On 12 Aug 2009, at 11:29 , Costello, Roger L. wrote:

>
> Here's what the XML Schema 1.1 specification says:
>
>   The <redefine> construct is *deprecated*



Quite correct.  While some may wish it were not so, the
current spec does deprecate the 'redefine' construct,
just as it defines conditional type assignment,
assertions, and other constructs, and just as it
repairs certain bugs in XSD 1.0 like the bug which
allows some restrictions of unions to accept values
which are not accepted by their base type, without
the processor noticing.

On several of these issues, including both the deprecation
of 'redefine' and the repair to the union bug, the WG has
expressed hope for feedback from the community. But it
would be misleading to say that the WG "is proposing to"
deprecate redefine, unless you also change the tutorial
to say that it "is proposing to" fix the union bug -- or
indeed to say of EVERYTHING in the spec that "the WG is
proposing" it.  No W3C spec is finally done until it's
a recommendation.

Those interested in the particular issue of schema
composition and the deprecation of redefine may be
interested in examining the test results shown in a
paper I wrote for the XML Schema working group some time
ago, which illustrates that while xsd:redefine is, as
observed earlier in this thread "a supported feature
of 1.0", processors which support it seem to have
varying views of what it means in certain concrete cases.

http://cmsmcq.com/2007/schema_composition/model.xml#t0results


-- 
****************************************************************
* C. M. Sperberg-McQueen, Black Mesa Technologies LLC
* http://www.blackmesatech.com
* http://cmsmcq.com/mib
* http://balisage.net
****************************************************************