Repository navigation
Python 2.7 heterogeneous list unbox problem #4087
Description
Reporting a bug
- I am using the
latest releasedmaster version of Numba (most recent is visible in
the change log (https://github.com/numba/numba/blob/master/CHANGE_LOG). - I have included below a minimal working reproducer (if you are unsure how
to write one see http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports).
Python 2.7 has an issue with this:
from numba import njit
import numpy as np
@njit
def foo(arg1, arg2):
return 10
a = [np.array([ True]), np.array([False, True, False])]
b = [np.array([1]), np.array([[ 0, 1, 2]])]
try:
print(foo(1, b)) # ok, TypeError
except TypeError:
pass
print(foo(a, b)) # brokenwhich gives:
test_pr_4077.py:15: RuntimeWarning: tp_compare didn't return -1 or -2 for exception
print(foo(a, b)) # broken
Traceback (most recent call last):
File "test_pr_4077.py", line 15, in <module>
print(foo(a, b)) # broken
SystemError: NULL result without error in PyObject_Call
seems like it's the unbox of the second list that's the problem where it types it as a types.List(types.Array(int64, 1,'C')) but the second element is a 2d array. Why it only impacts the second use case and Python 2.7 (3.x is fine) needs more investigation.
Activity
stuartarchibald commentedon Feb 6, 2020
Python 2.7 reached EOL on January 1st, 2020 and Numba 0.47 was the last release to support Python 2.7. Closing this issue as it is Python 2.7 specific, if this problem is reproducible on Python 3 with Numba >= 0.48 then please reopen it. Many thanks!