In io4dolfinx, there is the possibility to checkpoint a mesh with hdf5 backend. In dolfinx main repo, vtkhdf can also be used to write solution files, even if some work is still in progress (e.g. Improvements to VTKHDF format I/O · Issue #3582 · FEniCS/dolfinx · GitHub). Would it make sense to expose also some hdf5 compression options in the api? This could help in reducing the required storage for long time series or huge meshes.