CDATA sections may occur in Mixed content. CDATA sections, comments and PIs may occur in ANY content. Default values for IDREF attributes must match Name. Default values for ENTITY attributes must match Name. Default values for IDREFS attributes must match Names. Default values for ENTITIES attributes must match Names. Default values for NMTOKEN attributes must match Nmtoken. Default values for NMTOKENS attributes must match Nmtokens. Default values for NOTATION attributes must match one of the enumerated values. Default values for enumerated attributes must match one of the enumerated values. Non-syntactic validity errors in default attributes only happen if the attribute is in fact defaulted. Default values for attributes may not contain references to external entities. Even internal parameter entity references are enough to make undeclared entities into mere validity errors rather than well-formedness errors.