https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109229
--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Maybe we could do this:
--- a/libstdc++-v3/include/std/numeric
+++ b/libstdc++-v3/include/std/numeric
@@ -483,12 +483,16 @@ namespace __detail
_OutputIterator __result, _Tp __init,
_BinaryOperation __binary_op)
{
- while (__first != __last)
+ if (__first != __last)
{
- auto __v = __init;
- __init = __binary_op(__init, *__first);
- ++__first;
- *__result++ = std::move(__v);
+ auto __sum = __binary_op(__init, *__first);
+ *__result++ = std::move(__init);
+ while (++__first != __last)
+ {
+ auto __tmp = __sum;
+ __sum = __binary_op(__sum, *__first);
+ *__result++ = std::move(__tmp);
+ }
}
return __result;
}
We'd need something similar for inclusive_scan, reduce etc.
But I don't think this is actually allowed by the standard. It clearly lists
the forms allowed when invoking the summation operator, and binary_op(sum,
*first) isn't one of them.
The reason for the requirement that the result of the summation operation can
be converted to T is precisely so that the summation can be done in that type.
