| msg277811 - (view) |
Author: Mark Dickinson (mark.dickinson) *  |
Date: 2016-10-01 12:33 |
| The newly-added statistics.geometric_mean function appears to have no tests at all, with the exception of a single doctest: >>> geometric_mean([1.10, 0.95, 1.12]) 1.0538483123382172 Steve, Ned: what do you think about taking geometric_mean out of Python 3.6, leaving us time to get it into shape for 3.7? The implementation of geometric_mean is complicated, with many edge cases. The original code was committed without review in #27181, and it currently suffers from some serious bugs: see #27761, #28111, #28327 (the first of these affects only the tests for nroot; the second and third are behaviour bugs in geometric_mean). With the lack of tests, and Python 3.6 b2 looming, I think the best thing to do might be to defer the inclusion of geometric_mean to Python 3.7. I'll mark this as a release blocker until we decide what to do. |
|
|
| msg277812 - (view) |
Author: Steven D'Aprano (steven.daprano) * |
Date: 2016-10-01 12:50 |
| Unfortunately I'm in a pickle at the moment, I cannot build 3.6 on any of the machines I have available (three), and no time to upgrade them to something that will build 3.6. But I can provide some tests, if somebody is willing to review and check them in to 3.6. |
|
|
| msg277837 - (view) |
Author: Ned Deily (ned.deily) * |
Date: 2016-10-02 00:56 |
| I was hoping that the open issues could be resolved in time for b2 but that seems unlikely at this point. So I have to agree that this feature should be deferred to 3.7. Steven, can you make the necessary reverts on the 3.6 branch? |
|
|
| msg277851 - (view) |
Author: Steven D'Aprano (steven.daprano) * |
Date: 2016-10-02 05:12 |
| > The newly-added statistics.geometric_mean function appears to have no > tests at all That's weird and unfortunate. I certainly wrote tests, and I have a backup of them. I have no idea what happened. Attached is a patch that adds the tests, but obviously I haven't been able to run this until I resolve my gcc issues and can build 3.6 again. (I did run this some weeks ago, and they passed *then*, but I cannot be sure they still pass now.) If somebody would like to test this for me, it would be appreciated. |
|
|
| msg277950 - (view) |
Author: Inada Naoki (methane) * |
Date: 2016-10-03 10:57 |
| I run attached test, and saw following errors. On macOS 10.11 (XCode 8) $ hg summary parent: 104258:0d948a46c59a test_invalid_sequences seems don't have to stay in CAPITest. branch: 3.6 commit: 1 modified, 1 unknown update: (current) $ ./python.exe -m test.test_statistics -v ... ====================================================================== ERROR: test_negative_error (__main__.TestGeometricMean) (values=[-1]) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/inada-n/work/python/py36/Lib/test/test_statistics.py", line 1741, in test_negative_error self.assertRaises(exc, self.func, values) File "/Users/inada-n/work/python/py36/Lib/unittest/case.py", line 728, in assertRaises return context.handle('assertRaises', args, kwargs) File "/Users/inada-n/work/python/py36/Lib/unittest/case.py", line 177, in handle callable_obj(*args, **kwargs) File "/Users/inada-n/work/python/py36/Lib/statistics.py", line 567, in geometric_mean if isinstance(g, (numbers.Real, Decimal)): NameError: name 'g' is not defined ====================================================================== ERROR: test_single_value (__main__.TestGeometricMean) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/inada-n/work/python/py36/Lib/test/test_statistics.py", line 1587, in test_single_value self.assertEqual(self.func([x]), x) File "/Users/inada-n/work/python/py36/Lib/statistics.py", line 567, in geometric_mean if isinstance(g, (numbers.Real, Decimal)): NameError: name 'g' is not defined ====================================================================== ERROR: test_singleton_lists (__main__.TestGeometricMean) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/inada-n/work/python/py36/Lib/test/test_statistics.py", line 1762, in test_singleton_lists self.assertEqual(self.func([x]), x) File "/Users/inada-n/work/python/py36/Lib/statistics.py", line 567, in geometric_mean if isinstance(g, (numbers.Real, Decimal)): NameError: name 'g' is not defined ====================================================================== FAIL: test_multiply_data_points (__main__.TestGeometricMean) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/inada-n/work/python/py36/Lib/test/test_statistics.py", line 1796, in test_multiply_data_points self.assertApproxEqual(self.func(data), expected, rel=1e-13) File "/Users/inada-n/work/python/py36/Lib/test/test_statistics.py", line 227, in assertApproxEqual check(first, second, tol, rel, msg) File "/Users/inada-n/work/python/py36/Lib/test/test_statistics.py", line 247, in _check_approx_num raise self.failureException(msg) AssertionError: 6.27015835359916 != 695.9875772495068 values differ by more than tol=0 and rel=1e-13 -> absolute error = 689.7174188959076 -> relative error = 0.990990990990991 |
|
|
| msg277951 - (view) |
Author: Inada Naoki (methane) * |
Date: 2016-10-03 11:09 |
| Attached patch fixes first two errors. Last one error is from test. I've commented it in review tool. |
|
|
| msg277996 - (view) |
Author: Mark Dickinson (mark.dickinson) * |
Date: 2016-10-03 19:36 |
| One more geometric_mean issue (the current code can enter an infinite loop for Decimal inputs): #28351. Steve, I really think we should postpone to 3.7. I'm very happy to help out with all the necessary fixes, tests and review, but I'm not going to have time for that before 3.6b2. If we defer, that takes the pressure off a bit and lets us get a good solid implementation in for 3.7. |
|
|
| msg278026 - (view) |
Author: Steven D'Aprano (steven.daprano) * |
Date: 2016-10-04 10:08 |
| > Steve, I really think we should postpone to 3.7. [...] If these fixes have to be in by the next beta (10th Oct), I fear that you are right. I can build up to changeset 103135:8b74e5528f35, but not beyond. I will be able to rectify that, but realistically it won't be before beta2. I expect that harmonic_mean is okay to stay in though? I feel frustrated at how close to complete geometric_mean is, disappointed that I can't work on it at the moment, and embarrassed that I'm letting the team down. Sorry for the extra work I've put you through. I can do the removals -- probably tonight. |
|
|
| msg282604 - (view) |
Author: Inada Naoki (methane) * |
Date: 2016-12-07 09:19 |
| 3.6rc1 is released. What is status of this issue? |
|
|
| msg282607 - (view) |
Author: Inada Naoki (methane) * |
Date: 2016-12-07 09:45 |
| It had been rolled back. https://hg.python.org/cpython/rev/9dce0e41bedd Can we close this issue? #27181 is still open. |
|
|