In order to extend the existing software to deal with new algorithms the serialisation mechanism needs to be extended to parse any new data structures. This can be done by following the definitions and examples given above. The new SWAP_STRING definition statement should be included underneath the structure definitions in the relavant header file. In addition the serialisation routines and swap string variables need to be stored, in the tina_data_info_array by the function define_all_ts_handlers(), in the user defined slots (tset.c). When porting the serialisation code to previously unsupported processors slightly more care will need to be taken (see machine dependence below).