Use PyInt_AS_LONG macro instead of explicit inlining.
diff --git a/Python/ceval.c b/Python/ceval.c
index bcc2288..3625473 100644
--- a/Python/ceval.c
+++ b/Python/ceval.c
@@ -799,8 +799,8 @@
if (PyInt_Check(v) && PyInt_Check(w)) {
/* INLINE: int + int */
register long a, b, i;
- a = ((PyIntObject*) v)->ob_ival;
- b = ((PyIntObject*) w)->ob_ival;
+ a = PyInt_AS_LONG(v);
+ b = PyInt_AS_LONG(w);
i = a + b;
if ((i^a) < 0 && (i^b) < 0) {
PyErr_SetString(PyExc_OverflowError,
@@ -824,8 +824,8 @@
if (PyInt_Check(v) && PyInt_Check(w)) {
/* INLINE: int - int */
register long a, b, i;
- a = ((PyIntObject*) v)->ob_ival;
- b = ((PyIntObject*) w)->ob_ival;
+ a = PyInt_AS_LONG(v);
+ b = PyInt_AS_LONG(w);
i = a - b;
if ((i^a) < 0 && (i^~b) < 0) {
PyErr_SetString(PyExc_OverflowError,
@@ -1390,8 +1390,8 @@
/* INLINE: cmp(int, int) */
register long a, b;
register int res;
- a = ((PyIntObject*) v)->ob_ival;
- b = ((PyIntObject*) w)->ob_ival;
+ a = PyInt_AS_LONG(v);
+ b = PyInt_AS_LONG(w);
switch (oparg) {
case LT: res = a < b; break;
case LE: res = a <= b; break;