python setup.py test -n option is broken

[cdeil]

python setup.py test --help claims that one can use the -n option instead of --parallel:

$ python setup.py test --help
...
  --parallel (-n)         Run the tests in parallel on the specified number of
                          CPUs.  If negative, all the cores on the machine
                          will be used.  Requires the pytest-xdist plugin.
...

But actually this doesn't work with current astropy master:

$ python setup.py test -n
...
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "astropy/__init__.py", line 196, in test
    skip_docs=skip_docs)
  File "astropy/tests/helper.py", line 231, in run_tests
    "parallel must be an int, got {0}".format(parallel))
ValueError: parallel must be an int, got 

$ python setup.py test -n 4
invalid command name '4'

$ python setup.py test -n4
error: option -4 not recognized

Am I using it incorrectly or is this broken.
(I guess the -n shortcut could simply be removed?)

[mdboom]

Ah -- I see why this isn't working: it conflicts with the distutils built-in option of "dry-run", which is also "-n". We could change this to -j, to be consistent with make.

[embray]

Never caught that due to usually just using --parallel. +1 to -j.

[cdeil]

I've attached a commit that changes to -j in astropy.
The same change should be made in astropy-helpers.

@embray Can you cherry-pick that commit into astropy-helpers directly or should I make an extra PR there?

[embray]

Either way.

[cdeil]

Then I'll admit I'm lazy and not make a separate PR for astropy-helpers.

This change is not tested on travis-ci anyways, so is this ready to merge?

[astrofrog]

Actually should we even change it in astropy at all? Or should all changes be made only in astropy-helpers?

[embray]

Both, for now.

[cdeil]

I think the travis-ci fail is unrelated to this PR?
https://travis-ci.org/astropy/astropy/jobs/26160813
Can someone please restart that build?

[cdeil]

@embray Can you please label this "bug" and add the 0.4.0 or 0.4.1 milestone?

[embray]

This is pretty minor, but if you could just add a changelog entry (perhaps under "misc"?) then we can merge this.

[cdeil]

I added a changelog entry.

[astrofrog]

@cdeil - it looks like this needs rebasing

[cdeil]

rebased

[cdeil]

@astrofrog This fix needs to be merged into astropy-helpers, too.

[astrofrog]

@cdeil - could you open a pull request there? (also, I realized too late that this was in astropy-helpers - and should only really be fixed there)

[cdeil]

Here's the PR for astropy-helpers: astropy/astropy-helpers#29