[
Lists Home |
Date Index |
Thread Index
]
I agree with that. Keep in mind what was
not available when SGML was designed and how
much convergence XML could pick up for free
that SGML users had to bake in via the SGML.
Then the difference between invention and
adaptation, necessity and convenience is
manifestly clearer. Other aspects such as
well-formedness were already part of fielded
SGML systems if not conformant systems, and
over time, we are finding the shortcomings
of that approach have made it necessary for
new technical innovations (xml: namespaces) to
become more complex. Even informally deprecating
formal public identifiers is not a universally
accepted practice.
One can say as some have including me that
SGML made it easier for authors and XML for
programmers. That is mostly true. However,
without a good grounding in the state of the
industry when these were spec'd, it is misleading.
Yes, as Gavin said, fixed SGML Declarations were
de rigeur for the owners of SGML application
languages (eg, MIL IETMs), but we still had
to tweak them locally. So the flexibility had
advantages. If that was "too hard", we hired
smarter developers because contracting for
smarter users was not, nor ever is, an option.
The biggest problem in attempting to make
SGML go away is that now XML has to
work for non-Web and Web systems. It is
as if, tired of the food, pirates made
the cook walk the plank before they determined
if anyone else knew how to cook on a rolling
ship in a storm.
It is a hard job to bake one meal for all
crews and passengers. In fact, somewhat
impossible if pleasing them all is a goal.
len
From: Tim Bray [mailto:tbray@textuality.com]
At the end of the day XML's
main technical contribution may turn out to have been that it dragged
Unicode into the mainstream. -Tim
|