The current format is not very flexible and adding new output variables is somewhat challenging. I propose using something similar to this (unstructured grids), where the handling of the IO is performed by a third-party library (HDF5 in this case). This is already implemented as an export format #99; hopefully, we can reuse part of the work.
My idea is to separate the file into general information, nodal values, and element values. Backwards compatibility can be guaranteed if we keep the current file reader and switch readers depending on the file (file ending or magic numbers maybe?).
I will keep a list here of the data that needs to go into the file:
File ending seems fine. Then it's easier to know what reader the code is actually using. We need to think on a new extension name to substitute old ".hsol".
The current format is not very flexible and adding new output variables is somewhat challenging. I propose using something similar to this (unstructured grids), where the handling of the IO is performed by a third-party library (HDF5 in this case). This is already implemented as an export format #99; hopefully, we can reuse part of the work.
My idea is to separate the file into general information, nodal values, and element values. Backwards compatibility can be guaranteed if we keep the current file reader and switch readers depending on the file (file ending or magic numbers maybe?).
I will keep a list here of the data that needs to go into the file: