[jdom-interest] use-case driven vs use-case dominated

[Joseph Bowbeer]

These remarks aren't in response to any particular feature request or
response.  I just wanted to agree with Patrick Dowler in saying that we
should be careful not to let use cases dominate; that coolness does count,
and that we should be careful not to beat-down good design just because we
haven't yet found someone who is willing and able to defend their particular
use.

(I also agree with Philip Nelson that understanding what JDOM is and what it
isn't is really important.  For example, saying that JDOM *is* an XML
document representation and *not* a data representation or a data-binding
tool greatly simplifies JDOM's design space.)

Libraries vs. use-cases:

Relying on use-cases can be healthy for the design, in that it reduces drift
and bloat, but designing libraries (and programming languages for that
matter) is different than designing applications in that the library design
necessarily anticipates user experience.  It's impractical to develop a
standard library using the short-cycle iterative feedback approach that
user-centric design requires because the library is released infrequently
and users depend on its stability.

When the design anticipates user experience, then other more intuitive and
artistic factors should come into play; factors such as simplicity,
generality, conciseness, and even elegance (aka coolness?).  For example,
Java is an elegant combination of power and simplicity and that's why we
like it.  In these situations, designers must rely more on their own
abilities as creators -- to reach beyond the disparate list of RFEs and do
something better that will not only satisfy current users, but attract new
users as well.

--
Joe Bowbeer