Minutes_2019_10_01

Attendees: Ehsan, Stuart, Pearu, Aaron, Todd, Stan, Siu, James

0. Feature Discussion

• Release status
  • RC1
  • llvmlite branched, tagged, building v0.30.0rc1
  • numba pending
• 0.46.1
  • python 3.8
  • maybe numpy 1.17 patches
• Requests for 0.47 (last release for the year)
  • jitclass performance issues
  • llvm 9 trial
  • CTK libcudadevrt.a
  • CI needs to take 50% of current time
    • Val & Stu already looking at this
    • also checking Azure CI config to avoid wasting compute time
  • Caching:
    • transitive dependency
    • other issues: i.e. function as argument, with objmode
    • distributing cache
  • Immutable list and deprecating reflected list

1. New issues

• #4637 - numba.njit(parallel=True) LoweringError with astropy.units
  • parfors
• #4636 - np.argsort cannot work for matrix more than 1D
• #4635 - np.argsort still slowing down by jit
• #4634 - Method never returns
• #4632 - Unable to broadcast argument 1 to output array
• #4630 - Wrong answer from parfor with [:0] slice
• #4628 - CUDA device arrays are always 'A' order and they probably shouldn't be.
• #4627 - numba fallback because 'spdiags': cannot determine Numba type of <class'function'>
• **** #4625 - DUfunc ufunc methods don't work unless method already compiled
  • problem with ufunc.
  • TODO: Stan link to high-level issue
• #4624 - Support optional arguments axis, dtype and out for numpy.cumsum
• #4622 - Numba 0.46.0RC1 checklist
• #4620 - 200 times slower using jit for a combined groupby and rank
  • stu to reply
• #4618 - CUDA math.* typing incorrect
• #4616 - Error (It asked me to report it)
• #4608 - Accept overloaded functions as parameter
  • issue opened from last week discussion

Already Closed

• #4633 - Parallel execution crash
• #4631 - unsigned int type inference
• #4613 - CUDA Library Load (check prefix)

2. New Open PRs

• #4629 - Make numba.dummyarray.Array iterable
• Closed: #4626 - DO NOT MERGE. Merge of #4609 #4619
• #4623 - Fix issue 4520 due to storage model mismatch
• #4617 - Add coverage for np.argwhere
• **** #4615 - [WIP] Allow masking threads out at runtime
  • discussion on threading behavior and parallel semantic
• #4612 - Implement methods lower()/islower() for unicode based on Cpython
• #4611 - Implement method title() for unicode based on Cpython
• #4610 - Implement np.is* functions
• #4609 - Update CUDA Array Interface & Enforce Numba compliance

Already merged

• #4621 - Fix packaging issue due to missing numba/cext
• #4619 - Implement math.{degrees, radians} for the CUDA target.
• #4614 - Add documentation for implementing new compiler passes.
• #4607 - add dependency list to docs

4. Next Release: Version 0.46.0, RC=Sept 30

• Finish pending rewrite passes
• Document best practices for constructing new compiler pipelines
• Define autodiscovery system for Numba extensions (like numba_scipy or HPAT) that don't need direct user import
• Allow opt-in dispatching of functions by literal value
• Making caching aware of transitive dependencies
• Define declarative typing system for @overload (to be used in future releases)
• Numba Runtime C API for extensions to register reference-counted memory with the runtime.
• Start using new CI system in parallel with existing one
• Priority bug fixes:
  • Low performance of JIT method calls (requested by Pandas devs)
  • Others TBD