Remove dependency from nose

[utkbansal]

Fixes #

Changes made in this Pull Request:

PR Checklist

• Tests?
• Docs?
• CHANGELOG updated?
• Issue raised/referenced?

[kain88-de]

Can you remove nose as well from the Travis config. Then we can see if this really is the last thing that uses nose.

[utkbansal]

@kain88-de This is not the last thing that uses nose. I'm currently investigating what is still dependent on nose. Will update as I know more.

[utkbansal]

@kain88-de https://github.com/MDAnalysis/mdanalysis/pull/1574/files#diff-416947382aa4e650975916732248fef5R519 is failing. I tried using the debugger and the value of dihedrals[0] comes out to be 0.0 (float64).

[kain88-de]

The diff you are citing is to big.

[utkbansal]

@kain88-de testsuite/MDAnalysisTests/utils/test_distances.py line 519. Method is test_dihedrals in TestCythonFunctions class.

[richardjgowers]

Not sure why that would have started working, but it looks like the tests pass now? (Ie it raises the right Error again). This just needs a rebase now

[mnmelo]

This is weird indeed. @nestorwendt's recent PR #1566 went over the behavior of those lines (the one in question seems to be the check whether an undefined dihedral returns a nan).
There should be little numerical instability in this particular check, since it's the case where all four dihedral atoms' positions are [0., 0., 0.].

[utkbansal]

@kain88-de @richardjgowers This works on Travis, not on my system. Have tried reinstalling the package and testusite, but the tests still fail.

[kain88-de]

Looks good besides of the issue with marking a test when the import doesn't work. Please change all of those tests to use pytest.importorskip

[kain88-de]

correct here is pytest.importorskip. Note you need to use it inside of the function.

[kain88-de]

@utkbansal I don't know why the test don't work on your local laptop. It's hard to tell without being in front of the computer.

[utkbansal]

@kain88-de

    def chk_same_position(x_id, y_id, hasval='nan'):
        """Handling nan/inf: check that x and y have the nan/inf at the same
            locations."""
        try:
            assert_array_equal(x_id, y_id)
        except AssertionError:
            msg = build_err_msg([x, y],
                                err_msg + '\nx and y %s location mismatch:'
                                % (hasval), verbose=verbose, header=header,
                                names=('x', 'y'), precision=precision)
>           raise AssertionError(msg)
E           AssertionError: 
E           Arrays are not almost equal to 5 decimals
E           Cython dihedrals didn't match numpy calculations
E           x and y nan location mismatch:
E            x: array([ 0.     ,  0.     ,  3.14159,  0.50714])
E            y: array([     nan,      nan,  3.14159, -0.50714], dtype=float32)

[kain88-de]

Are your cython generated files up do date? We recently changed them so maybe you need to compile the cython modules against the current package version you are testing.

[utkbansal]

@kain88-de Can we merge this soon? Would save me from a lot of merge conflicts.