Python Version? about pysph

[Kekushke]

Sample output. Python 3.7.3 with PySPH Release 1.0a6
========================== short test summary info ============================ FAILED base/tests/test_octree.py::TestOctreeFor1DDataset::test_levels_in_tree FAILED base/tests/test_octree.py::TestOctreeFor1DDataset::test_parent_for_node FAILED base/tests/test_octree.py::TestOctreeFor1DDataset::test_plot_root - Ty... FAILED base/tests/test_octree.py::TestOctreeFor1DDataset::test_sum_of_indices_lengths_equals_total_number_of_particles FAILED base/tests/test_octree.py::TestCompressedOctreeFor1DDataset::test_levels_in_tree FAILED base/tests/test_octree.py::TestCompressedOctreeFor1DDataset::test_parent_for_node FAILED base/tests/test_octree.py::TestCompressedOctreeFor1DDataset::test_plot_root FAILED base/tests/test_octree.py::TestCompressedOctreeFor1DDataset::test_sum_of_indices_lengths_equals_total_number_of_particles FAILED parallel/tests/test_parallel.py::DumpLoadTestCase::test_dump_and_load_work_in_parallel FAILED solver/tests/test_solver_utils.py::TestOutputNumpy::test_dump_and_load_with_constants FAILED solver/tests/test_solver_utils.py::TestOutputNumpy::test_dump_and_load_with_partial_data_dump FAILED solver/tests/test_solver_utils.py::TestOutputNumpy::test_dump_and_load_works_by_default FAILED solver/tests/test_solver_utils.py::TestOutputNumpy::test_dump_and_load_works_with_compress FAILED solver/tests/test_solver_utils.py::TestOutputNumpy::test_that_output_array_information_is_saved FAILED solver/tests/test_solver_utils.py::TestOutputNumpyV1::test_load_works_with_dump_version1 FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_be_able_to_update_particle_arrays FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_correctly_update_domain FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_on_2d_data FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_when_arrays_have_different_props FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_with_changed_data FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_with_explicit_points_in_constructor FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_with_explicit_points_without_z FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_with_ghost_particles FAILED tools/tests/test_interpolator.py::TestInterpolator::test_should_work_with_multiple_arrays FAILED tools/tests/test_interpolator.py::TestInterpolator::test_that_set_interpolation_points_works FAILED tools/tests/test_sph_evaluator.py::TestSPHEvaluator::test_evaluation FAILED tools/tests/test_sph_evaluator.py::TestSPHEvaluator::test_evaluation_with_domain_manager FAILED tools/tests/test_sph_evaluator.py::TestSPHEvaluator::test_updating_particle_arrays = 28 failed, 681 passed, 267 skipped, 58 deselected, 110 warnings

[prabhuramachandran]

@Kekushke -- we build constantly on windows and Linux (and I personally use a Mac) so the requirements certainly work. Can you please let me know what platform, what python distribution you have used and the exact steps you followed? As regards Python versions, we test on 3.8 and 3.7 and it works on 3.6 as well.

[Kekushke]

Here is the environment and procedure :
Platform: Ubuntu 18.04
Python: 3.7.3
GCC: 8.3.0
Steps:
STEP1 : APT Installs build-essential python-dev libomp-dev openmpi-bin libopenmpi-dev libtrilinos-zoltan-dev python-pip python3 python3-venv
STEP2: Create python3 venv for using pip3, and activate the venv
STEP4: Download PySPH: 1.0a6.tar.gz
STEP5: Uncompressed and run pip3 --no-cache-dir requirements.txt and also for requirements-test.txt, and mpi4py.
STEP6: export ZOLTAN variables to use apt installed libraries
STEP7: make clean, make cleanall, python setup.py build and python setup.py install
STEP8: pysph -v test
Many tests fail.
Please note no special pip3 steps for pyzoltan as in the instructions at readthedocs.io. What is the right way to install pyzoltan for this version?
Anything I am missing? Suggestions for which pysph version to use? I think some requirements are not probably set to be == that could be the problem. Anything to do with gcc version?

[Kekushke]

Looks like pyzoltan is built without inclusion by pip3 install pyzoltan as per the build log. If pip3 install pyzoltan is executed before build and install, ALL tests fail.

[Kekushke]

After super clean install, still exact same error log as in posted in the 2nd message in the thread above:
28 failed, 681 passed, 267 skipped, 58 deselected, 85 warnings

[Kekushke]

Some more information:
13 errors are TypeError
8 errors are UnicodeDecodeError
6 errors are ValueError

[prabhuramachandran]

Please use master and not the pip version i.e. STEP4: Download PySPH: 1.0a6.tar.gz. The pip version will almost certainly be broken. Do the following, git clone the pysph repo, then run pip -r requirements.txt, since you need pyzoltan, pip install pyzoltan then python setup.py install or python setup.py develop.

There were a lot of bugs fixed due to changes with numpy upstream that are fixed in master. If things are still broken, can you send me at least one of the 28 test errors.

In any case this issue tells me that we absolutely have to push a release so things are easier to install -- even if we think there is a lot more to fix and do.

[Kekushke]

Many thanks to you. I will try this.

[prabhuramachandran]

Thanks, do let us know if master is broken, our CI tests all use master as do me and others using pysph.

[Kekushke]

Update: I managed to get a very old version to work for my work. It was struggle. I think for each release here on GitHub, it will be useful to have author provided requirements.txt and should separately also include dump of pip3 freeze; version of python, version of gcc, cython etc. This will allow to replicate any release made in the past. Currently, it is difficult to go back or forward due to too many varying conditions of dependencies. Great work here, but if it is made even more easy to install, it will be very useful. Many thanks again.

[Kekushke]

Update: Hi! Used MASTER using your notes. Got through, but got 6 XFAIL errors in pysph/base/tests/test_nnps.py::ExtendedZOrderNNPSSubdividedAsymTestCase::test_neighbors_* tests.
Again thanks. :-)

[prabhuramachandran]

@Kekushke -- this is a good idea, to include explicit requirements for a release. The xfail means expected failures, so I think those can be safely ignored. Those are experimental NNPS algorithms that are almost never used except if you are doing research in NNPS. So I think you should be good at this point. We are working towards a release though so hopefully this weekend or so that should happen.

[Kekushke]

Many thanks for your note. I will continue to test MASTER. I will definitely try new release when out. In new release it will be good to include not only requirements.txt but also separate dump of pip3 freeze, version number of gcc, version number of zoltan, version number of pyzoltan and all other dependencies etc. to recreate an old snapshot when needed. Again thanks :-)

[prabhuramachandran]

This is now fixed for the latest 1.0b1 release: https://github.com/pypr/pysph/releases/tag/1.0b1

[Kekushke]

Thank you so much!

…

On Sat, Mar 5, 2022 at 1:19 AM Prabhu Ramachandran ***@***.***> wrote: This is now fixed for the latest 1.0b1 release: https://github.com/pypr/pysph/releases/tag/1.0b1 — Reply to this email directly, view it on GitHub <#277 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJFBAL6WOG6UIUVCBDJKTFDU6MRLNANCNFSM4PVD7LOA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were mentioned.Message ID: ***@***.***>