Minutes_2019_01_08
Attendees: Siu, Stuart, Todd, Stan, Ehsan
-
ARM machine CI/CD offer https://github.com/numba/numba/issues/3308#issuecomment-451536356
- Stan to reply
-
CI cleanup:
- Remove extra services from numba and llvmlite repos
- Remove unneeded badges from README's / docs (like appveyor)
- Decide how to add test coverage monitoring back in
- Travis CI has good integration with other services, but too slow to run full test suite with coverage turned on
- Not sure how to handle secret tokens on Azure yet
- Secret tokens can't be enabled for PRs (or someone could commit test code changes to exfiltrate the token) so not sure how to have code coverage reporting on PRs
-
Numba 0.43 planning:
- Feb 26 RC date?
- Initial dictionary support
- Need to focus on finishing / closing stale PRs
- Investigate speeding up test suite (large sections of serial test execution)
- What other major features / initiatives?
- Time to start The Great Numba Code Cleanup?
- Need to augment
@overload
with more type information first - Then we can be sure to add it as we refactor
- Need to augment
- #3666 - LLVM IR parsing error in a large function
-
#3665 - PyPI wheels for numba contain incorrect hashes which lead to resolver failure on strict hash checking
- double check linux
- #3664 - Trace var types in undefined lists for better error message
- #3659 - Array comprehension fails against multidimensional array
- #3658 - KeyError when caching with multiple files
- #3656 - Threading improvements in roadmap
- #3653 - Return of tuple with optional type
- #3652 - Lowering error with automatic parallelization
- #3650 - nopython mode -- numpy.where function not working with True/False yield values supplied
-
#3649 - np.linalg.norm impl kwarg
ord
broken - #3644 - Dictionary support
- #3642 - gdb launch potential race condition (safe, just undesirable)
- #3641 - trouble with Literal{T} & undefined local
- #3639 - [Docs?] Numba creates multiple Python objects for an object returned more than once
- #3637 - config.threading_layer = 'intel-openmp'
- #3636 - Add outdated warning to old docs.
- #3635 - Forcing parallel and nogil for CC
- #3634 - kmeans list comprehension
- #3632 - Using CC cant use function defined in own module
- #3629 - Time for first njit compilation
- #3628 - why not I can complie the code using numba.jit
-
#3625 - Segfault using
generated_jit
andprofiler
- #3619 - FakeCUDAAtomic Add, Min and Max methods do not return the old value
- #3654 - List of lists with array type error
- #3647 - How to write a cuda kernel function that performs a regression(least square estimator )for each thread?
- #3626 - Numba 0.42 final release checklist
- #3623 - Issue with list comprehension
- #3621 - Nested ctypes calls using njit(cache=True) results in segmentation fault
- #3620 - Unsupported nested memory-managed object
- #3663 - Removes coverage computation from travisci builds
- #3661 - Access C Struct as Numpy Structured Array
- #3660 - Added Ignore message to test numba.tests.test_lists.TestLists.test_mul_error…
- #3657 - [WIP] np.max, np.min, np.nanmax, np.nanmin - support for complex dtypes
- #3655 - WIP: support numpy.average
- #3648 - WIP: Fail on @overload signature mismatch.
-
#3646 - [WIP] Log internal errors encountered in typeinfer
- looking for feedback
- #3645 - Enforce and fix that AbstractTemplate.generic must be returning a Sig…
-
#3640 - [WIP] PassThruContainer
- provides feedback for user
- #3638 - Fix frame injection in the dispatcher tracing path.
- #3633 - Support for np.extract
- #3630 - np.where with broadcasting
- #3627 - Support for np.trapz
-
#3615 -
Fix gdb support on RPiclosed
- #3662 - Replace six with numba.six
- #3651 - fix link to berryconda in docs
- #3643 - Remove support_literals usage
- #3631 - Update changelog
- #3624 - Fix invalid enabling of caching for function that references dynamic external addresses
- #3622 - Handle User Defined Exceptions with args.
- #3618 - Skip gdb tests on ARM.
- #3617 - Draft roadmap as of Dec 2018
- #3616 - Skip gdb tests on ARM.
- #3614 - Resolve #3586
-
#3604 - [WIP] Support for np.interp
-
#3592 - WIP: Add dead branch prune before type inference.
-
#3561 - Fixing the boolean copy from data
- In progress
-
#3556 - [WIP] Support for np.trim_zeros
- in progress
-
#3519 - WIP: fix-3457 support of numpy repeat.
-
#3468 - Add support for
np.clip
andndarray.clip
.- depends on fix on @overload_method for kwargs
-
3450 - [WIP] generated_jit for CUDA kernels
-
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
===========================
- Initial dictionary support
- ???