while trying to execute `['/Users/clivingston/workspace/pex/.pyenv_test/versions/2.7.10/bin/python2.7', '-', 'bdist_wheel', '--dist-dir=/var/folders/7n/x4k4tqh94k9gwm61g046m2340000gn/T/tmp39eob7fs']`",)
):
stdout:

stderr:
usage: setup.py [global_opts] cmd1 [cmd1_opts] [cmd2 [cmd2_opts] ...]
   or: setup.py --help [cmd1 cmd2 ...]
   or: setup.py --help-commands
   or: setup.py cmd --help

error: invalid command 'bdist_wheel'

I suspect this is specific to my machine, but shouldn't tox be installing wheel for this run?

[jsirois]

@CMLivingston did your test include 2ec7c0b? That has the fix for the error: invalid command 'bdist_wheel'.

FWIW: without the fix, running tox -epy27 -- -vsktest_activate_interpreter_different_from_current is much quicker and would reveal flake. If the PYTHONHASHSEED rolled the right way, setuptools would be inserted on the sys.path and imported before wheel was on the sys.path, and so distutils command discovery in setuptools would fail to see not-yet-present wheel.

[CMLivingston]

Nope, just pulled and it now works for me! Thanks.

[jsirois]

OK great. Thanks for giving this a spin @CMLivingston. I'll split that unrelated fix off to its own PR - deserves a test of its own. I'll then rebase this after that PR hits master.

[jsirois]

And the error: invalid command 'bdist_wheel' issue is old!: #158

[jsirois]

Depends on #532 which is the 1st two commits - only ff01ec0 needs review in this PR.