The reading component is less useful than one might think as clients still need to implement the writing component and that's already half the battle. If we make it validation-only, we could simplify the codebase and get rid of some unnecessary checks for HDF5 (except for variable-length strings and VLS arrays).
The reading component is less useful than one might think as clients still need to implement the writing component and that's already half the battle. If we make it validation-only, we could simplify the codebase and get rid of some unnecessary checks for HDF5 (except for variable-length strings and VLS arrays).