[
Lists Home |
Date Index |
Thread Index
]
Even where semantics are shared, it doesn't necessarily lead to a common
data structure. What may be an appropriate structure for XML serialization
of data may not work so hot as a object containment hierarchy.
I'm not talking about differences in physical data models, but differences
in how logical models with share semantics are structured. For example, I'm
trying to merge two similar but independently developed data models right
now. Both are built on different assumptions, and some entities are
different; but there are also entities that are 'missing' from one of data
models. It's not that they don't logically exist, it's that they're
implicit in one of the domains and explicit in the other.
Merging these two models involves the trick of hiding these 'extra'
(assumed) entities under existing interfaces for one of the domains.
Navigation to data fields may now involve more steps for this domain, but
it's not obvious from the interface; however, it will be explicit for the
other.
I guess I would say (arguably, I admit) that they're both operating on the
same physical model and the same semantics, but the logical models are
somewhat different.
|
