Minutes_2022_05_03
Siu Kwan Lam edited this page May 3, 2022
·
1 revision
Attendees: Siu Kwan Lam, stuart, Jim Pivarski, Shannon Quinn, Benjamin Graham, Nick Riasanovsky, Andre Masella, Luk, Guilherme Leobas, brandon willard
NOTE: All communication is subject to the Numba Code of Conduct.
Please refer to this calendar for the next meeting date.
- "How to become a maintainer?"
- What is maintenance?
- understand risk and managing risk to keep the project stable
- keep the project up-to-date: numpy, cpython, openmp, tbb, etc...
- triage and fixing bugs
- updating, managing IRC channel, forums, CI infrastructure
- managing resources for releases
- reviewing, approving PRs
- become code owner to added code and approved PRs
- maintaining, updating documentation
- maintainer concerns:
- bandwidth to respond to incoming contributions
- current maintainers are doing all the roles: architect, release managers, ...
- Questions:
- Can someone become a maintainer spending 10hrs on one of the maintenance task?
- create specific roles as the community grow
- How can contributors help maintainers?
- Stuart: good areas are:
- help on issue triage and answer user question
- Stuart: on code reviewing:
- taking code ownership on parts of numba and review PRs in those parts.
- Stuart: good areas are:
- What is preventing people in contributing?
- Knowing what area of contribution is the most helpful
- things that require coordination
- standardized review procedure
- Can someone become a maintainer spending 10hrs on one of the maintenance task?
- What is maintenance?
(Out of time. continue the following topics next week)
- Support
*args, **kwsin overloads
- #8007 - Wrong results when using jit(parallel=True)
- ** #8008 - meta-issue: Numba/NumPy interation+collaboration+discussion!
-
#8009 - Rewrite
numba.{carray,farray}support using@overload - #8013 - cc.distutils_extension() does not work in project with nested structure
- #8015 - "resize" attribute of numpy array
- #8018 - numba vectorize does not short circuit as expected
- #8011 - Use with construct to compile one time use blocks of code
- #8012 - Parallel fails when using .real and .imag attributes with numpy functions