" I hear you are returning to this forum from a 10 year sabbatical? "
I've never left in terms of reading it. I am building XML-based
systems again as a full-time job. It wasn't a sabbatical. It was a
"XML Is Done; Use It When Needed Because It's Fully Baked". And
therefore, it was best to stay quiet because I had nothing to
contribute. Once back in the game, this is a good place to ask
questions and compare experiences. Everyone here has done this long
enough that no serious question goes unanswered and otherwise, we know
each other well enough to ignore the rest.
Part of the laissez-faire attitude for some of us is that XML-Dev (as
development) once XML was published became a kibitzing forum and helps
people understand the specification. As a development list, very
little is developed here. Some abstractions are debated repeatedly.
It isn't a matter of discarding "new ideas" but very few new ideas are
presented. For the most part new people who don't realize the world
didn't begin when they entered it present old ideas and these are
debated politely in case new conditions have emerged. Minds aren't
closed; they are skeptical by practice.
You might want to search the pre-xml archives for a phrase
"object-oriented pixie dust". Creating objects with markup is a
perma-thread from the early days. The problem as I recall was some
optimizations simplify one problem for a little speed and create
numerous problems in other applications. Tim Bray called such ideas
"premature optimization" and the answer has been "good stuff for some
specialized processors doing specific tasks".
If you want a lexically concise answer, John Cowan is the better man.
Amelia made a concrete suggestion. That's the next wicket.
XML is used both on the web and off the web. The speed of reading XML
in and out of memory isn't the same constraint in all application
domains. Some XML documents are quite large by some standards.
Others not so much. When building large integrated document sets in
say technical publication applications, if an OID complicates the
production process and toolsets by a fraction without an offsetting
measurable and explainable gain, it will be discarded. If XML is
re-architected such that it is required, XML will be discarded.
len
