PCL tests failed after build from source using conda python 3.7 on Ubuntu 18.04

[szx0112]

I build and install pclpy as in docs. I tried to make it works with python 3.7 in conda env. After the installation, the auto- test showed many failures.

=========================== short test summary info ============================
FAILED test_features.py::test_compute_normals_simple - OSError: No such file ...
FAILED test_features.py::test_compute_normals_simple_given_output - OSError: ...
FAILED test_features.py::test_difference_of_normals_estimation_simple - OSErr...
FAILED test_features.py::test_fpfh_simple - OSError: No such file or director...
FAILED test_filters.py::test_extract_indices_simple - OSError: No such file o...
FAILED test_filters.py::test_radius_outlier_removal_simple - OSError: No such...
FAILED test_filters.py::test_radius_outlier_removal_alias - OSError: No such ...
FAILED test_filters.py::test_radius_outlier_removal_indices - OSError: No suc...
FAILED test_filters.py::test_radius_outlier_removal_negative - OSError: No su...
FAILED test_io.py::test_simple_io_pcd - assert 0 == 397
FAILED test_io.py::test_las_xyz - OSError: No such file or directory: '/home/...
FAILED test_io.py::test_las_rgb - OSError: No such file or directory: '/home/...
FAILED test_io.py::test_las_intensity - OSError: No such file or directory: '...
FAILED test_io.py::test_las_save_extra_dims - FileNotFoundError: [Errno 2] No...
FAILED test_io.py::test_las_read_normals - OSError: No such file or directory...
FAILED test_io.py::test_las_write_with_offset - FileNotFoundError: [Errno 2] ...
FAILED test_io.py::test_las_read_with_offset - FileNotFoundError: [Errno 2] N...
FAILED test_io.py::test_las_to_pcd - OSError: No such file or directory: '/ho...
FAILED test_octree.py::test_voxel_centroid_simple - OSError: No such file or ...
FAILED test_octree.py::test_voxel_centroid_api - OSError: No such file or dir...
FAILED test_octree.py::test_voxel_centroid_api_rgba - OSError: No such file o...
FAILED test_segmentation.py::test_supervoxel_clustering - OSError: No such fi...
FAILED test_segmentation.py::test_region_growing - OSError: No such file or d...
FAILED test_segmentation.py::test_region_growing_simple - OSError: No such fi...
FAILED test_segmentation.py::test_euclidean_cluster - OSError: No such file o...
FAILED test_surface.py::test_moving_least_squares_normals - OSError: No such ...
FAILED test_surface.py::test_moving_least_squares_alias - OSError: No such fi...
FAILED test_surface.py::test_moving_least_squares_no_normals - OSError: No su...
FAILED test_surface.py::test_moving_least_squares_no_normals_omp - OSError: N...
FAILED test_surface.py::test_moving_least_squares_with_output - OSError: No s...
FAILED test_surface.py::test_greedy_projection_triangulation_simple - OSError...
ERROR test_io.py::test_las_save_extra_dims - NameError: name 'WindowsError' i...
ERROR test_io.py::test_las_write_with_offset - NameError: name 'WindowsError'...
ERROR test_io.py::test_las_read_with_offset - NameError: name 'WindowsError' ...
=================== 31 failed, 58 passed, 3 errors in 1.70s ====================

After installation, I can import pclpy without error. However, I find I am not able to proceed some test cases. For example, the high-level code output1 = pclpy.moving_least_squares(pc, search_radius=0.5, compute_normals=False)

runs smoothly without error.

However, when I run it through PCL wrapper.

mls = pcl.surface.MovingLeastSquaresOMP.PointXYZRGBA_PointNormal()

It shows pcl has no attribute MovingLeastSquaresOMP.

Any comment is appreciated

[szx0112]

Switched to python-pcl due to lack of documentations.

[VisionaryMind]

Switched to python-pcl due to lack of documentations.

Python-pcl does not implement OpenMP, which is one of the reasons this library is unique. Were you eventually able to get OMP working with pclpy? Python-pcl is no longer maintained and, for all intents and purposes, obsolete. I do not see any recent activity on this repository, so it's possible pclpy is also soon to be obsolete.

This brings up the question --- what do we do if we need to implement multi-processing in python-based pipelines? It looks to me that in the absence of a functional wrapper, we will have to resort to developing our own modules.