Bill:
Please see the write-up I just sent out regarding this issue. I do believe
that usability is a major issue here, both in terms of the understandability
of the document structures and that of OO code used to process those
structures (the encapsulation argument).
What I'd like to know is why, with a couple of simple changes, we could
optimize for performance and usability *without impacting LCSC at all) we
don't do it. The lack of resources is not a good argument here, especially
when you sconsider what our extended documents look like. What are the
design criteria that account for the current design? Mostly, they are based
on normalization which has, I will admit, been successful in all but the two
limited cases I'm addressing - it produces an algorhithmic consistency where
before there was no solid basis for such consistency. I like it. But it has
a couple of small problems...
Anyway, it's all in the write-up...
Cheers,
Arofan