So I'm now not sure what the right answer is.

The crucial question is, is xs:string validly derived from s2?

It appears that it is, because clause 2.2.4 of Schema Component
Constraint: Type Derivation OK (Simple) [1] is satisfied:

  2.2.4 B's {variety} is union and D is validly derived from a type
  definition in B's {member type definitions} given the subset, as
  defined by this constraint.

where B is s2 and D is xs:string.

This is arguably a bug, because the facets of B are being ignored.  I
believe this is a known bug, but don't have time to check the issues
lists just now.

ht

[1] http://www.w3.org/TR/xmlschema-1/#cos-st-derived-ok
-- 
 Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh
                     Half-time member of W3C Team
    2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
            Fax: (44) 131 650-4587, e-mail: ht@i...
                   URL: http://www.ltg.ed.ac.uk/~ht/
[mail really from me _always_ has this .sig -- mail without it is forged spam]

From nobody@w... Fri Jan 07 12:36:48 2005
Received: from w