Minutes_2022_07_26

Numba Meeting: 2022-07-26

Attendees: Siu Kwan Lam, Andre Masella, Graham Markall, LI Da, Shannon Quinn, stuart, brandon willard, Guilherme Leobas, Val, Nicholas Riasanovsky, Luk NOTE: All communication is subject to the Numba Code of Conduct.

Please refer to this calendar for the next meeting date.

0. Discussion

• Released Numba 0.56.0 and llvmlite 0.39.0
• Numba 0.56.x for numpy 1.23 as soon as conda packages are available
• (suggest topic) - (name)
• Numba Governance (current draft) - Shannon Quinn
  • two years old o_O
  • conda is adopting a governance model (https://github.com/conda-incubator/governance)
  • initial goal to spell out how decision is made
  • Update groups on Github, cleanout the permissions
  • Answer the question: "who is in charge here?"
  • Process for people becoming a "maintainer"?
• Bytes issues due to ArrayCompatable - Nick Riasanovsky
  • https://github.com/numba/numba/issues/8286
  • https://github.com/numba/numba/issues/8285
  • https://github.com/numba/numba/issues/7921
  • https://github.com/numba/numba/issues/7174

1. New Issues

• #8257 - Segfault in np.split with negative axis
• #8260 - Numba 0.56.0 Checklist
• #8262 - msan support
• #8263 - NumPy 1.23 support
• #8268 - Use of uninitialized values for empty loops
• #8270 - Support np.any for scalar input
• #8272 - 'CUDAUFuncDispatcher' object has no attribute 'name'
• #8273 - Erratic warnings (NumbaPerformanceWarning: Grid size) when using @numba.vectorize(target = 'cuda')
• #8278 - Installation of numba==0.56.0rc1 not possible: ERROR: No matching distribution found for llvmlite<0.40,>=0.39.0dev0
• #8282 - Numba Wrapper Address Protocol (WAP) is broken
• #8284 - Support set_parallel_chunksize in the OpenMP threading layer
• #8285 - Slice of byte string has an incorrect result with boxing

llvmlite:

• #861 - llvmlite 0.39.0 Checklist
• #862 - MemorySanitizer (msan) support

Closed Issues

• #8261 - how to call guvectorize cuda
• #8280 - set_parallel_chunksize in 0.56 rc1 has no effect for the omp threading layer
• #8281 - Typo in Python demonstration on the web

2. New PRs

• #8256 - Migrate linear algebra functions from glue_lowering
• #8258 - refactor np.where to use overload
• #8259 - Add np.broadcast_to(scalar_array, ())
• #8264 - remove mk_unique_var from parfor_lowering_utils.py
• #8265 - Remove mk_unique_var from array_analysis.py
• #8266 - Remove mk_unique_var in untyped_passes.py
• #8267 - Fix segfault for invalid axes in np.split
• #8271 - [WIP] Implement some CUDA intrinsics with @overload, @overload_attribute, and @intrinsic
• #8276 - Move Azure to use macos-11.
• #8277 - Added error catching in Interval doctest
• #8283 - Clean up / remove support for old NumPy versions

llvmlite:

• #865 - Move Azure to use macos-11.

Closed PRs

• merged - #8269 - Cherry pick #8255
• merged - #8274 - Update version support table doc for 0.56.
• merged - #8275 - Update CHANGE_LOG for 0.56.0 final
• merged - #8279 - Cherry pick for 0.56 release

llvmlite:

• merged - #863 - Update changelog for 0.39.0 release
• merged - #864 - Update release date for 0.39.0 release.
• merged - #866 - Cherry pick for 0.39 release
• merged - #867 - Update CHANGE_LOG 0.39.0 final.

3. Next Release: Version 0.57.0/0.40.0, RC Jan 2023