> Most interesting were the recommendations for future committee work,
> which can be summed as: For God's sake, don't add anything; concentrate
> on errata and conformance tests.

That may be the groupthink at the end of the meeting, but it is not
what you see in many of pre-meeting submissions, unless you have
a very liberal definition of errata. Many people want reforms to
wildcards and UPA, in particular. When the participants go home
again, they will still need their changes, or be unable to cope
with mixed content, or runtime typing, or whatever, and wonder
"Did that workshop actually set my requirement back a step?"

Large user groups will have to show leadership with profiles.
And vendors will have to adopt a layered mentality defined outside
the XML Schemas Working Group: MicroSoft's submissions for example
manage to say that they don't want to support XML Schemas 1.1 but
that Office would like co-constraints (and, indeed, that there is
already have a kind of XPath based co-constraint system in infoPath).
So why treat co-constraints as something that the schema grammar needs
to support? Why not do it a layered way, with annotations containing,
in particular, Schematron assertions?  The pain of XML Schemas is
its monolithicity: the solution is not expansion or conformance suites
but proper layering.

I have a little blog "Snow Season in Schemaland" at
  http://www.oreillynet.com/pub/wlg/7372

Cheers
Rick Jelliffe