Is it true to say that when an element is validated against a type (as
distinct from being validated against an element declaration - for =
example
when validating within xs:anyType content) then:

   xsi:nil is allowed, and must (if present) be a valid boolean, but its
value is ignored.

In particular xsi:nil="true" is permitted but it neither allows nor =
requires
the content to be empty if the type itself does not permit empty =
content.

This seems to be what Validation Rule: Element Locally Valid (Type) is
saying, the relevant consideration being that "clause 3.2 of Element =
Locally
Valid (Element) (=A73.3.4) did not apply", because we were not =
validating
against an element declaration.

(Actually, I can't see a detailed chain of reasoning that allows me to
conclude that validation should fail if the attribute xsi:nil="fgjh" =
is
present, but I assume that in this case common-sense prevails.)

Test case elemZ033b is related, though in this case the instance is =
valid
whether xsi:nil is ignored or not.

Michael Kay
http://www.saxonica.com/


From mike@s... Sat Dec 09 23:19:20 2006
Received: from lisa