regressing the performance bugfix -- Guido wants the performance bug left
alone, because there can be no guarantee re the semantics of += vs + .
This commit is contained in:
parent
1d313ab9d1
commit
a253e183b8
@ -47,9 +47,6 @@ Core and builtins
|
|||||||
- obj.__contains__() now returns True/False instead of 1/0. SF patch
|
- obj.__contains__() now returns True/False instead of 1/0. SF patch
|
||||||
820195.
|
820195.
|
||||||
|
|
||||||
- builtin_sum() now uses PyNumber_InPlaceAdd, fixing a previous
|
|
||||||
performance bug for sum(list_of_lists) and similar cases.
|
|
||||||
|
|
||||||
Extension modules
|
Extension modules
|
||||||
-----------------
|
-----------------
|
||||||
|
|
||||||
|
@ -1840,7 +1840,7 @@ builtin_sum(PyObject *self, PyObject *args)
|
|||||||
}
|
}
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
temp = PyNumber_InPlaceAdd(result, item);
|
temp = PyNumber_Add(result, item);
|
||||||
Py_DECREF(result);
|
Py_DECREF(result);
|
||||||
Py_DECREF(item);
|
Py_DECREF(item);
|
||||||
result = temp;
|
result = temp;
|
||||||
|
Loading…
x
Reference in New Issue
Block a user