Extending the Data Structure

Before contemplating a permanent modification to the SNOMAN data structure contact the SNOMAN Coordinator to discuss the matter. To extend the data structure requires the following steps. In these instructions

xx is used to denote the name of the new bank.

yy the existing bank that supports it.

All the steps below, including the production of documentation, must be carried out before submitting the software for inclusion into SNOMAN.

1. Decide on the bank's content and format. The format should be kept as simple as simple as possible and, although not mandatory, it is recomended that the bank be organised to keep all data words of the same type (e.g. all floating point) together, possibly even leaving spare slots for future development. This keeps the format simple and minimises the memory overhead for the bank.

2. Decide how the bank fits into the existing data structure i.e. what existing bank supports it.

3. Choose a new bank name (no more than 4 letters) and a bank number if appropriate.

4. Choose bank word mnemonics and create the files:-
   • bank_xx.html

   • bank_xx.inc

   and modify the equivalent bank_yy files. Update the bank mnemonics section of:-

   • symbol_table.dat

5. Add array MMxx(5) to the bank definition file bnkdef.inc.

6. Make the following changes to inzebr.for:-
   • Add array Mxx(5) and use a DATA statement to define the first 4 words: its name, number of links, number of structural links and number of data words, leaving word 5 (the I/O characteristic) set to 0.

   • Modify the DATA statement for Myy for the new structural link.

   • Add 2 lines, calling UCOPY to copy Mxx to MMxx and calling MZFORM to set up the I/O characteristic (see the ZEBRA User Guide for more information on MZFORM).

7. If a public reference link is needed for the bank modify the file z.inc as follows:-
   • Add a description comment for the new bank.

   • Declare an integer Lxx.

   • exi EQUIVALENCE the link to the first free slot in Z_REF_LINKS.

   • Update the titles bank NTPC. This defines the relationship between banks and allows data from the new bank to be stored in user defined N-tuples.

The bank can now be created using such routines as MZLIFT (for examples using MZLIFT see MCG_GENERATOR).