Python bindings for JIGSAW: a Delaunay-based unstructured mesh generator.

pathlib is listed as a requirement for all python versions, but it is part of the system in python >=3.4 (i.e. all currently supported version of python). Because of this dependency, pip check fails when jigsawpy is installed for python 3.x. I would request that it either be removed as a dependency in the next release or that it be constrained to python versions before 3.4.

https://docs.python.org/3/library/pathlib.html

Because of the line: https://github.com/dengwirda/jigsaw-python/blob/b6e455c22696568f90a7428a49e03c980892c923/setup.py#L135 the tests directory is being included in the conda-forge (and presumably the pypi) package. This directory should be excluded, or better yet only jigsawpy should be included:

packages=find_packages(include=['jigsawypy', 'jigsawpy.*'])

This helps Windows find jigsaw.dll without a full path name.

A similar call seems not to be needed for Linux and not to work for OSX.

I created a patch identical to this in the conda-forge recipe: https://github.com/conda-forge/staged-recipes/pull/10842/commits/b8d1bc4d9cffadf1c8ca62069d5f6e676cc271e4

• Update to jigsaw-0.9.15.x.
• Fix an i/o bug in handling of jig_t objects.
• Fix an array-size bug in orthoball routines.
• Add support for "jumbled" meshes.
• Remove pathlib from REQUIRED in setup.py.
• Support ellipsoidal-type initial conditions.
• BREAKING: encode h(x) values as 32-bit floating point data to support "very-large" h(x) types.

Hello, I was wondering if there is any way to speed up the rDT mesh and Mesh optm parts of the jigsaw-python subtree source. Currently, I am compiling with gcc-11, using the standard CMake approach (maybe there's an optimization switch I haven't enabled?). I tried compiling it with intel 2018 and 2019 but I wasn't successful with the compilation. Any ideas on achieving some speed-ups will be greatly appreciated. Thanks, -J.

I don't find the exceptions raised by jigsawpy to be very intuitive, and I think that might be true for most python users.

As an example: https://github.com/dengwirda/jigsaw-python/blob/f87571945975b4882d8c27bdfff9275f6e46baa8/jigsawpy/savemsh.py#L613

I would expect something like:

    if (not isinstance(name, str)):
        raise TypeError("Argument 'name' to savemsh should be of type str")

As I understand it, the generic Exception class is only intended as a base class for exceptions and shouldn't be used directly.

Python is case sensitive so the use of NAME here also led to some confusion for me.

I was told that opts.verbosity = -1 should mean that no output went to the terminal window, just to the log file. But I am finding that I still get terminal output when I have this option set. I verified that the .jig file does, indeed have:

  VERBOSITY=-1

but I still see something like this to the terminal:

#------------------------------------------------------------
#
#   ,o, ,o,       /                                 
#    `   `  e88~88e  d88~\   /~~~8e Y88b    e    / 
#   888 888 88   88 C888         88b Y88b  d8b  /   
#   888 888 "8b_d8"  Y88b   e88~-888  Y888/Y88b/  
#   888 888  /        888D C88   888   Y8/  Y8/     
#   88P 888 Cb      \_88P   "8b_-888    Y    Y    
# \_8"       Y8""8D                             
#
#------------------------------------------------------------
# JIGSAW: an unstructured mesh generation library.  
#------------------------------------------------------------
...

It seems the intention is that one can create an icosahedral mesh without passing a mesh object: https://github.com/dengwirda/jigsaw-python/blob/b6e455c22696568f90a7428a49e03c980892c923/jigsawpy/init.py#L105 but when I try this:

        jigsawpy.cmd.icosahedron(opts, subdivisions)

I get:

Traceback (most recent call last):
  File "/home/xylar/miniconda3/envs/compass_jigsaw_mesh/bin/compass", line 33, in <module>
    sys.exit(load_entry_point('compass', 'console_scripts', 'compass')())
  File "/home/xylar/code/compass/separate_mesh_steps/compass/__main__.py", line 63, in main
    commands[args.command]()
  File "/home/xylar/code/compass/separate_mesh_steps/compass/run/serial.py", line 281, in main
    run_step()
  File "/home/xylar/code/compass/separate_mesh_steps/compass/run/serial.py", line 249, in run_step
    test_case.run()
  File "/home/xylar/code/compass/separate_mesh_steps/compass/ocean/tests/global_ocean/mesh/__init__.py", line 109, in run
    super().run()
  File "/home/xylar/code/compass/separate_mesh_steps/compass/testcase.py", line 190, in run
    self._run_step(step, self.new_step_log_file)
  File "/home/xylar/code/compass/separate_mesh_steps/compass/testcase.py", line 306, in _run_step
    step.run()
  File "/home/xylar/code/compass/separate_mesh_steps/compass/mesh/spherical.py", line 350, in run
    self.make_jigsaw_mesh(subdivisions)
  File "/home/xylar/code/compass/separate_mesh_steps/compass/mesh/spherical.py", line 376, in make_jigsaw_mesh
    jigsawpy.cmd.icosahedron(opts, subdivisions)
  File "/home/xylar/miniconda3/envs/compass_jigsaw_mesh/lib/python3.9/site-packages/jigsawpy/__init__.py", line 106, in icosahedron
    return jigsaw.icosahedron(
  File "/home/xylar/miniconda3/envs/compass_jigsaw_mesh/lib/python3.9/site-packages/jigsawpy/jigsaw.py", line 597, in icosahedron
    raise Exception("Incorrect type: MESH.")
Exception: Incorrect type: MESH.

It seems that this check: https://github.com/dengwirda/jigsaw-python/blob/b6e455c22696568f90a7428a49e03c980892c923/jigsawpy/jigsaw.py#L519-L520 should check for None:

    if (mesh is not None and not isinstance(mesh, jigsaw_msh_t)):
        raise Exception("Incorrect type: MESH.")

Are the method(s) used to install binaries sufficient? Currently, two options are supported:

1. setup.py build_external, etc builds and installs the jigsaw-cpp binaries within jigsawpy itself.
2. If no local binaries are found, a global search of the machine path is done to pick them up in, e.g. usr/local/bin, etc --- used by the conda-based workflow.

@jreniel has suggested installing to the python environment directory, either as a 3rd option, or, perhaps(?) a replacement for 1.

Should also work out what's needed to make jigsawpy available for direct install via pip.