PyArg_ParseTuple and Unicode

[Scottie]

I  am confused here, but this is not uncommon.  I am trying to get an
extension to handle both normal strings and unicode.  From my initial
reading of the PyArg_ParseTuple document, I thought the following
would work:

if( PyArg_ParseTuple(arg, "...u#...", ...) ) {
    ...Unicode ops...
}
else if( PyArg_ParseTuple(arg, "...s#...", ...) ) {
    ...plain string ops...
}


My understanding of "u" and "u#" was that they would fail on non-
unicode input (while "s" and "s#" pass both along).  The behavior I
see is different: "u#" gives me a pointer to the base of a vanilla
string, but divides the length by two.

Is this problem behavior in 2.2a4, or does the document need to
explain this better, or is there something I am just not understanding?
Or, of course, some combination of the three.

Thanks,
     Scott David Daniels,
    Scott.Daniels at Acm.Org