24 Mar
2015
24 Mar
'15
2:29 p.m.
[for just patch 1/4; I didn't look as closely at the others] Reviewed-by: Jeff Epler jepler@unpythonic.net
...with one half of a caveat (rounded up?): Like most or all of the originals, and like DIV_ROUND_CLOSEST just above it in kernel.h, the new code gives an incorrect answer if the temporary overflows at + unsigned long long _tmp = (x) + (__d) / 2;
As a doc improvement a remark could be added to a number of these functions, but there's little reason for that to be part of this series.
Jeff