test failure with the current master

[ev-br]

$ python -c'import numpy; import scipy; print numpy.__version__; print scipy.__version__'
1.6.2
0.14.0.dev-7c99934

Ubuntu Precise 64bit, default setup.

Running scipy.test("full") triggers this:

======================================================================
ERROR: test_inf (test_basic.TestLegendreFunctions)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/br/virtualenvs/scipy-dev/local/lib/python2.7/site-packages/scipy/special/tests/test_basic.py", line 2380, in test_inf
    assert_equal(lp[1][1,1:], np.inf)
  File "/home/br/virtualenvs/scipy-dev/local/lib/python2.7/site-packages/numpy/testing/utils.py", line 256, in assert_equal
    return assert_array_equal(actual, desired, err_msg, verbose)
  File "/home/br/virtualenvs/scipy-dev/local/lib/python2.7/site-packages/numpy/testing/utils.py", line 707, in assert_array_equal
    verbose=verbose, header='Arrays are not equal')
  File "/home/br/virtualenvs/scipy-dev/local/lib/python2.7/site-packages/numpy/testing/utils.py", line 617, in assert_array_compare
    val = comparison(x[~x_id], y[~y_id])
IndexError: 0-d arrays can't be indexed

----------------------------------------------------------------------
Ran 11108 tests in 514.657s

FAILED (KNOWNFAIL=152, SKIP=555, errors=1)

[argriffing]

I also remember seeing a legendre-related test error in my scipy master.

[pv]

Was numpy assert_equal changed between 1.6.2 and 1.7.1? Apparently yes...

[rgommers]

I don't remember any changes except for improving inf/nan handling. Which may be related for this PR. If it works for 1.7 just skip the test for older numpy; all assert_* functions were broken with inf/nan input.

[pv]

Fixed in 24f1d16