Repository navigation
memoryview on array with datetime/timedelta-dtype triggers SystemError on Python 3.5 #7270
Description
On Python 3.5, constructing a memoryview for an ndarray containing a datetime/timedelta-dtype leads to a SystemError: <class 'memoryview'> returned a result with an error set:
In [50]: memoryview(numpy.zeros(10, dtype=[("A", "m8[s]")]))
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
ValueError: cannot include dtype 'm' in a buffer
During handling of the above exception, another exception occurred:
SystemError Traceback (most recent call last)
<ipython-input-50-5d5ac6c085fa> in <module>()
----> 1 memoryview(numpy.zeros(10, dtype=[("A", "m8[s]")]))
SystemError: <class 'memoryview'> returned a result with an error set
Prior to Python 3.4, results for the same were unpredictable and bizarre, see this Stack Exchange question.
I am aware that datetime64 and timedelta64 do not support the buffer protocol (see #4983), but the another exception occurred should not happen.
This appears to be related to Python issue 23571, although I can't say I quite understand what's going on.
Activity
gerritholl commentedon Feb 17, 2016
Or should this still be considered an upstream bug?
augurar commentedon Aug 10, 2016
Yes, this is an upstream bug. It seems that
memoryview()is setting an error but returning a non-null result. Handling of this case changed in Python 3.5, so theSystemErroris expected behavior.eric-wieser commentedon Mar 1, 2019
I can't reproduce this any more in 1.15.4, despite the fact that nothing seems to have changed upstream.