Description
Just executing import numpy as np leaks memory:
Python version: 3.6.3
Numpy version: 0d749ad
This could also be a CPython issue, but other extensions I have tested don't leak
on 3.6.3. I have seen similar leak patterns with Valgrind if an extension did not use
the GC alloc functions or if there was a cyclic GC problem.
Example (there are more):
==8063== 1,709 bytes in 18 blocks are definitely lost in loss record 4,430 of 4,646
==8063== at 0x4C2A9A1: malloc (vg_replace_malloc.c:299)
==8063== by 0x41D06D: _PyMem_RawMalloc (obmalloc.c:73)
==8063== by 0x41D84C: PyObject_Malloc (obmalloc.c:479)
==8063== by 0x4F61AC: PyUnicode_New (unicodeobject.c:1281)
==8063== by 0x4FBC95: _PyUnicode_FromUCS1 (unicodeobject.c:2173)
==8063== by 0x4FC8A9: PyUnicode_FromKindAndData (unicodeobject.c:2244)
==8063== by 0x5C8392: r_object (marshal.c:1156)
==8063== by 0x5C863F: r_object (marshal.c:1218)
==8063== by 0x5C9136: r_object (marshal.c:1389)
==8063== by 0x5C994A: read_object (marshal.c:1487)
==8063== by 0x5CA29E: marshal_loads (marshal.c:1787)
==8063== by 0x4BC911: _PyCFunction_FastCallDict (methodobject.c:234)
==8063==