Don't test here for negative number to float power; that belongs in

floatobject.c.

Python/ceval.c

@@ -2282,15 +2282,6 @@ powerop(v, w)
 		err_setstr(TypeError, "pow() requires numeric arguments");
 		return NULL;
 	}
-	if (
-#ifndef WITHOUT_COMPLEX
-            !is_complexobject(v) && 
-#endif
-            is_floatobject(w) && getfloatvalue(v) < 0.0) {
-		if (!err_occurred())
-		    err_setstr(ValueError, "negative number to float power");
-		return NULL;
-	}
 	if (coerce(&v, &w) != 0)
 		return NULL;
 	res = (*v->ob_type->tp_as_number->nb_power)(v, w, None);