[
Lists Home |
Date Index |
Thread Index
]
On Mon, 18 Oct 2004 15:46:00 +0100, Michael Kay
<michael.h.kay@ntlworld.com> wrote:
> XQuery is not a big language. It is arguably smaller than XSLT 1.0. If
> vendors produce implementations that don't conform, this will probably be
> because they are trying to do things like translating the language into SQL,
> which is never going to work 100%. There's no good reason at all for native
> implementations to leave features out
Translating the language into SQL is exactly what I was thinking about
in the "effectively implemented in which types of software" line.
It's the distribution of needs across the real world of actual users,
not within a working group, that ultimately determines which features
are important. For example, XML 1.0 was clearly designed by document
people and there may be no good reason for document-oriented
applications to not implement all of it. For better or worse,
however, XML has gotten the most actual usage for data and message
exchange, and some of XML 1.0's features have proven to be highly
burdensome for those use cases. Something similar could happen with
XQuery, we shall just have to see.
It would be great if the investment in time needed to get XQuery to
Recommendation status does really lead to a spec with the quality that
there is nothing worth leaving out. It would be a something of a
first for W3C, -- or any consensus-driven process -- so I'm reserving
judgment. My point about "XML-I" is that I'd like to see the judgment
on which features in a spec are truly needed in which scenarios
ultimately made by users and implementers. The web services community
has determined from experience that a number of the features in the
SOAP spec add more complexity than they're worth, so they were removed
from the WS-I interoperability profile, irrespective of the thoughtful
work that went into the original specs. If it turns out that the
XQuery spec does strike a near-optimum balance between the needs of
its diverse audience, it will have been worth the wait!
|