[complib-talk] Storing compositions on the database

[Graham J. <gr...@ch...>]

We have discussed (under method categories) the topic of how do you store
compositions on the database, and Richard was going to put his thinking cap
on regarding it. If anyone on this list has any suggestions please say, as
this is a tricky problem and I accept Don and Richard's view that we have to
get this bit right from the start, as it is fundamental to the successful
design of Composition Library.

The question is:-

How do we represent compositions on the database in a way that will
encompass any composition that has been rung to date and also provide as
much flexibility as possible to minimise change for future developments?

I have been thinking about this further this week, and have come to the
conclusion that we cannot separate the way a composition is stored entirely
from either how it is entered or how it is formatted, because:

1. After a composition is input and saved, it needs to be represented in
exactly the same way if a subsequent edit is required.
2. In spliced, Method mnemonics will need to be supplied at the time of
input.
3. To format a composition in its normal compact form, information needs to
be captured at the time of input, as it would be infeasible to reconstruct a
multipart composition that has included, omitted or A/B blocks. It may also
be better to allow the submitter to influence the column headings used (e.g.
for Belfast do you have a columns headed I, O and V, or do you have a column
headed B and use I/V and I/O/V/V in the lines? In split tenors compositions
where the observation changes in each part, often course positions are used
rather than MWH.

To check whether a proposed approach works, we need to work examples through
end-to-end i.e. how input, how stored, how formatted.
   
Examples need to include:
1. Conventional BMWH compositions of regular even bell methods with several
types of call.
2. Half lead calls
3. Unusual starts
4. Spliced compositions included spliced at the half lead or another
position
5. Differential methods with multiple cycles
6. Multi-extent blocks
7. Stedman
8. Twin hunt methods
9. Other principles
10. Irregular methods
11. High numbers
12. Other oddities that we can think of

Graham