As my initial naming scheme for the IMP methods in the hdf5 module seems to have confused everyone, I've changed it to a new, hopefully clearer, one. Sorry about the change, but I think it will make the model much easier to figure out. The new methods treat the files as objects rather than IO in light of the persistent state that is associated with them.

Changes are - write_X -> add_X - save_configuration -> save_frame - load_configuration -> load_frame - read_all_X -> create_X

To be more consistent with similarly named IMP methods, all methods take the RootHandle as the first argument. If they take a frame, it is the second argument.

Hope this helps things. The module overview docs have been improved a bit too.