Minutes_2018_11_27

Numba Meeting: 2018-11-27

Attendees: Stuart, Todd, Siu, Ehsan

0.41.0 Release status
- Checklist https://github.com/numba/numba/issues/3537
- Ready to tag and build numba
0.42.0 Release Planning
- Shorter cycle
  - RC Dec 10; Final Dec 17
- LLVM 7
  - 1 windows bug
- Other string features/fixes
- Caching bugs:
  - parallel=True no-threads caching issues

#3542 - Anything else than os.cpu_count() for NUMBA_NUM_THREADS threated as 1
- could be import order
#3541 - numba.errors.LoweringError: Failed in nopython mode pipeline (step: nopython mode backend)
- Invalid List.append pass type check
#3537 - Numba 0.41.0 final release checklist
#3534 - Parfors reordering print statement in fusion.
- issues with aliasing of slice
#3533 - merging signatures and layout in guvectorized decorator
- feature request
#3529 - Lowering error when using njit on inner function and not passing arguments
- because some types can't be compile-time constants
- documented behavior
- need error message improvements
#3528 - numpy.median gives error in stencil
- codegen bug
- not median specific. issue with any reduction function it seems.
#3525 - "This should not have happened" ~ checking for value in numpy array
- bug in error checking/reporting
#3524 - yield + for issue
- bug confirmed
#3522 - prange induces race conditions, parallelization slows down code dramatically with range
- another reason for parallel diagnostic, add prange in range warning
- and more documentation/tutorial on parallel-accelerator
#3521 - Numba stencils unexpectedly upconvert float32 to float64 differently than numpy
#3518 - @stencil ignores cval when out kwarg supplied.
- PR: #3520
#3517 - Unused functions abort compilation

#3540 - numba.errors.LoweringError: Failed in nopython mode pipeline (step: nopython mode backend)
#3523 - @njit is mentioned, but not explained, in chapter 1 of the user manual

#3538 - Avoid future C-level assertion error due to invalid visibility
#3532 - Unregister temporarily pinned host arrays at once
- Siu will review
#3531 - Fix jitclass method inlining and avoid unnecessary increfs
- Siu will review
#3520 - Fix @stencil ignoring cval if out kwarg supplied.
#3519 - WIP: fix-3457 support of numpy repeat.
#3516 - [WIP] Typeof dtype values
- Siu will review

#3468 - Add support for np.clip and ndarray.clip.
- depends on fix on @overload_method for kwargs
3437 - Changes to accommodate LLVM 7.0.x
3450 - [WIP] generated_jit for CUDA kernels
3392 - Launch and attach gdb directly from Numba.
- needs work
3390 - typeinfer: use unknown_loc object instead of string literal
3162 - Support constant dtype string in nopython mode in functions like numpy.empty.
- Need to resolve #3195
3160 - First attempt at parallel diagnostics
- Stuart will implement Todd's suggestion
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

Merged old PRs

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