Unit tests fails with numpy 1.13.1 #577

onovy · 2017-10-31T20:47:43Z

Refs: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=876921

Thanks for fixing.

S.S................................................./usr/lib/python2.7/unittest/case.py:340: RuntimeWarning: TestResult has no addExpectedFailure method, reporting as passes
  RuntimeWarning)
.......S...S............................SSSSSSSSSSS.................S...................................................................................................................................................E.E...E.................................................................................................................................................................................................................................................................................................../build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_quantity.py:287: RuntimeWarning: to_compact applied to non numerical types has an undefined behavior.
  self.assertQuantityAlmostIdentical(q.to_compact(unit=unit),
.......................................F..F..........................................................................................................................................................................
======================================================================
ERROR: test_inplace_exponentiation (pint.testsuite.test_quantity.TestOffsetUnitMath) [with input = ((10, u'degC'), (2, u'')); expected_output = [u'error', (80173.92249999999, u'kelvin**2')]]
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/parameterized.py", line 116, in new_method
    return method(self, *param_values)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/parameterized.py", line 137, in newfunc
    return func(*arg, **kwargs)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_quantity.py", line 1165, in test_inplace_exponentiation
    self.assertEqual(op.ipow(in1_cp, in2).units, expected.units)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/quantity.py", line 961, in __ipow__
    self._magnitude **= _to_magnitude(other, self.force_ndarray)
TypeError: unsupported operand type(s) for ** or pow(): 'numpy.ndarray' and 'Quantity'

======================================================================
ERROR: test_inplace_exponentiation (pint.testsuite.test_quantity.TestOffsetUnitMath) [with input = ((10, u'kelvin'), (2, u'')); expected_output = [(100.0, u'kelvin**2'), (100.0, u'kelvin**2')]]
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/parameterized.py", line 116, in new_method
    return method(self, *param_values)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/parameterized.py", line 137, in newfunc
    return func(*arg, **kwargs)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_quantity.py", line 1165, in test_inplace_exponentiation
    self.assertEqual(op.ipow(in1_cp, in2).units, expected.units)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/quantity.py", line 961, in __ipow__
    self._magnitude **= _to_magnitude(other, self.force_ndarray)
TypeError: unsupported operand type(s) for ** or pow(): 'numpy.ndarray' and 'Quantity'

======================================================================
ERROR: test_inplace_exponentiation (pint.testsuite.test_quantity.TestOffsetUnitMath) [with input = ((10, u'degC'), (500.0, u'millikelvin/kelvin')); expected_output = [u'error', (16.827061537891872, u'kelvin**0.5')]]
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/parameterized.py", line 116, in new_method
    return method(self, *param_values)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/parameterized.py", line 137, in newfunc
    return func(*arg, **kwargs)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_quantity.py", line 1165, in test_inplace_exponentiation
    self.assertEqual(op.ipow(in1_cp, in2).units, expected.units)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/quantity.py", line 961, in __ipow__
    self._magnitude **= _to_magnitude(other, self.force_ndarray)
TypeError: unsupported operand type(s) for ** or pow(): 'numpy.ndarray' and 'Quantity'

======================================================================
FAIL: test_isfinite (pint.testsuite.test_umath.TestFloatingUfuncs)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_umath.py", line 617, in test_isfinite
    (self.q1, self.qm, self.qless))
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_umath.py", line 101, in _testn
    self._test1(func, ok_with, raise_with, output_units=None, results=results)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_umath.py", line 85, in _test1
    self.assertQuantityAlmostEqual(qm, res, rtol=rtol, msg=err_msg)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/__init__.py", line 117, in assertQuantityAlmostEqual
    np.testing.assert_allclose(m1, m2, rtol=rtol, atol=atol, err_msg=msg)
  File "/usr/lib/python2.7/dist-packages/numpy/testing/utils.py", line 1395, in assert_allclose
    verbose=verbose, header=header, equal_nan=equal_nan)
  File "/usr/lib/python2.7/dist-packages/numpy/testing/utils.py", line 778, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Not equal to tolerance rtol=1e-06, atol=0
At isreal with [ 1.  2.  3.  4.] joule
(mismatch 100.0%)
 x: array(False, dtype=bool)
 y: array([ True,  True,  True,  True], dtype=bool)

======================================================================
FAIL: test_isreal (pint.testsuite.test_umath.TestFloatingUfuncs)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_umath.py", line 609, in test_isreal
    (self.q1, self.qm, self.qless))
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_umath.py", line 101, in _testn
    self._test1(func, ok_with, raise_with, output_units=None, results=results)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/test_umath.py", line 85, in _test1
    self.assertQuantityAlmostEqual(qm, res, rtol=rtol, msg=err_msg)
  File "/build/1st/python-pint-0.8.1/.pybuild/pythonX.Y_2.7/build/pint/testsuite/__init__.py", line 117, in assertQuantityAlmostEqual
    np.testing.assert_allclose(m1, m2, rtol=rtol, atol=atol, err_msg=msg)
  File "/usr/lib/python2.7/dist-packages/numpy/testing/utils.py", line 1395, in assert_allclose
    verbose=verbose, header=header, equal_nan=equal_nan)
  File "/usr/lib/python2.7/dist-packages/numpy/testing/utils.py", line 778, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Not equal to tolerance rtol=1e-06, atol=0
At isreal with [ 1.  2.  3.  4.] joule
(mismatch 100.0%)
 x: array(False, dtype=bool)
 y: array([ True,  True,  True,  True], dtype=bool)

----------------------------------------------------------------------
Ran 779 tests in 40.510s

The text was updated successfully, but these errors were encountered:

cpascual · 2018-01-24T10:58:26Z

Is anyone looking into this?
I am concerned because if this is not fixed, pint may be dropped from debian (and our own package, taurus, would be dropped as well because it depends on pint)

jetheurer · 2018-02-14T14:24:39Z

Hey @cpascual, I've noticed this error appears in both numpy==1.13.0 and numpy==1.13.1 on Python 2.7 and I believe it is due to the lack of a __array_ufunc__ method on the _Quantity class. I'm going to look into making a fix along these lines.

@hgrecco Did you have any thoughts for this ticket?

dalito · 2018-02-14T15:21:56Z

Automated testing on Travis seems to be limited to numpy <= 1.11. So these failures are maybe not too surprising. Testing should be extended to include numpy 1.13.x and 1.14.

Numpy 1.13 included improvements to help other projects with subclassing ndarray. Maybe the numpy support in pint can be improved now when using newer numpy versions.

cpascual · 2018-02-19T09:54:04Z

Hi @jetheurer , any progress?
The deadline for debian packages to be included in Ubuntu LTS 18.04 is 1st of March. I am not sure if https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=876921 precludes pint (and the packages depending on it) to be part of the next Ubuntu LTS...

jetheurer · 2018-02-20T09:23:42Z

Hey @cpascual, I don't have anything yet, but I'll try to get something up by Saturday. Thanks for the 1st March info, that deadline is pretty close!

cpascual · 2018-02-20T10:37:50Z

Hi @jetheurer , do not worry: a debian dev told me that https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=876921 does not mean that pint will be dropped from Ubuntu

bonderado · 2018-03-26T04:36:15Z

For reference: pint just dropped off Debian Buster (https://tracker.debian.org/pkg/python-pint)

df7cb · 2018-04-05T07:06:30Z

All packages depending on pint have been removed from Debian Buster/testing last week. A fix would be very welcome.

cpascual · 2018-04-05T07:12:54Z

I am going to give this a try myself.
@jetheurer did you get something done?

jetheurer · 2018-04-05T13:28:02Z

@cpascual, go for it. I dont have anything to show at the moment.

cmft · 2018-04-05T14:49:17Z

Hi @jetheurer, @cpascual ,
I have a look on that, and the problems seem to be in:

assertQuantityAlmostEqual method (in testsuite.init.py)
The problem is in np.testing.assert_allclose(m1, m2, rtol=rtol, atol=atol, err_msg=msg)
The np.testing.assert_allclose seems to be changed. At naked eye the values seems to be equals.
self._magnitude **= _to_magnitude(other, self.force_ndarray) ( in pint/quantity.py)
If you replace this call for this code the problem is solved.

self._magnitude = np.power(self._magnitude,                                                                                                                   
                                             _to_magnitude(other,                                                                                                                       
                                             self.force_ndarray))

jetheurer · 2018-04-08T06:13:08Z

@cmft That looks like a pretty quick fix. For some reason I thought the issue was more complicated than that - but maybe it wasn't too bad after all!

coreycb · 2018-04-11T19:20:52Z

@cpascual hi, were you planning to submit a patch for this?

cpascual · 2018-04-12T07:51:47Z

Hi @coreycb , yes, I am working on the patch. The problem is that I am not sure on how to approach it. For the moment I am focusing on the tests that give Error (the ones that test Quantity.ipow). I am also in touch with @cmft (we are co-workers) and we are trying to understand the ultimate cause of the failures with new numpy versions.

cpascual · 2018-04-12T10:55:02Z

update: I managed to fix the ipow errors and the test_isfinite fail. Now I am working on the test_isreal. Once I get it I'll make the PR

cpascual · 2018-04-13T18:11:18Z

Update: I finally created the PR. Anyone interested, please have a look at it in #630

@hgrecco

630: Fix unit tests errors with numpy >=1.13 (Fix 577) r=cpascual a=cpascual This PR fixes the 4 test errors/failiures reported in #577 and adds new travis builds with numpy 1.14 3 of the 4 problems are relatively trivial to solve. The remaining one (`test_isreal`) is not so trivial because it involves allowing to compare a *non-dimensionless* quantity against zero. The reason is that `numpy.isreal(x)` essentially does `x.imag !=0` if the `imag` method is implemented in x... and since Quantity reimplements .imag to return units (see #171), the check always returns False if x is a non-dimensionless quantity. As mentioned, this can be solved by allowing comparisons against zero. Actually, this has already been requested in #194, but it was not implemented. In this PR, I implemented the comparison-to-zero feature and provide several new unit tests for it. AFAIK this implementation takes into account the concerns expressed by @hgrecco in #194. But if the feature is not wanted, we can just revert the last 3 commits of this PR and then `test_isreal` will left with an expected failure flag. The rules that I implemented for the comparison-with-zero are: ``` a) comparing against (non-quantity) zero is allowed for any quantity b) comparing against zero-magnitude quantities of incompatible dimensionality raises a Dimensionality error, except for the case of equality, for which it always returns False Notes: 1.- Numpy arrays of zeros are also supported and the comparison rules apply elementwise 2.- In the case of non-multiplicative units, the rules above apply after converting to base units if the autoconvert offset flag is set. Otherwise they raise an OffsetUnitCalculusError. ``` Fixes #577 Co-authored-by: Carlos Pascual <[email protected]>

df7cb · 2018-04-16T13:36:56Z

Thanks for the fix! Could we have a new release with it, or should we rather cherry-pick it for the Debian package?

cpascual · 2018-04-16T14:13:50Z

Could we have a new release with it,

About this, I cannot say... maybe @hgrecco can say something.

or should we rather cherry-pick it for the Debian package?

I just tested cherry-picking the following 4 commits on top of the 0.8.1 tag : the tests work fine.

hgrecco · 2018-04-16T14:23:32Z

There are still several things pending until 0.9 but I could release 0.8.2 if everyone agrees.

cpascual · 2018-04-16T14:33:16Z

but I could release 0.8.2 if everyone agrees.

That would be nice, and it would simplify packaging in debian...

onovy · 2018-04-16T14:36:58Z

hi, fyi: I already uploaded cherry-picked version of this fix to Debian this morning.

hgrecco · 2018-04-16T16:08:01Z

I have just updated CHANGES. Please notice that we will be releasing a 0.8.2 in which we have dropped support for Python 2.6, 3.0, 3.1 and 3.2

dalito mentioned this issue Feb 16, 2018

Numpy array assigning temperature in different units gives DimensionalityError #618

Open

cpascual mentioned this issue Apr 5, 2018

Problem parsing magnitude that starts with "0" #614

Open

cpascual mentioned this issue Apr 13, 2018

Fix unit tests errors with numpy >=1.13 (Fix 577) #630

Merged

bors bot closed this as completed in #630 Apr 13, 2018

jondoesntgit mentioned this issue Jul 27, 2018

Use 'µm' as abbreviation of micrometer? #666

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Unit tests fails with numpy 1.13.1 #577

Unit tests fails with numpy 1.13.1 #577

onovy commented Oct 31, 2017

cpascual commented Jan 24, 2018 •

edited

Loading

jetheurer commented Feb 14, 2018

dalito commented Feb 14, 2018 •

edited

Loading

cpascual commented Feb 19, 2018

jetheurer commented Feb 20, 2018

cpascual commented Feb 20, 2018

bonderado commented Mar 26, 2018

df7cb commented Apr 5, 2018

cpascual commented Apr 5, 2018

jetheurer commented Apr 5, 2018

cmft commented Apr 5, 2018

jetheurer commented Apr 8, 2018

coreycb commented Apr 11, 2018

cpascual commented Apr 12, 2018

cpascual commented Apr 12, 2018

cpascual commented Apr 13, 2018

df7cb commented Apr 16, 2018

cpascual commented Apr 16, 2018 •

edited

Loading

hgrecco commented Apr 16, 2018

cpascual commented Apr 16, 2018

onovy commented Apr 16, 2018

hgrecco commented Apr 16, 2018

Unit tests fails with numpy 1.13.1 #577

Unit tests fails with numpy 1.13.1 #577

Comments

onovy commented Oct 31, 2017

cpascual commented Jan 24, 2018 • edited Loading

jetheurer commented Feb 14, 2018

dalito commented Feb 14, 2018 • edited Loading

cpascual commented Feb 19, 2018

jetheurer commented Feb 20, 2018

cpascual commented Feb 20, 2018

bonderado commented Mar 26, 2018

df7cb commented Apr 5, 2018

cpascual commented Apr 5, 2018

jetheurer commented Apr 5, 2018

cmft commented Apr 5, 2018

jetheurer commented Apr 8, 2018

coreycb commented Apr 11, 2018

cpascual commented Apr 12, 2018

cpascual commented Apr 12, 2018

cpascual commented Apr 13, 2018

df7cb commented Apr 16, 2018

cpascual commented Apr 16, 2018 • edited Loading

hgrecco commented Apr 16, 2018

cpascual commented Apr 16, 2018

onovy commented Apr 16, 2018

hgrecco commented Apr 16, 2018

cpascual commented Jan 24, 2018 •

edited

Loading

dalito commented Feb 14, 2018 •

edited

Loading

cpascual commented Apr 16, 2018 •

edited

Loading