Clause 3 of Schema Component Constraint: Simple Type Restriction
(Facets) in section 3.14.6 of Structures appeals to the concept of one
set of facets R 'constituting a restriction' of another set B, with
respect to a 3d set S.  This raises two issues.

First, the appeal is confusing because the concept in question is not
really one of restriction but of an operation sometimes called 'set
overlay' or 'priority union' (or probably other things as well): take
everything from one set, and everything from the second set except
those members which match elements in the first set (for some
definition of matching other than identity, since if matching is based
on identity this is just a union).  Given the sets
  
    B = {minInclusive = 0, maxInclusive = 20}
    S = {maxInclusive = 100}
    R = {minInclusive = 0, maxInclusive = 100}

then B, S, and R satisfy the definition, but R does not actually
constitute anything one would be inclined to call a restriction of B.

It does constitute the overlay of B with respect to S, or the priority
union of S and B (in that order).  

So I think the term should be changed.

Second, in the face to face meeting this morning, Henry and Sandy
suggested that the purpose of this component constraint was not really
to constrain components but to provide a definition of this term.
They proposed that we should delete it here and move the relevant bits
to the place in Datatypes which was, as far as they could tell, the
only place that refers to this constraint.

So we need to decide what to do about this proposal to move the
constraint or its contents.