Closed bdice closed 1 year ago
https://github.com/NVIDIA/cuda-python/blob/6782a648bc728cdf55e280eb20631748b8c15627/setup.py#L104
This line assumes that pyparsing has been imported, but it is not imported in the setup.py script.
pyparsing
setup.py
This appears to be a transitive dependency of pyclibrary, and cuda-python's setup.py expects version 2.4.7. However, the latest version of pyclibrary has a much looser pinning on pyparsing>=2.3.1,<4: https://github.com/MatthieuDartiailh/pyclibrary/blob/1d4dbfc207afee3fd80b72e94c60d47d8263d49a/setup.py#L48
pyclibrary
pyparsing>=2.3.1,<4
As of this writing, the latest pyparsing is 3.0.9. https://github.com/pyparsing/pyparsing/releases/tag/pyparsing_3.0.9
I'm not sure how to reconcile this. Is the error message out of date? Does a dependency on pyparsing need to be added? Is pyparsing 3.0.9 allowable or not?
Here is an example traceback I encountered while attempting to package cuda-python for CUDA 12 on conda-forge (work in progress: https://github.com/conda-forge/cuda-python-feedstock/pull/33).
Processing $SRC_DIR Added file://$SRC_DIR to build tracker '/tmp/pip-build-tracker-8n7yddr7' Running setup.py (path:$SRC_DIR/setup.py) egg_info for package from file://$SRC_DIR Created temporary directory: /tmp/pip-pip-egg-info-3mflzan5 Preparing metadata (setup.py): started Running command python setup.py egg_info Parsing headers in "/home/conda/feedstock_root/build_artifacts/cuda-python_1682019216274/_h_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold/targets/x86_64-linux/include" (Caching False) Missing header cuda.h Missing header cudaProfiler.h Missing header cudaEGL.h Missing header cudaGL.h Missing header cudaVDPAU.h Parsing driver headers Missing header driver_types.h Missing header vector_types.h Missing header cuda_runtime.h Missing header surface_types.h Missing header texture_types.h Missing header library_types.h Missing header cuda_runtime_api.h Missing header device_types.h Missing header driver_functions.h Missing header cuda_profiler_api.h Missing header cuda_egl_interop.h Missing header cuda_gl_interop.h Missing header cuda_vdpau_interop.h Parsing runtime headers Missing header nvrtc.h Parsing nvrtc headers Traceback (most recent call last): File "<string>", line 2, in <module> File "<pip-setuptools-caller>", line 34, in <module> File "/home/conda/feedstock_root/build_artifacts/cuda-python_1682019216274/work/setup.py", line 103, in <module> if pyparsing.__version__ != '2.4.7': NameError: name 'pyparsing' is not defined error: subprocess-exited-with-error × python setup.py egg_info did not run successfully. │ exit code: 1 ╰─> See above for output.
CI link (this link will eventually expire, so the relevant portion is copied above): https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=693718&view=logs&j=d0d954b5-f111-5dc4-4d76-03b6c9d0cf7e&t=841356e0-85bb-57d8-dbbc-852e683d1642
Resolved in CUDA Python 12.2.0. Closing.
Thanks @vzhurba01!
https://github.com/NVIDIA/cuda-python/blob/6782a648bc728cdf55e280eb20631748b8c15627/setup.py#L104
This line assumes that
pyparsing
has been imported, but it is not imported in thesetup.py
script.This appears to be a transitive dependency of
pyclibrary
, and cuda-python'ssetup.py
expects version 2.4.7. However, the latest version ofpyclibrary
has a much looser pinning onpyparsing>=2.3.1,<4
: https://github.com/MatthieuDartiailh/pyclibrary/blob/1d4dbfc207afee3fd80b72e94c60d47d8263d49a/setup.py#L48As of this writing, the latest pyparsing is 3.0.9. https://github.com/pyparsing/pyparsing/releases/tag/pyparsing_3.0.9
I'm not sure how to reconcile this. Is the error message out of date? Does a dependency on
pyparsing
need to be added? Ispyparsing
3.0.9 allowable or not?Here is an example traceback I encountered while attempting to package cuda-python for CUDA 12 on conda-forge (work in progress: https://github.com/conda-forge/cuda-python-feedstock/pull/33).
CI link (this link will eventually expire, so the relevant portion is copied above): https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=693718&view=logs&j=d0d954b5-f111-5dc4-4d76-03b6c9d0cf7e&t=841356e0-85bb-57d8-dbbc-852e683d1642