> Also, would it be better for the diagnostic to point to the "const" in the function rather than the "f"? Probably, but the location of the const on a function does not appear to be available. http://reviews.llvm.org/D4479