Create a shareable Jupyter notebook for Chaste

[kwabenantim]

Summary

• Enable users to use PyChaste with only a browser, without having to install any additional software.
• Make it easy to create and share Chaste tutorial notebooks for others to modify and experiment with.

[kwabenantim]

Colab Notebook Issues

• With condacolab, we can potentially install chaste and dependencies from conda.
• The default Python in Colab is 3.10, but the current chaste conda package only supports Python 3.7.
• Python 3.7 support in condacolab is currently an issue: Python 3.7 conda-incubator/condacolab#45
• Overwriting the default Python in the notebook's base conda environment can successfully change the Python version to 3.7 (i.e. remove pinned packages and run conda env update -n base -f environment.yml --purge). However, the required colabtools package only works with Python 3.8+.
• Creating a constructor script for condacolab with all chaste dependencies appears to work, but fails to install, probably due to lack of support for Python 3.7.

Possible Workarounds

• Rebuild PyChaste on conda with Python 3.10 support.
• Create a JupyterHub server with chaste pre-installed.

[kwabenantim]

Rebuilding PyChaste with Python 3.10 support.

• The current chaste conda package was limited to Python 3.7 mainly because it's the highest supported version by chaste dependencies from conda-forge, particularly VTK <=8.2.
• Rebuilding the chaste conda package for Python 3.10 will require building a custom VTK for conda.
• We can try to build an older version of VTK e.g. VTK =7.0.0 with support for Python 3.10, which will also temporarily resolve the scene rendering issue for shared notebooks (Scene rendering is broken #1).

[kwabenantim]

Building VTK 7.0.0 for conda with Python 3.10 support

• VTK on conda-forge is built with the vtk-feedstock scripts.
• Building a custom version requires modifying the build scripts and applying patches to the old VTK source code.
• PR: Build VTK 7.1.1 from conda feedstock with Python3.10 #3

Issues

• sha mismatch: Fixed by updating sha string for vtk7.0.0 in meta.yaml
• Patch error for 9987_try_except_python_import.patch : Disabled optional patch (temporarily)
• CMake error - issue with gcc version: Fixed by adding gcc11-support.patch
• CMake error - could not find osmesa: Changed build variant to osmesa (temporarily) so mesa gets installed from yum
• MPI error -mpi4py/vtkmpi4py/src/mpi4py.MPI.c:106357:24: error: 'PyThreadState' {aka 'struct _ts'} has no member named 'exc_type'; did you mean 'curexc_type'?: Bumping mpich down from 4.* to 3.*
• vtkWebGLExporter error -No module named 'vtk.vtkWebGLExporter': Disabled test for vtkWebGLExporter (temporarily)
• ModuleNotFoundError: No module named 'vtkmodules':

[kwabenantim]

VTK 7.1.1 with Python3.10 built and uploaded to Anaconda: https://anaconda.org/PyChaste/vtk

[kwabenantim]

• PyChaste built with Python3.10 and uploaded to anaconda: https://anaconda.org/PyChaste/chaste

[kwabenantim]

• Now testing PyChaste in Colab