Minutes_2018_12_11

Numba Meeting: 2018-12-11

Attendees: Stan, Siu, Stuart, Oscar, Todd

Release Plan
- LLVM 7
  - shipped dev build on almost all platforms
  - failing on cuda-sim + real hw (not a release blocker)
- lld patch?
  - proposal on adding lld binding in llvmlite
  - needs review
- buildfarm status
  - build machines online but lack VPN access
  - will likely delay RC release
Numba master repo branch prune
- agreed
Policy on impl NumPy function support that requires strings (Py2.7 is a problem)
- Implement bytes thus py2.7 "string"
2018 digest
- blogpost + notebook
- Stuart to make list of functionality
LLVM Perf events effort
- reason perf doesn't work is missing notify perf events listener in MCJIT
- OrcJIT already have the listener. We may try the MCJIT-OrcJit-Replacement.

#3584 - Numba fails with numba.errors.LoweringError: Failed in object mode pipeline
#3582 - Error when calling parallelized function from compiled function
- short term write guard in parfors against optional (+ probably F order?)
  - type-inference enhnacement
#3579 - Typing empty list across branches breaks array ctor
- type inference error
#3575 - Error in documentation of numba.objmode context manager
#3574 - simulate bug func_or_sig vs fn_or_sig named parameter
#3573 - Incorrect LoweringError when assigning to complex128 .imag
#3571 - DeviceNDArray needs to expose 'contiguous' and/or 'c_contiguous' flag as property(s) on flagsobj
#3569 - Fortran array signature does not accept array that is both F and C contiguous
#3565 - Numba 0.41.0 bug: Cannot unify Literalint and int64
#3562 - itertools support

#3568 - Function using np.roots and np.real doesn't compile with nopython=True
- should add np.asarray

#3583 - Use ir.unknown_loc for unknown Loc, as #3390 with tests
- ready to merge
#3581 - Fixing minor typo in reference/types.rst
- ready to merge
#3580 - Use alias information when determining whether it is safe to reorder statements.
- Ehsan to review
#3578 - Align cuda and cuda simulator kwarg names.
- ready to merge
#3577 - Fix example and docs typos for objmode context manager.
- ready to merge
#3576 - Add special copy routine for NumPy out arrays
- ready to merge
#3572 - Support for np.ptp
- in-progress
#3570 - Minor documentation fixes for numba.cuda
- ready to merge
#3567 - Fix unifying literal types in namedtuple
- waiting
#3564 - Docs: 5minguide: Correct @njit alias vs @jit
- Closed
#3563 - Support for np.roll
- Waiting on author
- basically fine

#3566 - Skip alignment tests on armv7l.
- note: need to fix type-system to deal with alignment properly

#3561 - Fixing the boolean copy from data
- In progress
#3558 - Fixes array analysis for inplace binary operators.
- Needs review
#3556 - [WIP] Support for np.trim_zeros
- in progress
#3552 - Handle broadcast arrays correctly in host->device transfer.
- Siu needs to review
#3549 - ParallelAccelerator caching improvements
- ~~Siu needs to add skip tests~~
- Needs review
#3519 - WIP: fix-3457 support of numpy repeat.
#3520 - Fix @stencil ignoring cval if out kwarg supplied.
- Discussion on going
#3468 - Add support for np.clip and ndarray.clip.
- depends on fix on @overload_method for kwargs
3450 - [WIP] generated_jit for CUDA kernels
3392 - Launch and attach gdb directly from Numba.
- ready for review
3162 - Support constant dtype string in nopython mode in functions like numpy.empty.
- Need to resolve #3195
3134 - [WIP] Cfunc x86 abi
- Needs re-review
3046 - Pairwise sum implementation.
#2999 - Support LowLevelCallable
#2942 - Fix linkage nature (declspec(dllexport)) of some test functions
#2894: [WIP] Implement jitclass default constructor arguments.
#2817: [WIP] Emit LLVM optimization remarks

===========================