[
Lists Home |
Date Index |
Thread Index
]
Michael Champion wrote:
>I heard from my colleague Dave Remy, who is on the program committee
>for the W3C Schema Experience Workshop in a couple of weeks, that most
>of the experience reports they've received talk about data-oriented
>use cases and not document-oriented use cases. It's apparently not
>too late to submit an experience report or attend the workshop
>(although it's probably too late to get on the formal program).
>
>Once again I'd encourage people to write down a page or two of
>thoughts and send them in. See
>http://www.w3.org/2005/03/xml-schema-user-cfp for details.
>
>[obligatory disclaimer here] Maybe lots of people have had the
>experience of building document-oriented apps with XSD, but it was so
>painful no one wants to remember it and write it down. [duck]
>Nevertheless, the process might be therapeutic, and W3C need to hear
>the experiences, Nobody is going to whitewash or suppress unfavorable
>reports, I'm quite sure that W3C, WS-I, vertical standards bodies, and
>the vendors have open minds about what the reality is and what is to
>be done about it.
>
Does "I know of almost no large uses of XSD for documents" count? A
non-experience report?
The crowd I come in contact with (mostly the kinds of users who used
SGML before: military,
legal, pharmeceutical) are very conservative as far as editing
environments: if they are using
FrameMaker they are stuck with DTDs, if they are using text editors or
markup editors (like
Topologi's) they usually want to have character entity references and so
keep with DTDs,
if they are using Word->XML conversion they hard code some schema
(usually a DTD)
into the converter code, and so on.
Furthermore, document-oriented systems very often have large files that
tree-based tools
(such as those that use XML Schemas) often won't open successfully.
Furthermore, the large companies tend to have evolving systems and
evolving schemas,
so they don't get much value from switching from one grammar to
another. And, where
they do want more modeling power, RELAX NG is more powerful than XSD. (And
if the site has no tools for applying default values from a XSD, it
again loses out to DTDs
and is on par with RELAX NG.)
The trouble with moving to XML Schemas is that it may involve
substantial re-tooling.
For example, the recent release of Epic editor supports XML Schemas:
users who have
a small number of document types expressed in XML Schemas but many
expressed with
DTDs have to figure out whether it is more practical to convert the XML
Schemas to
DTDs or upgrade their editing/production environments. (Or just to avoid
XSD.)
Finally, publishing data is rife with co-occurrence constraints: serious
publishers may
be more interested at implement Schematron.
So, realistically, I only expect to see widespread XML Schemas adoption for
documents piggybacking on the release of new industry standards:
S-1000D and DITA
for examplein the short-term. These don't come out very often, compared
to data-oriented standards,
and the promulgators of these may decide to put out DTD or RELAX NG
versions just to
aid adoption.
In the long-term, retooling and upgrades will make XSD-using tools more
available, and it
will make XSD much more viable, though still not particuarly interesting.
Cheers
Rick Jelliffe
|
