RE: Picking the Tools -- Marrying processing models to data model s

[Jeff Lowery <jlowery@scenicsoft.com>]

> From: W. E. Perry [mailto:wperry@fiduciary.com]
...
> Put differently, the data structure exhibited by an 
> XML instance is
> still capable of sufficiently variable realization in process 
> as to bridge the
> lack of shared data definition between the autonomous nodes 
> of the internetwork
> topology.

Still, there are constraints on data models, at least the data models you
modify and intend to hand back to me. XML Schema allows me large scope for
instruction as to how I think the data model should behave, at least for
individual entities, plus some rules for cardinality relationships between
entities.

Where XML Schema isn't useful is in describing sophisticated constraints
based on relations, e.g. "if X>1 then Y must be <2 if Z is also true".
Something like Schematron may help, but objects are also well suited for
defining sophisticated relational constraints (along with all the other
stuff they do). So, for complicated rules regarding data values, we still
need objects; nonetheless, I would contend that 80% or better of all data
constaints aren't complicated or are rules regarding relations, and can
therefore be handled by a declarative schema. So forcing the weight of
objects on all designs seems like overkill for most cases.

Maybe somebody can frame this argument a little better... <:-P