David Forslund wrote:
>
> At 02:55 PM 11/22/99 -0500, Gunther Schadow wrote:
> >...
>
> >CORBAmed is like any other interface standard, you need to twist
> >you internal model and build a mapping to CORBAmed's model.
>
> But you can go from UML to the interface model directly, so the "twisting"
> is primarily in the language mapping from the IDL to the particular language.
>
Might I suggest that the twisting is a matter of perspective as to what
exists as what I will call the primary or fundamental entities of the
system design. I sometimes call these atoms, but that is slippery
itself. Dave is coming from the perspective of interfaces being the
lowest meaningful level of the system and Gunther, I think, is coming
from viewing the data model in the sense of a SQL like table
implementation as the lowest meaningful level. SO this particular
discussion is crossing over itself at the conceptual level.
Way back in my early database modeling days, using the then 'hot'
Entity Relationship modeling tools, I quickly learned that one man's
(why are there few women on this list?) attribute was another woman's
entity! or vice versa......
S/MIME Cryptographic Signature
