[
Lists Home |
Date Index |
Thread Index
]
At 10:30 PM 7/6/2002 -0700, Dare Obasanjo wrote:
><Uche>
>These strike me as precisely the sorts of matters that are not even yet
>ready for standardization.First of all, querying a lot of persistent XML
>and viewing non-XML data in XML forms are very different matters, and very
>different needs. The latter is useful regardless of how much data is in
>play, and whether or not it is persistent. It is not ready for
>standardization because it is such a varied matter. The RDBMS >vendors
>all have different approaches to the problem, and on the OO front there
>are things such as JXPath. I'm not sure why this diversity needs to be
>scrapped. The former, I think, might eventually make sense to
>standardize, a la SQL/OQL, but I think that the practice of it is still
>taking shape.
>
></Uche>
>
>In this area I completely agree with you 100%. Standardization should
>occur after some practice has been established and the pros & cons of
>various approaches can be weighed.
Well, we *did* compare XPath, XML-QL, XQL, XSLT, YATL, Lore, and SQL in
conjunction with our use cases. This comparison led to the initial design
of XQuery.
>Instead we have a standards process driven by research interests where
>core aspects of the technology in the standard have never seen the light
>of day outside a non-research/academic setting. Unfortunately, the XML
>database vendors are to blame for this (Yes, us included) for wanting
>standardization early and not wanting to fragment the market.
Implementation experience is needed before XQuery should be accepted as a
Recommendation. And the W3C has been listening on this front. I doubt very
much that we will be adopted unless it is clear that XQuery is
implementable. Remember, XQuery is still a Working Draft. It is becoming
more stable, but still changing.
By 'core aspects', I assume you are speaking of the static typing of the
Formal Semantics, which has not yet been widely implemented?
>My personal opinion would involve scrapping the XQuery effort and
>revisiting the issue in two or three years when lots of practice was
>established. Of course, the realities of the industry make that merely a
>renegade opinion and not an indication of the future.
Some people fear that doing things this way would mean that the entire
industry must follow the lead of the big companies, who would develop the
technologies that would later be standardized. Smaller companies would be
excluded from developing new languages like XQuery, because people are
afraid to rely on languages that are used only in the products of a small
company. Coming up with a standard after several years of experience with
competing vendors is not always straightforward - C# and Java are extremely
similar, but when do you think we will agree on a standard to replace these
two languages?
I think that following this approach would help large vendors to make their
users dependent on them. I don't think that it would really favor users.
We would be better off with a process for revisiting W3C specifications
after some period of time. For instance, perhaps a Recommendation should
have a lifetime of 3-5 years, after which it is replaced with a real standard.
Jonathan
|
