By default, HDF5 objects are stored with an internal creation timestamp. This means that if the same file is written twice, the content, and hence the checksum, will change. This can be disabled: (http://docs.h5py.org/en/stable/high/group.html)
create_dataset(...,track_times=False)
It's possible there are other timestamps stored, not associated with datasets, in which case I can't immediately find anything about them on the h5py website.
I don't think the dataset creation time is useful for us, and omitting it means we can tell that a repeated run produced the same result (via the checksum).
By default, HDF5 objects are stored with an internal creation timestamp. This means that if the same file is written twice, the content, and hence the checksum, will change. This can be disabled: (http://docs.h5py.org/en/stable/high/group.html)
create_dataset(...,track_times=False)
It's possible there are other timestamps stored, not associated with datasets, in which case I can't immediately find anything about them on the h5py website.
I don't think the dataset creation time is useful for us, and omitting it means we can tell that a repeated run produced the same result (via the checksum).