Minutes_2021_08_10

Numba Meeting: 2021-08-10

Attendees: Siu Kwan Lam, Todd Anderson, Luk, stuart, Guilherme, Graham

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

PSA:
- RC3 out. Hopefully last of this series.
- RC testers group: https://numba.discourse.group/g/RC_Testers
  - no obligation, just for notification purpose
Calling cython things from Numba (general issue about calling functions that take np arrays)
- (long digression into profiling)

#7293 - The |= operator (set.__ior__) causes errors when used instead of set.update
#7292 - FR: initializing variable-shaped arrays; lists/arrays as arguments to array constructors and other functions
#7291 - unable to use List.empty_list of type jitclass or deferred_type in jitclass.init
*** #7290 - global flag to default all numpy / numba calculations to be on float32
#7287 - the error traceback infomation of import numba.
#7286 - NameError in stencil when cval is np.nan or np.inf
#7284 - CUDA: TestCudaArrayInterface.TestKernelArg has invalid accesses
#7282 - BUG/DOC: Missing definition in docs
#7281 - readthedocs builds of stable are failing
#7279 - Errors of Ingest in scanpy which is caused by Numba.
#7278 - Make non-NumbaError exceptions during compilations hard errors
#7277 - Refactor: consolidate the TLS stacks used to track target configuration
#7276 - Refactor: rename one of the TargetConfig classes
#7274 - Numba 0.54.0rc3 Checklist
#7268 - Running jit_module on an existing module? (And document calling jit(...)(func) manually)

#7294 - Continuation of PR #7280, fixing lifetime of TBB task_scheduler_handle
#7289 - Implement np.tensordot()
#7288 - Handle cval as a np attr in stencil generation.
#7285 - CUDA: Fix OOB in test_kernel_arg
#7283 - Added NPM to Glossary and linked to mentions
#7280 - Fix lifetime of task_scheduler_handle
#7270 - Add support for numpy.moveaxis/normalize_axis_index and normalize_axis_list in nopython mode
#7266 - CUDA: Skip multi-GPU copy test with peer access disabled
#7265 - Fix linspace to use np.divide and clamp to stop.

Request for 0.55
- broadcast_to https://github.com/numba/numba/pull/7119
  - driven by need for vectorize support in aesara