Minutes_2021_03_23

Numba Meeting: 2021-03-23

Attendees: Todd, Ehsan, Farah, Graham, Guilherme, Hadia, Jim P., Ryan C., Siu, Stu, Val.

NOTE: All communication is subject to the Numba Code of Conduct.

0.53.1 status
- https://github.com/numba/numba/milestone/55
- Plan to release (tag) on Thurs
Mentorship program
- 4(5) participants
- starting soon
- not sure about the exact protocol, we'll have to figure everything as we go
#6846 keyword-only & **kwargs
- keyword-only:
  - bound back to py2
  - existing code doesn't
  - needed fix: overhaul class Signature and friends
  - It looks like it was intended, but not implemented correctly. This is the code that I hit when trying to use keyword-only arguments naively with the overload decorator: code snippet that includes "if kwonly...".
  - Jim has a work-around: https://github.com/numba/numba/issues/6846, so nothing is needed at this time.
- **kwargs
  - maps to LiteralStrKeyDict
  - difficulty: hard (ETA: ~weeks)
JimP announces new project https://github.com/scikit-hep/vector
JimP: another problem is for vector to support multiple backend; numba's dispatcher is getting in the way
- possible solution: register_jitable

**** #6846 - Implement keyword-only or **kwargs handling for numba.extending.overload
#6845 - cache=True with same wrapped fn but different targetoptions [request]
#6839 - BUG: segfault in pandas rolling test
#6838 - Error on keyword arguments to parallel functions
#6835 - Structured scalar values underflow on access
#6832 - llvm_pass_timings fails to parse some raw data on 32-bit

#6842 - Error with Moran_Local()
#6831 - test_no_accidental_warnings fails with packaging >= 20.5
#6827 - CUDA Mapped Array Has No Object "stream" when attempting to access __cuda_array_interface__

#6850 - Fix up more internal warnings
#6849 - Add __hash__ for IntEnum
#6848 - Ignore .ycm_extra_conf.py
#6847 - [WIP][QUESTIONS] An attempt to extend Enum support
#6844 - CUDA: Remove NUMBAPRO env var warnings, envvars.py + other small tidy-ups
#6843 - cross-referenced Array type in docs
#6841 - CUDA: Fix source location on kernel entry and enable breakpoints to be set on kernels by mangled name
#6840 - CUDA: Remove items deprecated in 0.53 + simulator test fixes
#6834 - Remove illegal "debug" kw argument
#6833 - Raise TypeError exception if numpy array is cast to scalar
#6826 - Fix regression on gufunc serialization

#6837 - Ignore warnings from packaging module when testing import behaviour.
#6836 - CUDA: Documentation updates
#6830 - CUDA: Use relaxed strides checking to compute contiguity
#6829 - Fix accuracy of np.arange and np.linspace
#6828 - Fix regression in CUDA: Set stream in mapped and managed array device_setup