Re: [xml-dev] Has XML run its course?

["W. E. Perry" <wperry@fiduciary.com>]

XML has certainly run its course as a unitary body of practice, which I believe
is Simon's fundamental point. Nowhere in the ten original principles was the
long term preservation of interoperability elevated to an explicit goal. The
assumption appears to have been that the rulemaking (or 'recommendation'
development) authority of the W3C would provide a method of extension to the
original XML specification, while ensuring the moral suasion to hold the
consensus of a professional community for an ever-expanding body of practice.
Yet the significant innovation of XML 1.0 was, first, simple
well-formedness--as divorced from validity on terms of a priori
expectations--and, second, the explicit recognition of WF's corollary:  that
for many cases well-formedness is the only reasonable syntactic prerequisite to
processing. It was the revelation found in simple well-formedness which first
attracted many of us to XML and which, in the face of the W3C's exercise of
authority to expand dogma, afforded the sufficient basis for our protestant
faith in the original syntactic specification.

The W3C is in the unenviable situation of extending XML by generalization to
encompass every special interest which can demand its attention, while the
alternative is perfecting tools through particularization and speciation using
the relatively simple syntactic approach offered by XML 1.0. By definition,
such speciated tools do not work in the 'general case', but they do their
particular jobs well and are quickly and easily adaptable to changes in the
instance documents they encounter. This by contrast to the entire family of
processors which begin from the original premise of validity:  that the initial
task of a tool is to ensure that an instance upon which it might operate first
conforms to fixed prior expectation. The former is a basis for responsive,
adaptive improvement, while the latter--by its nature static--must, to admit
progress at all, be supplemented with a method of extension like the W3C's
rulemaking authority. In the former case, we never expect interoperability of
processes, but we gain a common, perhaps even portable, body of data or
documents which, designed to one narrow set of expectations, may nonetheless be
manipulated usefully by widely different processes never anticipated--and
certainly never accommodated for--by the document authors. In the latter case,
interoperability of process becomes a tactically necessary goal in order to
insure the continued, indispensable acquiescence of the developer and user
community for an ever-expanding body of  'recommended' practice.

Respectfully,

Walter Perry