Guilherme Leobas

> Please let me know what you think of this approach. Hi @luk-f-a, appreciate your efforts in this PR. I like this new approach. Also, could you take a look...

Hi @dlee992, I believe this PR is waiting on a second review from Stuart. But feel free to give your comments as well.

> BTW, can I directly push my patches into this PR? Didn't do this before, and not sure how to do this. You can create a branch based on Lucio's...

Current failure happens on `llvmlite/bindings/targets.py::TargetData.get_pointee_abi_size`. Below is the last five frames of the stacktrace: ```python [48] /Users/guilhermeleobas/git/numba/numba/core/base.py(1220)wrapper() -> return fn(*args, **kwargs) [49] /Users/guilhermeleobas/git/numba/numba/experimental/structref.py(305)codegen() -> alloc_size = context.get_abi_sizeof(alloc_type) [50] /Users/guilhermeleobas/git/numba/numba/core/base.py(1105)get_abi_sizeof() ->...

Yes!

Some tests are failing with the following traceback: Traceback ``` ====================================================================== ERROR: test_pytree_inputs_cpu (__main__.TestJvpCPU.test_pytree_inputs_cpu) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/guilhermeleobas/git/pytorch/torch/testing/_internal/common_utils.py", line 2739, in wrapper method(*args, **kwargs) File "/home/guilhermeleobas/git/pytorch/torch/testing/_internal/common_device_type.py",...

The pass_manager failures do not seem to be related to this PR and I cannot reproduce them locally

@jeanschmidt, ~I believe this PR was already reverted.~ here: https://github.com/pytorch/pytorch/pull/122328 edit: Nevermind, commit is still there. #122328 wasn't merged.

@pytorchbot merge

@pytorchbot merge