Lists Home |
Date Index |
"Alaric B. Snell" wrote:
> if you were into 'loose coupling' in the sense that many XML-for-data fans seem
> to use it, then you would *always* have to explicitly convert data models
> between interchange and processing, and hence never use processing tools that
> don't require conversion, n'est pas?
C'est vrai, except that I would call that the necessary work of data
instantiation which must be done to satisfy the particular input demands of any
specialized process on every occasion of processing any instance data. If you
build your process's data expectations to the common denominator of the data
which arrives then you sacrifice the very autonomous, idiosyncratic expertise
which makes your process uniquely valuable: common denominator => common value;
unique, expert data selection and instantiation => uniquely valuable processing.