filprofiler
filprofiler copied to clipboard
Conda packaging failing to build with failing tests
As far as I can tell, in malloc.py:
- malloc() works
- second realloc() that increases claims to work but isn't tracked.
- final shrinking realloc() does work.
It's very confusing and might be a real bug but only happens in this one place...
Maybe im
is buggy?
For now I'm just skipping tests...
Installing numexpr
from PyPI on 3.9 (but not numexpr
from Conda-Forge, and not older Pythons) made tests pass?!