Gmsh 4.4.1 in FEniCS? Meshio

mesh
15

I had the same problem in the FEnics Docker container that I downloaded from the downloads page. To install meshio one would create a different docker image using the process mentioned in the FEnics Containers page.The contents of the Dockerfile using instructions here would look like,

FROM quay.io/fenicsproject/stable:current
USER root
RUN apt-get -qq update && \
    apt-get -y upgrade && \
    apt-get clean && \
    pip install --upgrade pip && \
    pip install meshio[all] && \
    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
USER root

However this would give the error

ValueError: Invalid high library version bound (invalid high library version bound)

This is because the HDF5 version that ubuntu-18(container used by Fenics) installs by default is 1.10.0, which can be easily verified by the following command

dpkg -l | grep hdf5
ii  libhdf5-100:amd64                      1.10.0-patch1+docs-4                amd64        Hierarchical Data Format 5 (HDF5) - runtime files - serial version
ii  libhdf5-mpich-100:amd64                1.10.0-patch1+docs-4                amd64        Hierarchical Data Format 5 (HDF5) - runtime files - MPICH2 version
ii  libhdf5-mpich-dev                      1.10.0-patch1+docs-4                amd64        Hierarchical Data Format 5 (HDF5) - development files - MPICH version

But pip install meshio[all] would install h5py-2.9 which would require HDF5-1.10.3 or above.The one h5py version that would work with this is 2.7.1.But installing this using pip gives the following error ,reasons for which I am unable to find out now.

Segmentation fault (core dumped)

So the workaround is installing h5py from ubuntu repository only to maintain continuity of versions.The Dockerfile could look like

FROM quay.io/fenicsproject/stable:current
USER root
RUN apt-get -qq update && \
    apt-get -y upgrade && \
    apt-get clean && \
    apt-get -y install python3-h5py && \
    pip install --upgrade pip && \
    pip install meshio[all] && \
    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
USER root

That solved the problem.Obviously, the workaround can be used in cases of direct installation .Just install h5py from official repository and follow usual steps as mentioned

1 Like