Laird Nelson wrote:

> The conventional wisdom seems to be something like this:
>
        [Randy Stafford]  <snipped>  Good summary, Laird.  Good post in
general.

> (Now, my opinion.)
>
> In practice--in any kind of system that's more complicated than a simple
> web store or a school library or something along those lines--it is not
> very common at all to run across dependent objects (i.e. objects that
> are wholly owned by one class).  Typically if your design is good there
> aren't very many wholly contained objects at all, in fact.
>
        [Randy Stafford]  My experience agrees with this.  I think there is
some threshold of domain complexity at which it becomes difficult to do
things like map entity beans one-for-one with domain objects, O/R map your
domain model, separate "dependent objects" from non-"dependent objects",
etc.  I don't have a good feel for where that threshold is (or what is the
statistical distribution of domain complexity in the population of
applications today), but I suspect the threshold is a function of the
numbers of abstractions and multiple-cardinality relationships (especially
deep tree or lattice structures) in your domain model.  And the threshold is
certainly somewhere between the relatively trivial examples found in printed
books and what you encounter on some (but not all) real development
projects.

        I'm not sure there is necessarily any relationship between the
number of wholly-contained objects in a domain model and the "goodness" of
the design.  "Goodness" of design is difficult to qualify much less
quantify, and design decisions might seek to reduce (without overly
sacrificing the power of object modeling) the conceptual complexity inherent
in the problem domain to a lesser degree of complexity required to implement
the business requirements at hand (perhaps with an eye towards future
business requirements).

        Best Regards,
        Randy Stafford
        Senior Architect
        GemStone Professional Services

===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST".  For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".

Reply via email to