Serialising vs modelling

[Martin Hill]

> But at least you need in the interface a method
> public string serialise()
> or
> public String toString()
> or whatever name method, but a one which can propose a way to exchange
> a whole object in once.

Really we should be, as you say, first modelling the relationships between our 
various bits of data - the structure of our data.

When we come to serialising it, we don't need to worry about the actual method 
call made on an object - this is an implementation issue.  It might be java 
coders make use of the 'serializable' interface, other languages do other 
things, but this is independent of modelling or indeed working out the form of 
the serialisation (eg the XML elements).

Indeed, to be clear about this, we shouldn't be concerned at this modelling 
stage about the *method names* at all.  We should just be interested in the data 
relationships, both activity and static.  Method names will largely depend on 
the conventions of the languages being used - bear in mind we will probably have 
FORTRAN implementations as well as Java.

Which might help to simplify some of the discussion... if we're really lucky... :-)

Cheers,

MC