[FS] Inconsistency in the semantics of sum() function

Hi,

It seems to me that there is an inconsistency between Formal Semantics and 
Functions & Operators specs on the semantics of sum().

FO spec section 15.3.5 says that any value of type untypedAtomic must be 
converted into a double value, and for the one argument sum() function the 
double value 0.0 must be return for an empty sequence.  However, the FS 
spec section 6.2.6 provides inference rules that require the conversion of 
 untyped atomic  values into integer values, and says that sum(()) = 0 
(the integer value 0) :
"The fn:sum function has two forms. The first form takes two arguments: 
The first argument is the input sequence and the second argument is the 
value that should be returned if input sequence is empty. In this case, 
the result type is the union of the target type T and the type of the 
second argument.

                statEnv |- Expr1 : Type1 
        Type3 = convert_untypedAtomic(prime(Type1), xs:integer) 
                Type3 can be promoted to T 
        statEnv |- Expr2 : Type2      statEnv |- Type2 <: 
xdt:anyAtomicType ? 
 
--------------------------------------------------------------------------------
                statEnv |- fn:sum (Expr1, Expr2) : T | Type2 

The second form of fn:sum takes one argument, with an implicit second 
argument of the integer value 0. In this case, the result type is the 
union of target type T and xs:integer.

                statEnv |- Expr1 : Type1 
        Type3 = convert_untypedAtomic(prime(Type1), xs:integer) 
                Type3 can be promoted to T 
 
--------------------------------------------------------------------------------
        statEnv |- fn:sum (Expr1) : T | xs:integer 
"

Best regards,
Achille Fokoue