Graham Markall comments

Results 776 comments of


                                            Graham Markall

NotImplementedError When creating an empty structured array in parallel

Also marking as "doc" since we should state this limitation, and leaving "ice" on because we shouldn't be presenting a `NotImplementedError` in this situation.

Move Numba to LLVM's NewPassManager

> I'm not sure what work needs to be done on Cuda side The CUDA target doesn't use the pass managers at all - this is expected to be handled...

Move Numba to LLVM's NewPassManager

/azp run

Move Numba to LLVM's NewPassManager

gpuci run tests

Move Numba to LLVM's NewPassManager

/azp run

Move Numba to LLVM's NewPassManager

/azp run

Move Numba to LLVM's NewPassManager

gpuci run tests

Move Numba to LLVM's NewPassManager

gpuci run tests

Move Numba to LLVM's NewPassManager

From discussion out-of-band: we will look at whether we can follow up with another PR for the pass timings.

objects from the jitclass lose dynamic features about the class attributions

> A example of changing class attributions: Members of a jitclass have to be provided to the `jitclass` decorator, or can be inferred from type annotations. See: https://numba.readthedocs.io/en/stable/user/jitclass.html#the-decorator-jitclass