Re: [xml-dev] Is the XML Schema Test Collection still alive

[Michael Champion <michaelc.champion@gmail.com>]

On 4/13/05, Gregor <gregor@mrboogieman.com> wrote:

> 1. Is the schema test collection (especially the microsoft test case) still
> usable e.g. uptodate with XML Schema 1.0se and generally thought to be
> correct? 

I'm investigating this at Microsoft and hope to get more information.  


> 3. An Implementation problem:
> What parser set up did the original contributors use to test Xerces? (I get
> Xerces2 (Java Parser 2.6.2) to validate the Schema4Schema against its dtds
> but I run into problems trying to validate a Schema against the
> Schema4Schema or is there some easyer way to tell a valid XML Schema from an
> invalid one?

This touches on a larger issue of schema interoperability and
"validating the validators".   There is a W3C workshop on the general
question of Schema user experiencesin June -
http://www.w3.org/2005/03/xml-schema-user-cfp  It will be interesting
to see what others conclude.  My impression is that there are a lot of
industry schemas out there that do not actually conform to the spec,
and a lot of validators that do not enforce the spec.  That puts
people in a bind, since they don't know what to trust, and the schema
spec is far too cryptic for most people to judge for themselves. Does
anyone violently disagree with this assessment?  I would strongly urge
people who have practical experience in this area to participate in
the workshop, or at least talk about your experiences on xml-dev or
schema-dev so that we all have the benefit of your information.

By the way, most of the people who are trying to sort this all out at
Microsoft believe that W3C and the industry should be investing energy
in sorting out XML 1.0 and XSD  1.0 interoperability challenges and
building confidence that tools and schemas are correct before
complicating the picture by moving to new versions of these specs. 
Many people don't appear to realize, for example, that XML 1.1 (the
latest Recommendation) and Schema 1.0 (the latest Recommendation)
can't really be used together.  Of course, one plausible way forward
is to move to Schema 1.1 ASAP, but we believe that the more prudent
alternative is to stabilize the foundation before adding new floors on
top of what is now a bit shaky.  This argument has been made most
forcefully with respect to "Binary XML", but it applies to other areas
where people wrestle with real-world interop challenges.