numba: Tests using float hashing fail on win32+Py3.10 for 0.55rc1

Details are sparse at present but essentially the hashing algs for floats were updated to deal with a change in NaN handling for Python 3.10. Tests that use this path on windows 32 are failing with a lowering error manifesting from llvmlite, message is Type of #3 arg mismatch: i32 != i64 which usually indicates a intp (platform int size) related problem.

About this issue

Original URL
State: closed
Created 2 years ago
Comments: 17 (14 by maintainers)

Commits related to this issue

Fix #7713: Ensure _prng_random_hash return has correct bitwidth When the hash width is 32 bits, get_next_int32 needs to be used because that returns a 32-bit integer, as opposed to get_next_int, whic... — committed to gmarkall/numba by gmarkall 2 years ago
Merge pull request #7748 from gmarkall/issue-7713 Fix #7713: Ensure _prng_random_hash return has correct bitwidth — committed to numba/numba by sklam 2 years ago
Merge pull request #7748 from gmarkall/issue-7713 Fix #7713: Ensure _prng_random_hash return has correct bitwidth — committed to esc/numba by sklam 2 years ago

Most upvoted comments

(unfortunately it is a pain that we - again - can’t use the current numpy version (1.22.1) with numba - is there as specific reason for this or the requirement has been put into the numba source out of an overabundance of caution?)

#7754 will track progress on NumPy 1.22 support.

gmarkall on Jan 18, 2022

numba now installs clean from pip in my source Python source install on Fedora 35/python 3.10.1.
Much appreciated!

also managed to install on M1 (MacOS 12.1)

brew install llvm@11
pip3 install -U --force-reinstall numpy==1.21.5
LLVM_CONFIG=/opt/homebrew/Cellar/llvm@11/11.1.0_3/bin/llvm-config pip3 install numba

2sn on Jan 17, 2022