-;;; In the float case, we pick off small arguments so that compiler can use
-;;; special-case operations. We use an exclusive test, since (due to round-off
-;;; error), (float most-positive-fixnum) may be greater than
-;;; most-positive-fixnum.
+;;; In the float case, we pick off small arguments so that compiler
+;;; can use special-case operations. We use an exclusive test, since
+;;; (due to round-off error), (float most-positive-fixnum) is likely
+;;; to be equal to (1+ most-positive-fixnum). An exclusive test is
+;;; good enough, because most-positive-fixnum will be one less than a
+;;; power of two, and that power of two will be exactly representable
+;;; as a float (at least until we get 128-bit fixnums).