Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

test failure on Fedora rawhide with python3.5 #278

Closed
keszybz opened this issue Nov 24, 2015 · 9 comments

Comments

Projects
None yet
4 participants
@keszybz
Copy link

commented Nov 24, 2015

This appears to be different than other failures with python3.5 that were already fixed...

http://koji.fedoraproject.org/koji/taskinfo?taskID=11968168
with python3-numpy-1.10.2-0.1rc1.fc24 and python3-3.5.0-5.fc24.

======================================================================
FAIL: joblib.test.test_hashing.test_hashes_stay_the_same_with_numpy_objects('94fb1c385e1e955b74ead88ffaf6dcc8', '6028954a4c71750d74e66414bbab227b')
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
AssertionError: '94fb1c385e1e955b74ead88ffaf6dcc8' != '6028954a4c71750d74e66414bbab227b'
======================================================================
FAIL: joblib.test.test_hashing.test_hashes_stay_the_same_with_numpy_objects('aaafcb9154f904ad2d387312af7ca907', 'd32171e5bcf0e60034a7923bfea59e08')
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
AssertionError: 'aaafcb9154f904ad2d387312af7ca907' != 'd32171e5bcf0e60034a7923bfea59e08'
======================================================================
FAIL: joblib.test.test_hashing.test_hashes_stay_the_same_with_numpy_objects('a271389ac1271a618dbd3e10b688a3c3', '7989a9bbb5b7334efeda618843ab3ecc')
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
AssertionError: 'a271389ac1271a618dbd3e10b688a3c3' != '7989a9bbb5b7334efeda618843ab3ecc'
----------------------------------------------------------------------
Ran 626 tests in 11.397s
FAILED (SKIP=1, failures=3)
@sergiopasra

This comment has been minimized.

Copy link

commented Nov 24, 2015

This is not a Python 3.5 error. It happens in Python 2.7 also but only in i386

@sergiopasra

This comment has been minimized.

Copy link

commented Nov 24, 2015

The failures happen with release joblib-0.9.3 and with master

@lesteve

This comment has been minimized.

Copy link
Contributor

commented Nov 24, 2015

I'll try to see if I can reproduce. As a sanity check, you don't happen to have the test results with numpy 1.10.1 by any chance ?

@ignatenkobrain

This comment has been minimized.

Copy link

commented Nov 24, 2015

@lesteve python3-numpy-1.10.2-0.1rc1.fc24.x86_64

@sergiopasra

This comment has been minimized.

Copy link

commented Nov 24, 2015

@lesteve I have tested this with a virtualenv installing everything with pip. numpy is 1.10.1
To reproduce you have to test in a i386 machine. x86_64 works fine

@lesteve

This comment has been minimized.

Copy link
Contributor

commented Nov 24, 2015

I can reproduce, I'll look into it, thanks for the bug report!

@lesteve

This comment has been minimized.

Copy link
Contributor

commented Nov 24, 2015

I can reproduce, I'll look into it, thanks for the bug report!

Just for completeness I am only seeing the test failures on Python 3.5 with a 32bit python. Edit: I actually see failures with both Python 2.7 and Python 3.5 with a 32bit python.

@lesteve

This comment has been minimized.

Copy link
Contributor

commented Nov 25, 2015

So the short story on this one: the arrays produced by np.random.randn are not bit-identical on 32bit and 64bit. Not sure why because both of them are float64 arrays:

import numpy as np

rng = np.random.RandomState(0)
arr = rng.randn(3)
print(repr(arr.tobytes()))

Here is the output with the difference highlighted:


32bit:
'\xa5\xcf\xe5\xf3\x8e9\xfc?R\x89\xcc\xfa,\x9c\xd9?\xad\xc9"R\xd2Q\xef?'

64bit:
'\xa5\xcf\xe5\xf3\x8e9\xfc?S\x89\xcc\xfa,\x9c\xd9?\xab\xc9"R\xd2Q\xef?'
                           ^                         ^
                           |                         |

As a side-note, this problem doesn't seem to show up on Windows since we are testing both 32bit and 64bit on AppVeyor.

@lesteve

This comment has been minimized.

Copy link
Contributor

commented Nov 25, 2015

Should be fixed in 002d6f0.

@lesteve lesteve closed this Nov 25, 2015

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.