Attempting to encode a bad DOMString with stringToBinary()

In writing the test for stringToBinary(), I’m left wondering what to do
when a DOMString contains malformed UTF-16.  For example:

  stringToBinary('\ud800', 'UTF-16');

DOM 3 Core says that DOMStrings are represented using a sequence of
16-bit units which are interpreted as UTF-16.  ECMAScript doesn’t
require that a String contain valid UTF-16, though.  Should we add some
text to say that an ENCODING_ERR is thrown if the string passed in is
not valid UTF-16?

Thanks,

Cameron

-- 
Cameron McCormack ≝ http://mcc.id.au/