Thank you for the comment.  Unless we run into unexpected difficulty,
I expect the editors will do as you suggest.

Created attachment 489 [details]
Wording proposal of 19 September 2007

A wording proposal is attached, which adds a note after the definition
of list, describing the constraint on item types, and a similar
note after the definition of union, describing what may and what
must not be a member type of a union.

The wording proposal also contains a change to the description of
the simple type definition component, allowing unions with
empty sequences of member types; strictly speaking this change is not
directly related to the issue raised here (it is a change we
agreed to make when we adopted the current definition of xs:error),
but while preparing this wording proposal I happened to notice
that we had not made the necessary correction yet, so I took the
opportunity to do it now.

N.B. this wording proposal has NOT had the benefit of review by the
other editors.

I've added the keyword 'needsReview' to signal that WG action is needed
on the wording proposal in the attachment.

Looks good to me.

On 21 September, the Working Group accepted the wording proposal of
19 September found in the attachment.  I'm closing this issue accordingly.

Michael, if in your role as intermediary between the XML Schema WG and
the QT WGs you would report this resolution to QT and then signal
their assent to or dissent from the resolution in the usual way, we'd
be grateful.

The wording proposal adopted 21 September and mentioned in comment #5
was integrated into the status-quo document in October 2007.  I'm 
setting the keyword field to 'resolved' to reflect that fact.