This
is one area where Schematron shines. The schema author can define the diagnostic
messages to present to the user. Schematron seems to be the only schema language
that is actually designed to be useful to document authors. Other schema
languages just focus on being a formalized language for expressing constraints
without any focus on usability aspects -- especially for documents that are in
the process of being constructed.
I'm
not sure that standardized error messages are the solution, though. I don't
think a large committee will solve this problem well. I think vendors need to
put more effort into having more friendly and useful error messages. This *can*
be done. C++ compilers are notorious for their obtuse error messages, but I have
used compilers (Metrowerks comes to mind) that do a far better job of providing
clear, useful error messages than other vendors' compilers. XML Schema may not
help, here, but that's no excuse for some of the obtuse error messages that many
validators emit. They won't be able to do as well as Schematron, but they could
do far better than they currently do if the vendors/developers simply made the
committment to address this.
However, I will say that I think it would be great if we had some
lightweight standards for specific annotations that could be used to support
tools -- e.g. tooltip text, diagnostic messages for invalid constructs, and
XLinks pointing to documentation. It would be great to have interoperable ways
of leveraging schemas more fully to drive tools,
and in a way that is genuinely useful to users who often won't be XML or schema
experts.
Jonathan,
I'd also like to suggest that the XML Schema WG produce
a list of recommended error messages for validator
authors.
I realize that this is a very big job, and it's hard to
produce a complete list, but it would be very helpful
if
all conforming validators emitted the same message
for the
same error. Perhaps a starting point would be
to ask
validator authors to list the error messages
they use
together with the circumstances that trigger
them.
I'm currently trying to implement XML Schema using
Xerces
and some of the messages are very obscure, and
from some
of the postings on this topic it appears the
same error
can produce error messages that have vastly
different
levels of meaning.
Apologies to the list if this mail appears in HTML
format.
We are using Exchange and it puts all e-mail
into HTML format!
We are waiting on an upgrade to fix
this.
Regards,
Rob Griffin
Quest Software
E-mail: Rob.Griffin@oz.quest.com
Web
site: http://www.quest.com
> -----Original Message-----
>
From: Jonathan Robie [mailto:jonathan.robie@softwareag.com]
> Sent: Friday, March 22, 2002 1:36 AM
> To: xml-dev@lists.xml.org
> Subject:
[xml-dev] W3C XML Schema Test Collection
>
>
> Executive
summary:
>
> 1. XML
Schema test suites exist, and are being maintained by
> the W3C. This
> may be one of our best
points of leverage.
> 2. The public is encouraged
to add cases to the test suites
> 3. Some test
results are available online for various products
>
> Here are the details:
[snip]
