Andrew Ho wrote:

> The group have become quite silent each time after I describe the fallacy
> of the m x n worry. It seems that my proposal is not obvious and perhaps
> outrageous. So, it is highly probable that no one have tried to implement
> this silly design :-).
>

The solution for translating data built according to one template or archetype
into data which complies with another (target) template/archetype is really a
generic post-hoc translator, relying on formal reasoning about each archetype, and
rules for converting data. If archetypes are formally enough defined, and have an
interface definition (what I call a "query profile") then it should be possible to
do translation in a general manner. So you don't write a translator, but you do
write something equivalent - some rules and statements in the archetypes, to
enable a generic translator to do the work. The full analysis of what this is has
yet to be done...(to my knowledge anyway).

- thomas beale


Reply via email to