Remove python 2 specific code

[stephenworsley]

As requested in #3457, Python 2 specific code has been removed.

[bjlittle]

@stephenworsley It seems that the whole point of the data_op property was convenience i.e., to do the right thing across Python2 and Python wrt div and truediv.

Now that we're Python3, I'm just questioning the need for the existence of the data_op property.

Can't we just use operator.truediv or / directly inline for the tests? i.e.,

    def test_unmasked_div_zero(self):
        # Ensure cube behaviour matches numpy operator behaviour for the
        # handling of arrays containing 0.
        dat_a = np.array([0., 0., 0., 0.])
        dat_b = np.array([2., 2., 2., 2.])

        cube_a = Cube(dat_a)
        cube_b = Cube(dat_b)

        com = dat_b / dat_a
        res = self.cube_func(cube_b, cube_a).data

        self.assertArrayEqual(com, res)

Also drop the import operator as well 😉

[stephenworsley]

So it looks like what's going on here is that these classes are actually inheriting a bunch of tests from https://github.com/SciTools/iris/blob/master/lib/iris/tests/unit/analysis/maths/__init__.py

data_op is an abstract property in the parent class where it is used in other tests

iris/lib/iris/tests/unit/analysis/maths/__init__.py

Lines 28 to 31 in 66423f8

	@abstractproperty
	def data_op(self):
	# Define an operator to be called, I.E. 'operator.xx'.
	pass

It seems like the current usage makes sense in this context.

[bjlittle]

@stephenworsley Cool.

I guess what I'm pushing is that the whole data_op property as a concept is now defunct and should be removed. For me it's simply technical debt that should be purged, right?

[stephenworsley]

I don't think it is defunct, it allows code reuse for the tests in iris.tests.unit.maths.__init__ over multiple different operations, see also addition for example https://github.com/SciTools/iris/blob/master/lib/iris/tests/unit/analysis/maths/test_add.py

[bjlittle]

Okay, so it's generic and not specific, thanks for clarifying 👍

[bjlittle]

As the above comment.

[bjlittle]

@stephenworsley So is this good to go? Do you have more changes in-bound?

[stephenworsley]

@bjlittle Thats all I can find for the moment, If we find any others they can go in another pull request.

[bjlittle]

@stephenworsley Great detective work, thanks 😄