Test case w/unsupported datatypes

I'm working through the testcases and wondering about this test case:

       http://www.w3.org/2001/sw/DataAccess/tests/#extendedtype-ne-fail

which tests the decision that opaque types can not be tested for != . I
understand the intent, but wonder how that is justified given the definition
of A != B as fn:not(RDFterm-equal(A, B))? I guess A != B is in fact
something like:

	if A and B are literals of unsupported type 
		A != B is false
	else
		A != B is fn:not(RDFterm-equal(A, B))?  

Is that right? Some clarification of this in the docs would be good if so.

FWIW, I think it would have been a better decision to say that RDFterm-equal
of unsupported datatypes is a type exception if the lexical values aren't
the same (and so != would be as well). 

Thanks,

-Geoff