It isn't that unexplored; it isn't that well documented.
XML is good for HAS-A; not too good for IS-A.
XML is good for data objects; not so good (in fact, rotten)
for object-oriented objects.
XML is pretty good for tables, but not so good
for enforcing relational normalization rules.
XML is primarily syntax. The rest are applications
of languages that use that syntax or orthogonal
systems designed to process data objects that use
that syntax.
Many of the mistakes I encounter come of trying
to make of XML what it is not. XML can express
many styles of data laden trees, but that is
where it stops. The two ways guaranteed to
get howls are to design an XML type that exactly
mimics the implementation, then claim that is a
generalized solution (usually, a claim that
it should be a standard) or to create a
generalized solution (usually a standard) and
hand that to an implementer who knows it
doesn't apply to his or her problem (the
case for LegalXML is weak outside the US,
and I suspect this is true of many e-gov
and e-business specifications for XML
document types and schemas).
XML Doesn't Care. That's why it scales.
XML implementors have to care; that is why
they freak out a lot.
len
-----Original Message-----
From: Mike Champion [mailto:]
I definitely agree that this is a deep, and largely unexplored topic
that needs a lot of thought, research, experimentation, etc.
