Lists Home |
Date Index |
> The people that will carve out the small and slim successor to XSLT
> (and XPath, and XQuery, and DOM, and XML schemas, etc. etc.) are the
> end users and those who learn from them. Call it Mother Nature,
> Father Darwin, the Invisible Hand, or whatever ... all of us on this
> list, and our customers/readers/competitors, are going to bury the
> mistakes and cross-breed the successes.
So with that in mind, should we bother trying to make those specs that
are obviously complex-and-doomed-to-eventual-obsolescence better?
Perhaps it would be best to encourage the designers to make XPath 2.0
as complex as possible, so that it's easier for "XPath NG" to be
simple and effective in comparison?
I'd be content with this course of action were it not for the "XML
Schema effect". XML Schema is complex, not suited to the requirements
of many of its putative users, lacks a convincing reference
implementation, and has a beautifully simple alternative. However, it
has the lion's share of the schema market and I guess it'll be at
least another year before we see a new version of the schema language
blessed by the W3C (and thus with the W3Cs pulling power with
implementers and users).
Like you, I doubt it's possible for a committee to come up with a
specification that's clean and simple, especially when that committee
is formed from two working groups with different sets of goals and
assumptions. In that regard, XSLT/XPath 2.0 is a miracle of
engineering -- it could certainly be much worse. I also seriously
doubt that a collection of people with over a year of opinion-forming,
consensus-building and plain hard work will make any major changes on
the basis of a niggling thing like community opinion. On the other
hand, maybe it's short-termist of me, but the idea of at least two
years of helping people through their XSLT/XPath 2.0 problems fills me
with dread. It's hard to know what to do.
What can we do to help Mother Nature, Father Darwin and the Invisible
Hand (strange mental picture, that) on their course?