This issue against Structures should be dealt with at the same time as 
bug 2246 against Datatypes.

(In reply to comment #0)
> Clause 1.3.2 of Schema Component Constraint: Derivation Valid
> (Restriction, Simple) in section 3.14.6 of Structures reads:
> 
>     1.3.2 If there is a facet of the same kind in the {facets} of the
>     {base type definition} (call this BF), then the DF's {value} is a
>     valid restriction of BF's {value} as defined in [XML Schema:
>     Datatypes].
> 
> The concept of one facet value being 'at least as strong as' another
> value for the same facet is a very useful one, and it is good that
> Structures appeals to it here.  The problem is that this is a throw
> which the Datatypes spec does not catch very well: the notion is
> reasonably well covered with various component constraints about
> restrictions that are valid or invalid, but the Datatypes spec does
> not provide any single definition for the terminology used in the call
> from Structures, or for any other term for the concept.

See discussion, comment 5 of bug 2246.  This needs to be harmonized with 3.15.6, Schema Component Constraint: Simple Type Restriction (Facets).

(In reply to comment #1)
> This issue against Structures should be dealt with at the same time as 
> bug 2246 against Datatypes.

It's not clear (at least to me) that bug 2246 has anything to do with this one.  This does not mean that there might need to be a change to Datatypes to fix this one, of course.  Bug 2246 addresses the question of whether when deriving using a particular facet in the presence of that facet in the base type, the now-redundant base type's facet should be included in the derived datatype's {facets} set.  See bug 2246 and the fix needed for 2246 in Structures proposed in 4034.

For this bug, reference should be made specifically to the various schema component constraints specified separately within the Datatypes description of each kind of facet, rather than amorphously to an unspecified definition in the Datatypes spec.  If a Datatypes definition of "valid restriction" is wanted, a new bug should be entered for it.  (I don't pretend to know how that bug should be stated, or what to suggest as a fix.)

(In reply to comment #0 and #3)
from comment #0:
> Clause 1.3.2 of Schema Component Constraint: Derivation Valid
> (Restriction, Simple) in section 3.14.6 of Structures reads:
> 
>     1.3.2 If there is a facet of the same kind in the {facets} of the
>     {base type definition} (call this BF), then the DF's {value} is a
>     valid restriction of BF's {value} as defined in [XML Schema:
>     Datatypes].

from comment #3:
> For this bug, reference should be made specifically to the various schema
> component constraints specified separately within the Datatypes description of
> each kind of facet, rather than amorphously to an unspecified definition in the
> Datatypes spec.  If a Datatypes definition of "valid restriction" is wanted, a
> new bug should be entered for it. 

Perhaps a Note following the clauses of "Derivation Valid (Restriction, Simple) which says:

   Note:  "Valid restriction" in clause 1.3.2 and 3.1.2.5 means that
   the facet in question satisfies the constraints given separately in
   the appropriate subsection of [reference to Datatypes 4.3] which
   describes that particular kind of facet.

Then no change to Datatypes is required.
   

The Working Group accepted a wording proposal resolving this issue
at its face to face meeting in New Orleans 31 January - 2 February.
Final amendments to the wording were approved 23 February.  Accordingly,
I'm marking this issue resolved.